Mental property and funding in Synthetic Intelligence

Government abstract

This examine commissioned by the UK Mental Property Workplace (UKIPO) investigates necessary components influencing decision-making on UK funding in Synthetic Intelligence (AI). Particularly, it analyses the extent to which that is influenced by the UK mental property (IP) framework presently in place.

Definitions of funding, market functions and AI functions have been established. This gives a scope for the examine and a rationale for sampling views throughout a spread of related actors within the sector.

We gathered data from completely different sources. We first reviewed the literature on AI and AI funding. We then carried out a sequence of face-to-face stakeholder interviews, facilitated by a ready set of questions, through the interval from November 2021 to January 2022. We additionally benefited from discussions with a number of members of the UKIPO group and different events already engaged in associated UKIPO research on IP and AI.

The conclusions from the literature survey are summarised beneath.

  • the UK is quantity three on the planet as a vacation spot for personal funding in AI (behind United States of America (USA) and China) and primary in Europe in 2020 (Zhang, et al., 2021);

  • UK-based candidates and inventors rank sixth by way of patenting exercise measured by the variety of AI associated patent households (UKIPO, 2019)

  • London is the primary hub in Europe for AI improvement exercise

  • the UK has a big pool of AI and software program expertise, with necessary analysis institutes targeted on AI analysis and improvement (R&D)

  • the UK shares the benefit of being naturally English-speaking with the a lot bigger USA ecosystem

  • the UK is nice at growing AI options however is missing in funding and AI firm development past the start-up stage. The stakeholder interviews present some components of response on this

The face-to-face interviews present further conclusions, summarised beneath.

  • the UK IP Framework shouldn’t be a basic consideration for funding by the individuals on this examine

  • different components dominate funding selections: the supply of AI-skilled individuals; the popularity of expertise hubs and universities; authorities grants and entry to early-stage funding. Total, the relative energy of those components encourages AI funding within the UK

  • AI corporations have an internationally dispersed workforce from the start-up part onwards. That is decided by the necessity to discover AI expertise wherever on the planet. Alternatives for residence engaged on software program improvement come up for worldwide graduates returning to their residence international locations

  • the UK IP framework is broadly considered trusted and dependable and never of concern in funding selections, perceived as a background hygiene issue. Nonetheless, it was famous that early-stage AI start-ups lacked expertise of utilizing the UK IP framework

  • buyers usually require patents to assist funding selections. Patents present third-party opinions on the distinctiveness of the expertise and a ‘saleable asset insurance coverage’ within the occasion that the corporate ceases buying and selling

  • expertise sector individuals had a usually perceived false impression that ‘core AI software program shouldn’t be patentable’. The same remark was made in a lately revealed DCMS and OAI commissioned report on AI (Westgarth, et al., 2022). This report additionally notes that the worth of AI is derived from its software to unravel an present downside. But, patent safety for AI-related innovations is obtainable throughout all expertise areas. Patent innovations embodying algorithms or software program are allowable in the event that they make a technical contribution. The UKIPO, just like the European Patent Workplace (EPO), grants many hundreds of patents for pc and software program carried out innovations together with within the subject of AI (UKIPO, 2019). Thus, they do present safety for innovators on this subject. The UKIPO’s Handbook of Patent Observe gives in depth steerage on the patenting of software program and pc carried out innovations

  • as start-ups develop into the small and medium enterprise (SME) class, understanding of patents and copyrights improves, however understanding of the IP framework for commerce secrets and techniques stays weaker

  • there seem like two communities of curiosity utilizing and/or growing AI expertise and the UK IP framework. The expertise sector group tends to concentrate on patents and depends on patent legal professional corporations whereas the media sector is served by copyright solicitors

  • presently there’s a huge hole between these communities with little mixing of experience. Integrating these two areas {of professional} experience would assist early-stage AI corporations develop business methods to grasp the total potential of their key property by way of safety, fundraising and exploitation

The UK has established a powerful popularity as a spot to incubate AI companies. Nonetheless, it’s seen as too small a market to develop massive corporations and to safe the bigger investments wanted for main long-term development. Many younger corporations plan an early exit or a transfer to the far bigger USA market with its popularity for risk-taking and enormous funding. The UK too typically loses the larger-scale financial advantages of AI enterprise development to bigger economies. Challenges in scaling up for UK progressive companies can also be highlighted within the Proof paper of the UK Innovation Technique (BEIS, 2021).


  • AI – Synthetic Intelligence

  • BEIS – Division for Enterprise, Power, and Industrial Technique

  • CEO – Chief Government Officer

  • COVID-19 – Coronavirus Illness attributable to the SARS-CoV-2 virus

  • CTO – Chief Know-how Officer

  • DCMS – Division for Digital, Tradition, Media, and Sport

  • DTSA – Defend Commerce Secrets and techniques Act

  • EPFL – École Polytechnique Fédérale de Lausanne

  • EPO – European Patent Workplace

  • EU European Union

  • GDP Gross Home Product

  • IBM Worldwide Enterprise Machines

  • IP – Mental Property

  • IPO – Mental Property Workplace

  • IT – Info Know-how

  • M&A – Mergers and Acquisitions

  • ML – Machine Studying

  • NLP – Pure Language Processing

  • OECD – Organisation for Financial Co-Operation and Growth

  • ONS – Workplace for Nationwide Statistics

  • PCT – Patent Cooperation Treaty

  • PE – Personal Fairness

  • R&D – Analysis and Growth

  • SME – Small and Medium Enterprise

  • UKIPO – UK Mental Property Workplace

  • USA – United States of America

  • USPTO – United States Patent and Trademark Workplace

  • WIPO – World Mental Property Group

  • VC – Enterprise Capital


The principal goal of this report is to analyze the present determinants in funding in AI worldwide and the position of IP. For rising corporations within the UK, entry to funding could be a main hurdle (Workplace for Nationwide Statistics (ONS), 2019). Moreover, it’s unclear whether or not using IP and the flexibility to guard progressive concepts serves as an attractor for funding.

This examine goals to know whether or not the UK IP framework performs a task in incentivising funding in AI expertise and AI companies within the UK. It additionally seems to know the position of different components in funding selections within the UK relative to the IP framework.

We regard the IP framework as encompassing not solely the formal framework as outlined by the UKIPO and to a lesser extent the EPO, but additionally different wider dimensions. These dimensions embrace the associated authorized and business companies occupation and the judicial/authorized enforcement framework in place for the safety and litigation of IP rights. It additionally accounts for the broader IP training and coaching ecosystem.

Nonetheless, it must be famous that AI functions, services are worldwide with international attain and alternative past the UK.

We additionally study the query of entry to coaching datasets for AI fashions and the supply of such knowledge for corporations of various sizes and sectors. Moreover, we contemplate how labelling or different modification of the information may doubtlessly have an implication on its possession.

Background and former research

The appearance of AI is considerably altering the best way we do enterprise. It’s reshaping how society as an entire works and it’ll proceed to induce doubtlessly main adjustments for years to return. AI has grow to be an integral a part of many fashionable manufacturing and manufacturing processes. On this setting, massive quantities of knowledge are being mixed with excessive computing energy to result in enhancements in effectivity, capability, and output. To encourage implementation and adoption of AI primarily based options, authorities insurance policies should promote a reliable AI ecosystem and accountable AI R&D actions (Organisation for Financial Co-Operation and Growth (OECD), 2019).

On this mild, many governments worldwide are growing AI Frameworks. The UK Authorities Nationwide AI Technique was lately introduced and revealed by the Workplace for Synthetic Intelligence (Division for Digital, Tradition, Media, and Sport (DCMS), 2021). Such a nationally endorsed ecosystem helps SMEs in transitioning to an AI-focused surroundings and ensures that they continue to be aggressive each throughout the UK and globally. This dialog extends to the position performed by the IP framework, regulatory framework and financing in its wider Innovation Technique (Division for Enterprise, Power, and Industrial Technique (BEIS), 2021).

The UK authorities recognises the significance of AI to the UK financial system by means of the institution of a number of AI targeted establishments. These embrace the Workplace for AI (OAI), a BEIS and DCMS joint initiative, and the AI Council, an impartial professional committee offering an advisory position. The federal government additionally commissioned a number of studies, together with the AI Roadmap (UK AI Council, 2021) and the Nationwide AI Technique (DCMS, 2021) developed by OAI.

Moreover, there was important parliamentary curiosity on this matter, as indicated by the Choose Committee report (Home of Lords, 2018). The work so far has not focussed consideration on the doubtless necessary space of IP, which can present incentives for funding. IP, along with an enforcement regime, could affect funding by providing authorized safety for innovation.

Inquiries to be addressed on this examine

The UKIPO has the next goals and targets in relation to this requirement.

  • overview the literature on the determinants of investments in AI, and the position of IP
  • determine predominant actors of the AI sector within the UK and overseas and the character of their contribution
  • assess the determinants of AI funding
  • present proof on whether or not IP performs a task in incentivising AI funding within the UK
  • replicate variations between IP rights

How we outline AI for the aim of this examine

We now have thought of the definition of AI for the needs of this examine, notably with reference to the subject of funding. There are a number of definitions of AI (Council of Europe, n.d.; Wikipedia, 2021) that advanced over time and with the context by which it’s thought of.

  • Turing Take a look at: The one factor that issues is the exterior behaviour of the machine. Does equipment present clever behaviour? (Turing, 1950)

  • John McCarthy: “Each facet of studying or another function of intelligence will be so exactly described {that a} machine will be made to simulate it”. (McCarthy, 2011)

  • John R Searle: “The appropriately programmed pc with the proper inputs and outputs would thereby have a thoughts in precisely the identical sense human beings have minds.” However “even when a machine completely simulates human behaviour, there’s nonetheless no cause to suppose it additionally has a thoughts.” (Searle, 1980)

For instance, AI has been described as intelligence demonstrated by machines, versus the pure intelligence displayed by people or animals.

Within the UK Authorities Industrial Technique White Paper (BEIS, 2017), AI and the data-led financial system are outlined within the following phrases.

Synthetic intelligence (AI) is a set of applied sciences with the flexibility to carry out duties that might in any other case require human intelligence. These embrace visible notion, speech recognition, and language translation.

Machine studying (ML) is a kind of AI that enables computer systems to study quickly from massive datasets with out being explicitly programmed. In lots of instances, the adoption of ML applied sciences brings advantages of improved productiveness and cost-savings for a enterprise.

Knowledge-led financial system a digitally linked financial system that realises important worth from linked, large-scale knowledge that may be quickly analysed by expertise to generate insights and innovation.

The 2018 Home of Lords Choose Committee on AI adopted an enhanced model of the above definition. It added: “that AI techniques right this moment often have the capability to study or adapt to new experiences or stimuli”. We now have taken this as a helpful foundation in adopting a definition acceptable for this report.

There may be long-standing debate on whether or not machines can really be clever. Nonetheless, what issues right here is whether or not a machine can simulate intelligence to the diploma mandatory to offer a selected set of purposeful capabilities.

There are numerous components concerned in what is taken into account to be intelligence, together with abstraction, self-awareness, studying, planning, downside fixing and demanding pondering. Within the context of an AI a few of these are far more difficult to attain at any time limit than others. Intensive R&D has been making, and continues to make, fast progress tackling these challenges and realising options. AI improvement and its functions depends on the simulation of many components historically contributing to intelligence at any given time to unravel an issue.

AI has typically been categorised on a scale from Slender (or ‘weak’) to Basic (or ‘sturdy’) AI.

  • Slender (or ‘weak’) AI
    Slender AI, which usually performs a single job or a set of intently associated duties, is the most typical type of AI. Examples embrace climate apps, digital assistants, or knowledge evaluation to optimise a selected enterprise operate.

Though the functionalities of the AI in functions could also be slender and should concentrate on rising effectivity, they’ve the ability to rework techniques. Developments proceed to affect how companies work and the way we stay on a worldwide scale. Highly effective examples embrace machine studying for the evaluation of medical knowledge, medical imaging, fraud detection, and driver behaviour monitoring and prediction. Not too long ago, UK-based Monolith AI, along with Imperial School London, is utilizing AI to find out the manufacturability of metallic elements (The Producer, 2022).

Typically, when a way reaches mainstream use (e.g. optical character recognition), it’s now not thought of as AI; this phenomenon is described because the ‘AI impact’.

  • Basic (or ‘sturdy’) AI
    Basic AI, which depends on sentient machines emulating human intelligence, pondering strategically, abstractly, and creatively, remains to be in its infancy. In precept, Basic AI has the flexibility to deal with a spread of complicated duties and maybe exhibit empathy. Whereas machines can carry out some duties higher than people (e.g., knowledge processing), this absolutely realised imaginative and prescient of Basic AI doesn’t but exist in actuality (Müller & Bostrom, 2016; Grace, et al., 2018; Gruetzemacher, et al., 2019; Avin, 2019). That’s the reason human-machine synergy is essential in right this moment’s world, as AI stays an extension of human capabilities, not a alternative.

We be aware additionally that the (Home of Lords, 2018) report refers back to the “AI improvement sector, slightly than an AI sector, on the grounds that it’s principally a selected sub-group throughout the expertise sector which is presently designing, growing and advertising and marketing AI techniques, however a number of sectors of the financial system are presently deploying AI expertise, and plenty of extra will doubtless be a part of them within the close to future.”

We now have taken the same view right here. Slightly than focus particularly simply on finish functions, we glance additionally on the panorama of AI performance developments, one stage again from the top functions. It is a area by which necessary instruments and platforms are being developed and modules on which end-user functions are constructed. AI performance platform examples embrace:

  • pc imaginative and prescient platforms
  • video recognition
  • good robotic expertise
  • gesture management
  • speech recognition
  • speech translation
  • machine studying platforms
  • pure language processing
  • digital assistants
  • suggestion techniques
  • context computing

A 2019 WIPO report exhibits that ML is the predominant AI approach utilized in over a 3rd of AI patents filed worldwide (WIPO, 2019). This AI subject has additionally skilled continued development. The principle areas of software are telecommunications (15%), Life Sciences (12%) and private units/computer-human interactions (11%). Telecommunications is a subject of serious software of AI applied sciences. The principle gamers concentrate on completely different AI fields and functions. Baidu focuses on deep studying, Toyota and Bosch on transportation and Philips and Siemens on life sciences. While China figures extremely total, Chinese language Universities additionally stand out. 17 of the highest 20 educational gamers that patent are primarily based in China (WIPO, 2019). Nearly 1 / 4 of peer-reviewed AI publications in 2019 originate from China alone (Zhang, et al., 2021).

There was 400% enhance within the variety of revealed AI patent functions worldwide over the past decade . There was a doubling of patenting exercise by the UK AI sector over the identical interval, with 88% of AI patents first filed within the UK additionally protected elsewhere. In distinction, the USA and China had 53% and 19%, respectively, of patents protected elsewhere. UK-based candidates and inventors are ranked sixth worldwide by way of absolutely the stage of AI patenting exercise. Total, there are extra US-based candidates submitting for AI-related patents within the UK than UK-based candidates. Know-how associated to neural networks has proven important development the world over over the previous 5 years. The UK particularly has seen bigger proportional will increase than the worldwide developments in AI-related applied sciences associated to move, picture processing and telecommunications.

AI improvement stage

For the needs of this report, we concentrate on AI applied sciences with good potential for financial impact and at a stage of early software requiring monetary funding. Areas for prime potential embrace functions of ML, pc imaginative and prescient, and pure language processing (NLP) as indicated within the chart beneath (Determine 4.1). This chart exhibits that all kinds of AI applied sciences are utilized throughout a variety of industries. Nonetheless, it’s clear that usually talking probably the most extensively used types of AI, regardless of business, are ML, pc imaginative and prescient and NLP.

Determine 4.1 Patent households by co-occurrence of software subject with AI sort (Supply: 2019 WIPO Know-how Pattern report on Synthetic Intelligence

Figure 4.1 - A heat map chart showing the highest number of patents filings in 2019 occurring in Machine Learning, Computer Vision and Speech Processing

We be aware that what is taken into account AI expertise (given the ‘AI Impact’) is itself a shifting goal. We concentrate on mature AI applied sciences which have grow to be routine leaving apart future potential applied sciences. We thus retain a concentrate on the near-term functions that present technical and financial benefit.

IP and intangible property

IP within the UK is most often understood by way of the authorized rights obtainable associated to copyright, designs, patents and commerce marks. Nonetheless, throughout the business world of IP, wider intangible property are often mentioned. These could embrace registered and unregistered IP rights, in addition to different property (branding, software program and coaching). The worth of intangible property is usually poorly mirrored on an organization’s stability sheet (Haskel & Westlake, 2018).

Within the context of this report we contemplate registered and unregistered IP to incorporate the next:

Desk 5.1 Overview of Mental Property Rights

Registered rights Unregistered rights Knowledge/Database rights
Patents Unregistered commerce marks Knowledge*
Utility fashions Unregistered designs Databases
Registered commerce marks Copyright  
Registered designs know-how  
  Commerce secrets and techniques  

The desk above outlines each registered and unregistered IP rights, together with database rights. Whereas knowledge in its pure type as details, figures and knowledge shouldn’t be protectable as an IP asset, the gathering of such knowledge or impartial works within the type of a database, is afforded database proper safety beneath the Copyright and Rights in Databases Rules 1997. Software program code has copyright safety. Not all software program and algorithms could also be patented within the UK however software program and algorithms or different related mathematical strategies are patentable if they supply a technical impact. Traditionally the US has taken a much less restrictive strategy to patent exclusions on this space in comparison with the UK and Europe. In an AI context, novel functions of AI could also be patentable within the UK in addition to primary functionalities however provided that they supply a technical contribution

Enhancements in productiveness and automation additionally permit for improved use of human capital, whereby a extremely educated workforce is refocused on greater worth actions. A 2019 ONS report signifies that the most important capitalised intangible asset within the UK is pc software program and databases (ONS, 2019). Funding in pc software program accounted for a complete funding of just about £31 billion in 2019. AI represents a big subsection of this complete. A current DCMS report discovered that in 2020, 432,000 corporations within the UK spent a complete of £16.7 billion on AI applied sciences (DCMS, 2022). It additionally exhibits that common spend per enterprise will increase with measurement.

Globally, strain to undertake and develop new AI applied sciences will increase, with AI incorporation seen throughout many industries (Judah, 2018; Mckinsey Digital, 2020; Enterprise Insider, 2017). Companies have to make investments in inner infrastructures to see important value financial savings regarding elevated productiveness or elevated enterprise alternatives associated to AI adoption. Massive multinationals within the pharmaceutical, biotech, automotive (Kearney, n.d.) and monetary sectors (OECD, 2021) are additionally making important investments in AI (Mckinsey Digital, 2020). Investments are made each internally and externally. AI start-ups are targets for fast acquisition by recognized AI builders reminiscent of IBM but additionally by different industries reminiscent of prescription drugs and insurance coverage corporations. AI cannot solely provide important value financial savings in productiveness but additionally open new enterprise alternatives. For instance, UK-based Deepmind was acquired by Google in 2014, Mellanox by Nvidia in 2019, and Swedish-based Vi-Well being by UK-based Numan in 2021 (Mobihealthnews, 2021). On the similar time, many profitable UK AI corporations have entered into improvement agreements with business. As an example, AI-led pharmatech firm Exscientia lately accomplished an IPO on Nasdaq with $510m in financing. The corporate additionally has agreements in place with Sanofi (Exscientia, 2022) and Bristol Myers Squibb (Exscientia, 2021).

A 2019 Gartner report primarily based on a CIO Agenda survey confirmed that as much as 14% of enormous corporates use AI in some type (Gartner, 2019). This survey gathered knowledge from over 3,000 respondents throughout 89 international locations and all main sectors in 2018. A 2020 McKinsey survey famous that roughly 50% of companies surveyed adopted an AI performance (Mckinsey Digital, 2020). This survey collected knowledge from 2,395 respondents throughout corporations of various sizes, business, speciality and in numerous international locations. The principle areas of software of AI in such instances are chatbots, course of optimisation, and fraud evaluation throughout the insurance coverage, software program, retail, and IT sectors. These already actively utilizing AI usually tend to allocate extra of their digital budgets for its continued use (Mckinsey Digital, 2020).

AI-related corporations have on common greater valuations and lift extra in particular person funding rounds in comparison with others (MMC Ventures, 2019; IFC, 2020). Nonetheless, the effectiveness of those investments will be improved by higher steerage from governments and regulatory authorities on the design and implementation of AI applied sciences.

AI can be utilized in virtually all industries and at many factors alongside the worth chain. Consequently, AI has attracted important enterprise capital (VC) funding. Between 2012 and 2017, VC investments in AI-related enterprise actions rose from roughly US$600 million in 2012 to over US$14 billion in 2017 (IFC, 2020). There was heavy funding in ML applied sciences throughout a number of industries (IFC, 2019). It is because ML is probably the most developed and mature AI subject able to be utilized to present services or products. That is additionally mirrored in patent submitting knowledge (WIPO, 2019; UKIPO, 2019) and scientific publications (Zhang, et al., 2021).

There was a big enhance in international funding in AI in the previous few years. As indicated beneath in Determine 6.1, it has been estimated that international company funding in AI reached virtually US$ 68 billion in 2020 (Zhang, et al., 2021).

Determine 6.1 World company funding in AI 2015-2020; Supply Statista and Synthetic Intelligence Index Report 2021 (Zhang, et al., 2021)

Figure 6.1 - A bar chart showing from 2015-2020 showing the increasing global corporate investment in AI in US$

China and the USA lead funding within the AI area (Determine 6.2). There may be little or no funding in rising markets, excluding India, though most research have checked out developed economies (IFC, 2020). Complete funding in AI in China reached CNY 175 billion in 2020 (about GBP 20 billion) from CNY 101 billion in 2019 (about GBP 11 billion). Common funding worth elevated from CNY 151 million (about GPB 17 million) to CNY 266 million (about GBP 29 million) (Statista, 2021).

Determine 6.2 Personal Funding in AI in 2020 by nation – Supply: Synthetic Intelligence Index Report 2021

Figure 6.2 - A histogram showing the amount of private investment per country in AI for 2020, with the US in 1st position, followed by China and the UK

Determine 6.2 above means that China is behind the US by way of complete non-public funding in AI in 2020. Nonetheless, the Chinese language authorities investing closely in AI R&D at each an area and nationwide stage. Consequently, a big quantity of funding in AI in China will come each straight and not directly from authorities.

Common funding measurement in China is greater than the USA. US AI start-up companies raised round US$4.4 billion from 155 investments, whereas in China the same quantity was raised in 19 investments (ABI Analysis, 2017). A Chinese language Governmental Assertion “A Subsequent era synthetic intelligence improvement plan” consists of an IP technique for AI (Chinese language State Council , 2017; New America, 2017; Nesta, 2017). This technique gives certainty and confidence for companies to behave and start investing in AI. Since 2017 China set the purpose for important breakthroughs by 2025 and world chief place in AI by 2030 (South China Morning Publish, 2018). China has had a head begin on each AI-related analysis and patenting (Determine 6.3). The previous Pentagon chief software program officer acknowledged that the USA “may have no competing combating likelihood in opposition to China in 15-20 years” (van der Merwe, 2021).

Determine 6.3 Geographic breakdown of origin of world AI journal publications 2000-2020. Supply: Synthetic Intelligence Index Report 2021 (Zhang, et al., 2021)

Figure 6.3 - A chart showing the number of AI journal publications over the last 20 years from China, the US and EU, and indicating the China was active already from about 2005

AI funding within the USA, China and the EU accounted for over 93% of the full non-public fairness funding in AI between 2011-2018 (OECD, 2019). On this interval, UK-directed funding in AI accounted for a complete of 55% of complete European funding. On the idea of funding as a share of GDP, the USA leads at 0.12%, with China and UK at 0.085% every.

Globally, the primary AI hubs for improvement are Silicon Valley, New York, Beijing, Boston, London, and Shenzhen. Silicon Valley is the highest hub for start-up companies and is the seat for a lot of VCs investing on this space.

Though India is the second largest rising market, it beforehand lagged behind different markets reminiscent of China. Nonetheless, this can be altering because it has been anticipated that funding in AI in India will attain roughly US$1 billion by 2023 (Tejaswi, 2021).

Because it stands, India ranks in eighth place by way of non-public funding in AI, behind Europe and notably the USA and China. Nonetheless, a current State of AI report (Deloitte , 2021) suggests that the majority Indian companies plan to extend their AI spend. This report relies on a survey of over 300 senior executives in India. The report exhibits that Indian start-ups are seeing fast returns on earlier investments, giving them a aggressive edge. Companies within the retail & client sector, telecoms and monetary companies sectors exhibit excessive ranges of AI adoption. Nonetheless, these within the pharmaceutical and manufacturing sectors, in addition to hospitality, are lagging behind of their AI adoption. Moreover, India is recorded as one of many international locations experiencing highest development in AI hiring from 2016-2020, in addition to the very best AI expertise penetration charge, forward of each US and China, in second and third place respectively (Zhang, et al., 2021). A 2021 report estimates that AI has the potential so as to add US$957 billion of further gross worth to India by 2035 (Accenture, 2021). In 2019 the Indian authorities fashioned a committee to push for improvement of a proper AI coverage and technique to attain India’s AI ambitions (Zhang, et al., 2021).

educational analysis in AI, China has taken a transparent lead on growing AI-based improvements in universities. In 2019 over 100 of the highest 500 AI-focused educational patent candidates worldwide have been from China (WIPO, 2019). 20 have been from the US and Korea, and 4 from Japan and Europe. Nonetheless, whereas China leads in utilized areas (facial recognition and NLP), it lags behind in {hardware} improvement required to construct this AI ecosystem. Consequently, China presently depends on western corporations reminiscent of Nvidia, Arm, Intel, AMD in addition to quite a few earlier stage AI chip producers. Up to now the primary challenger from China on AI chip manufacturing has been Baidu.

Between 2019 and 2020, there was a transfer in direction of functions in drug discovery, autonomous autos, gaming academic functions (Determine 6.4). These are prone to have grown because of the results of COVID-19 and the related international lockdowns on working habits, leisure actions and economies.

Determine 6.4 Change in focus for AI software fields between 2019 and 2020. Supply -Synthetic Intelligence Index Report 2021 (Zhang, et al., 2021)

Figure 6.4 - A histogram showing changes in the AI private investment focus areas between 2019 and 2020. And indicating significant growth in 2020 in Drug Discovery applications

AI within the UK

The UK is above common in its AI readiness stage (Bughin, et al., 2019; McKinsey & Firm, 2019; Digis, 2021). This index depends on seven AI enablers. Three are AI-specific: AI analysis actions, AI startup funding and automation potential. The remaining 4 are foundations for AI: digital absorption, innovation basis, human capital and ICT connectedness. Total, the UK is doing comparatively nicely by way of AI analysis actions, AI startup funding, digital absorption, innovation basis and ICT connectedness. Nonetheless, it’s lagging behind by way of automation potential and human capital. The UK can also be a pacesetter within the AI rollout in Europe together with Scandinavia. It has been estimated that, by 2030, wider implementation of AI within the UK may lead to a 22% enhance to the UK financial system (present GDP). The UK was ranked second behind Singapore by way of readiness for AI adoption in a authorities AI readiness index (Oxford Insights, 2019). Nonetheless, important funding remains to be required, particularly in scaling up its present digital property and competencies. The examine discovered that the relative readiness of the UK within the roll out of AI depends on seven enablers, specifically:

  • present analysis actions
  • availability of start-up funding
  • potential for automation in business
  • digital absorption
  • innovation basis
  • human capital
  • ICT

Nonetheless, the examine additionally discovered that whereas the UK has good native analysis in AI, funding remains to be decrease than in China and the USA. Moreover, adoption by UK companies stays low. In different phrases, the UK is nice at growing AI options however is missing within the wider scale adoption of those options (Oxford Insights, 2019).

A report commissioned by DCMS discovered that round 15% of UK companies (totalling 432,000) have adopted at the very least one AI expertise. 2% are presently piloting AI and 10% plan to undertake at the very least one AI expertise sooner or later (DCMS, 2022).

This report confirmed that AI options for knowledge administration and evaluation are most prevalent within the UK, with 9% of corporations having adopted them. That is adopted by NLP and era (8%), NLP (7%), AI {hardware} (5%), pc imaginative and prescient and picture processing and era (5%). The IT and telecommunications and authorized sectors presently have the very best charge of adoption, respectively 29.5% and 29.2%. The sectors with the bottom adoption charges are hospitality (11.9%), well being (11.5%), and retail (11.5%).

It additionally confirmed that giant corporations usually tend to undertake a number of AI applied sciences within the UK. Nonetheless, there are progressive corporations utilizing a number of AI applied sciences throughout all measurement classes. Bigger corporations are extra doubtless to make use of 4 or 5 AI applied sciences to help of their enterprise actions.

The extent of funding in AI within the UK, primarily by means of VC, PE and M&A exercise, is greater than elsewhere in Europe, the variety of AI patents filed within the UK is decrease than in France and Germany (Bughin, et al., 2019). Nonetheless, the truth that UK AI actions are usually early stage could also be associated to this. Earlier idea and improvement stage AI improvements are predominantly targeted on the software program and algorithms. As such, they’re extra simply protected by copyright or commerce secret. Later stage AI functions may have a technical impact and, on this respect, can usually be protected by patents.

The AI enterprise sector tends to be dominated by the big multinationals. Nearly all of high-volume patent assignees are primarily based within the USA (Microsoft, Alphabet, IBM, Fb, Yahoo, Intel, Amazon, Apple and Qualcomm) (Cowan & Hinton, 2018). They’re adopted by Japan (Sony, Fujifilm, NEC and Xerox), the European Union (Siemens and Philips) and at last Korea (Samsung).

In 2017, DCMS and BEIS commissioned an impartial report on the expansion of AI within the UK (Corridor & Pesenti, 2017). The report gives suggestions on using IP by universities and analysis institutes producing AI associated innovations. The power to switch the innovations and IP must be improved. This is able to be sure that universities are usually not solely searching for to guard AI innovations but additionally actively searching for to commercialise and monetize these. On this context, the overview recommends standardisation of spin-out actions and customary licensing practices. Moreover, there could also be methods of additional incentivising company buyers at an earlier stage by co-developing standards and phrases.

The event of a focused Nationwide AI Technique (DCMS, 2021) by the UK authorities is seen as a well timed and constructive step. That is consistent with related methods being introduced elsewhere in Europe (Die Bundesregierung, 2020; Rijksoverheid, 2019; Authorities of Eire, 2021). Nonetheless, there are nonetheless basic points to be resolved for the UK to grow to be an AI “superpower” inside ten years (Hayes, 2021). These embrace an elevated openness from AI suppliers about how the core IP works.

The current UK Nationwide Knowledge Technique session can also be related in an AI context (DCMS, 2020). It units out core pillars for greatest use of knowledge within the UK, which embrace knowledge high quality, expertise, availability and its accountable use. Nonetheless, some studies query whether or not the UK’s goal to grow to be an AI superpower goes in opposition to necessary AI regulation on knowledge management (Verdict, 2021). Article 22 of the GDPR regulation, protecting regulation of AI, was integrated into UK legislation by means of the Knowledge Safety Act 2018. However there are issues that following exit of the UK from the European Union, facets of this regulation could also be modified or eliminated (Pierides & Roxon, 2021). This is able to prioritise innovation above regulation. EU regulation and laws surrounding the implementation of AI and knowledge sharing are seen as basic elements of AI innovation and development (European Fee, 2020).

The Alan Turing Institute, the British Requirements Establishment (BSI) and the Nationwide Bodily Laboratory (NPL) will lead a brand new authorities initiative (DMCS and OAI, 2022). The goal of this initiative is to develop AI international technical requirements and sensible instruments for companies to roll-out AI improvements. This is able to thereby contribute to putting the UK in a number one place within the international adoption of AI.

Who’re the primary actors and actor varieties in relation to funding in AI

The completely different actor varieties listed beneath have completely different ranges of involvement within the AI ecosystem. Some companies obtain funding to develop an AI system for their very own use or for others. Others make the funding internally for growing an AI functionality. Likewise, exterior institutional buyers could put money into a variety of AI companies, throughout a number of factors of the worth chain and throughout a number of software areas. The relative significance of AI for particular person companies additionally adjustments throughout actors. For some it is going to be the primary services or products they produce. For others, the AI stack could merely present a small a part of a a lot bigger software-led providing, together with different cutting-edge applied sciences. For the aim of simplicity, we have now grouped actors right into a small variety of basic clusters recognising that there could also be overlapping between, and sub-groups inside, clusters.

We determine 4 kinds of buyers in AI:

  • authorities funding, which for this report we assume to be led by components aside from the IP framework (as authorities defines the framework by means of laws). Nonetheless, the output of educational funding by authorities our bodies will probably be thought of along with IP-related necessities for funding or grant funding

  • corporations making direct inner funding in services or products to create AI-related performance through the use of their very own inner assets

  • corporations making exterior investments, the place they purchase services or products from third events, have enterprise funds, or fund exterior analysis

  • skilled Buyers reminiscent of enterprise capital or seed funding buyers, the place the funding is primarily for monetary acquire

We search to know whether or not funding selections are affected by the perceived capability to guard or safe them by means of the present UK IP framework.

As well as, funding will be thought of alongside the AI part worth chain as buyers could put money into a number of of the elements of AI. We determine six elements throughout the AI worth chain:

Chips and {hardware}. Firms growing semiconductor chips that assist AI functions (examples can be Intel and Qualcomm).

Platform and infrastructure. A {hardware} structure or software program framework that enables AI software program to run (reminiscent of Google’s AI platform, Microsoft Azure and TensorFlow).

Frameworks and algorithms. AI frameworks permit sooner and simpler improvement of AI functions (examples can be Microsoft CNTK and Caffe).

Enterprise options. A class of enterprise software program that harnesses superior AI strategies to steer digital transformation (examples can be Digital Genius and Afiniti).

Vertical Trade Options. Vertical (particular market sector) AI options (in particular market sectors) are about searching for to unlocking completely new alternatives slightly than simply optimizing present alternatives. Examples can be: Benevolent AI in healthcare; Zest finance in finance and insurance coverage; and FarmersEdge in agriculture.

Corporates. Massive firms (reminiscent of Ford and Boeing in automotive and transport; and Disney in retail and media).

We additionally thought of whether or not the kind of AI software (see 4.1) could present a 3rd dimension. Nonetheless, we discovered no proof from our interviewees that this issue made any related distinction.

In abstract, funding determinants, together with the IPO framework, are prone to be depending on mixtures of the kind of investor and the goal market sector. Because of this, our interviews coated a spread of various varieties and sectors.

A current examine by Deloitte (Deloitte Growth LLC, 2021) analyses MedTech start-ups, a lot of which incorporate AI capabilities. The examine exhibits that early stage funding is especially comprised of founder self-investment, buddies & household and angel buyers. VCs and institutional buyers or company enterprise arms play a extra marginal position. Enterprise Scanner knowledge (2019) confirmed that the majority VC funding focuses on ML. A report excessive $44bn had been invested by Q3 2021 with 72% at seed and early stage. There have additionally been report exits in 2021 of 94 AI corporations by Q3.

Drivers of funding in AI

Abilities and expertise

An necessary determinant for AI improvement within the UK is the big pool of AI and broader software program expertise obtainable. That is partly because of the success of the UK in attracting expertise from different international locations –being a fascinating vacation spot for these growing AI improvements. In 2019, the UK had the second largest variety of software program builders in Europe (15% of the full). London is the most important expertise hub in Europe. Importantly, the UK has a further profit in attracting innovators within the AI area, specifically by being an English-speaking nation. When US builders enter the European market, they usually transfer to the UK (Bughin et al., 2019). This has been additional strengthened by the muse of a brand new AI Council in 2019, which established doctoral coaching actions in AI all through the UK. Equally, the UK counts necessary analysis institutes, together with the Alan Turing Institute, the Leverhulme Centre and the Ada Lovelace Institute, all specializing in AI R&D.


Regardless of sturdy educational analysis resulting in a big potential workforce and important funding within the space, the UK generates a low variety of profitable AI associated companies in comparison with the USA (Bughin, et al., 2019). The Choose committee report on Synthetic Intelligence gives suggestions to deal with this shortfall (Home of Lords, 2018). It notes that universities and analysis institutes have to introduce clear and commonplace practices for licensing IP and forming spin-out corporations. There additionally seems to be an related lag in adoption of AI at scale within the UK in comparison with Europe and the USA, as indicated in Determine 7.1 (Bughin, et al., 2019).

Determine 7.1 AI expertise diffusion within the UK versus Europe and the USA primarily based on % share of corporations adopting AI at scale. Based mostly on supply (Bughin, et al., 2019)

Figure 7.1 - A bar chart showing how AI technology diffusion in the UK lags behind both Europe and the USA

Scale-up and funding tradition

There seems to be a pattern amongst early-stage UK SMEs to plan an exit by means of massive multinational acquisition slightly than search to develop organically. Early-stage UK deep tech companies are inclined to exit earlier than a second or third spherical of funding, in comparison with their US counterparts. For instance, within the Tech Nation Future Fifty Programme for digital companies (Tech Nation, 2001), the variety of exits accomplished by startups accelerated by Future Fifty reached 55 as of mid-2021 (Statista, 2022). Nonetheless, that is additionally partially because of the difficulties skilled by UK-based deep tech development corporations in making an attempt to realize series-B funding. Many transfer their base to the USA, particularly because of the ease of entry to a well-established funding ecosystem (British Enterprise Financial institution, 2021).

The UK market could also be seen as too small with restricted entry to funding sources to scale up. The European market has various laws and language necessities. In distinction, the US market represents a big homogenous and English-speaking market with important capital obtainable and enormous corporations seeking to purchase start-ups.

This will even be a part of a worldwide pattern of fast exit for AI corporations by means of M&A, as indicated in Determine 7.2 (CB Insights, 2021). Moreover, it has been estimated that in 2021 the VC worth from IPOs of AI and ML semiconductor corporations reached over US$10 billion (TechMonitor, 2021).

Determine 7.2 Price of firm exits for AI start-ups worldwide from 2019-2021 and kind of exit. Based mostly on supply – CB Insights Synthetic Intelligence in Numbers Report (CB Insights, 2021)

Figure 7.2 - A bar chart showing global AI start-up exists from 2019-2020 and whether exit was via M&A or IPO

Regulatory framework

In a written proof report back to the Home of Lords, Balderton Capital indicated that the primary dangers to AI improvement and use (Clever, 2017). Balderton Capital considers that failure to introduce a correct regulatory framework may even see worldwide operators make the most of gaps within the UK market. That will undermine the UK human capital base and doubtlessly enhance nationwide safety dangers. This will already be occurring as a lot of the patented UK AI-related R&D actions are outsourced or owned by massive multinationals (Cowan & Hinton, 2018).

At a nationwide stage, a number of actions will be taken to strengthen and regulate the AI improvement and use (World Financial institution, 2021). This may be finished by means of the institution of a transparent regulatory framework and technique. This would come with strengthening the nationwide or regional IP regime the place required, modifying competitors guidelines/coverage in addition to regulatory sandboxing. The state can also present coverage readability to companies, as China has finished (European Fee, 2021; Chinese language State Council , 2017; Home of Lords, 2018). By offering regulatory certainty, it’s attainable to allow companies to proceed with confidence.

Availability of coaching knowledge

An extra limiting issue, for international locations such because the UK, is the inadequate availability of distinctive coaching knowledge. This is because of quite a few causes together with failures to recognise the worth of such knowledge, in addition to knowledge safety concerns. Even when a singular AI expertise or answer is developed, there could also be an absence of appropriate coaching knowledge obtainable within the UK. This contrasts with extra open markets such because the US, or state-controlled markets reminiscent of China. Thus, there could also be alternatives for big corporations to license their knowledge to be used by third events. Authorities departments and organisations can also be capable of provide using massive quantities of (anonymised) knowledge to UK companies. This strategy could assist keep away from the scenario noticed in Canada (Cowan & Hinton, 2018). Though important authorities funding was made in AI analysis, a lot of the remaining patented innovations are owned by massive multinationals not headquartered in Canada.

IP safety

A current examine exhibits that the widespread adoption of AI throughout a number of industries and sectors could affect the requirement for IP safety of AI expertise (Hilty, et al., 2020). That is due to the very fast innovation cycles and fast product or system alternative. It could not make sense to put money into formally defending a software program system by patent if it turns into out of date earlier than the patent is even granted. Consequently, the usual assumption that IP safety is itself an incentive for additional innovation could now not be legitimate.

Moreover, the problem in defending software program innovations in a well timed method, particularly with a patent, could restrict software program start-ups’ capability to draw funding. It could additionally hinder their capability to generate revenues by means of business use of the IP rights, for instance, licensing offers. Consequently, IP safety within the AI area could also be justified to guard the rewards from funding in AI to create and invent.

This justification resonates strongly with many within the funding group. A robust IP portfolio, comprising a combination of various IP rights, demonstrates enterprise maturity of the borrower. Importantly, it additionally gives recourse for the investor to get well a few of its funding if the enterprise is unsuccessful.

Nonetheless, main technological adjustments are minimal attributable to very fast improvement instances and continued incremental enhancements. Most adjustments lead the elevated calculation pace and improved coaching knowledge units. Regardless of this, unique and novel AI software program and algorithms might be protected by copyright or commerce secrets and techniques and associated technical contributions by patents.

Nations growing AI, together with the UK, might want to develop coherent nationwide methods and frameworks to keep away from IP leakage (Cowan & Hinton, 2018). This consists of offering mechanisms reminiscent of patent swimming pools or specialist funding funds to make sure they will stay aggressive in opposition to the US and China. On this context, there stays the query of whether or not the UK IP framework in a wider definition may assist mitigate a few of these perceived dangers.

Literature survey-based conclusions

Based mostly on our survey of the literature so far, in addition to an rising understanding of financial actions in AI, it’s attainable to attract quite a few conclusions:

  • machine studying is by far the main space within the AI area and is the kind of AI most generally adopted throughout business. That is largely as a result of in every case it’s fixing an business or enterprise particular want, thereby rising productiveness, and lowering prices

  • within the UK, most funding in start-ups and SMEs comes from the founders, household and buddies, angel buyers and, in some cases, VC corporations. Nonetheless, at an early stage, VCs are prone to be extra restricted within the quantity of funding they supply and the kind of corporations they’re keen to put money into (GrantTree, 2021)

  • improvement of, and consequently funding in, AI is predominantly targeted on the USA and China. Nonetheless, throughout the European context, the UK is main improvement exercise and funding in AI

  • partnership and cooperation throughout AI analysis are restricted, whereas battle and litigation additionally seem like fairly low stage (WIPO, 2019)

  • moreover, compared to a number of international locations together with Germany, France, Japan and South Korea, the patent output of the UK is decrease (Bughin, et al., 2019). This can be as a result of the UK seems to wrestle to show AI analysis into profitable enterprise (Home of Lords, 2018). A number of components could clarify this. Some UK actions are early-stage and never sufficiently developed for patenting. It could even be attributable to UK corporations selecting to file within the USA or Europe as a substitute of the UK. Nonetheless, the lifecycle of software program is usually brief (~5 years) and early-stage developments will be outdated by new releases and upgrades

  • within the broader context of AI funding within the UK and likewise additional afield, it’s clear that, whereas IP could also be a part of the equation, it’s by far not a very powerful issue (Bughin, et al., 2019). Entry to human expertise capital, entry to knowledge for coaching AI, and a dependable authorized and regulatory system are additionally essential conditions for funding

  • an organization’s choice to put money into the UK appears to be much less influenced by the UKIPO/EPO framework than by associated skilled authorized and consulting companies. These embrace UKIPO IP audit and IP entry grants enabling business recommendation for early-stage companies (Mathys & Squire LLP, 2021). Mediation companies, IP rights enforcement, belief within the judicial system, and the overall regulatory surroundings additionally appear to matter extra

  • further financial levers, together with tax incentives, may also affect bigger corporations’ selections to find within the UK (TechUK, 2021; BEIS and DCMS, 2019; OECD, 2021). Tax incentives embrace patent packing containers or R&D tax credit

  • an open response to a Authorities Session paper highlighted the necessary position of IP in cementing the UK place in AI and attracting further expertise and funding (Lord Clement Jones, 2022). They contemplate a number of choices for modification of the present IP framework, a few of which replicate these of respondents we interviewed. These choices included introducing a lowered copyright safety time period for AI generated copyright works and leveraging enterprise incentives. In addition they talk about different choices to incentivise knowledge homeowners to licence coaching knowledge to be used

  • AI remains to be a closely white male dominated space. Solely 22% of all AI staff within the UK are feminine. Nonetheless, 45% of all UK AI start-ups have a feminine co-founder (Aashind, 2021). Moreover, based on the 2021 AI Index Report, 45% of North American AI graduates interviewed have been white (Zhang, et al., 2021). This demonstrates an absence of range within the AI area that will affect on alternatives for the widest attainable improvement and implementation of AI.

In-person interviews and questions

In an effort to acquire probably the most from the business interplay, nose to nose interviews have been carried out for all of the discussions. Sometimes, these assembly lasted 60 minutes or over, utilizing MS Groups or Zoom calls. It must be famous that through the examine interval, varied Covid-19 restrictions have been in place. Nonetheless, the face-to-face strategy was discovered to allow very open discussions. It additionally allowed interviewees to precise opinions that might have been constrained by a written format.

Two questionnaires have been designed to facilitate the interviews with these particular stakeholder teams within the AI worth chain. One questionnaire was directed to firm actors (each SMEs and Company) and one other to buyers. The questionnaires are offered in Appendix 2 of this report. The goal was to know their respective views on actions and funding in AI. It was additionally necessary to make sure that the questions and format of the interviews have been sufficiently just like keep away from creating any bias.

Within the first occasion a sequence of stakeholders throughout the next teams have been chosen:

In every goal organisation, we usually sought to interview the CEO, CTO, Chief IP officer or related senior individual. It is because they’re almost definitely to have an consciousness of the event, use and safety of AI within the enterprise. They’re additionally almost definitely to have an opinion on the primary determinants within the development of AI primarily based actions and funding. A listing of the stakeholder organisations, companies and buyers interviewed is offered in Appendix 1.

These stakeholders have been approached straight, the place direct electronic mail addresses or cellphone numbers have been obtainable. The place no direct contact particulars have been obtainable, we sought to hook up with related teams through LinkedIn and different platforms. The place the potential interviewee expressed an curiosity in taking part, we then adopted up through electronic mail and/or cellphone calls. Interview invites have been despatched out in batches, with quite a few invites despatched per week and interviews over the following two weeks.

A complete of 96 people from completely different companies and organisations have been approached, with 34 responding and 21 interviewed. This equates to an total response charge of near 22%. 13 respondents indicated that they didn’t want to take part and weren’t interviewed. An summary of the breakdown of these interviewed is proven in Desk 9.1 beneath.

Desk 9.1 Overview of variety of respondents, interviewees, and responses charges on this examine

  Numbers %
Complete contacted 96  
Total response charge 34 35%
% Interviewed (from contacted) 21 22%
% Interviewed (from responded 21 62%
% Declined 6 6%

The response charge amongst corporations and buyers was related at 33% and 31%, respectively. Over half (61%) of respondents have been corporations and 13% have been buyers (Determine 9.1). Buyers included a mixture of angel buyers, enterprise capital and company enterprise. A smaller proportion of interviewees additionally belonged to accelerators (9%), universities (9%) and different AI associated organisations (9%). Amongst corporations, 42% have been massive, 36% SME and 21% micro. A major majority of respondents interviewed have been male (86%), as indicated in Determine 9.2.

Interview individuals

Determine 9.1 Cut up of respondents interviewed on this examine based on entity sort

Figure 9.1 - A pie chart showing the breakdown of interviewed actor type, with companies being the largest group at 61%.

Determine 9.2 Gender cut up of these interviewed

Figure 9.2 - A pie chart showing the gender breakdown of those interviewed, indicating that 86% were male and 14% were female

Regardless of the big selection of various business pursuits amongst stakeholders interviewed, responses tended to converge.

It has been highlighted in a number of interviews undertaken as a part of this examine that it isn’t all the time attainable to guard AI innovations by patent. Many software program and AI corporations, particularly start-ups and SMEs, typically depend on a mixture of copyright and commerce secret utilized in tandem with confidentiality agreements.

Nonetheless, this has additionally raised the next questions:

Is the present set of IP rights adequate to guard the improvements arising from this quickly evolving AI subject and software program extra broadly?

Is it time to contemplate a brand new sort of IP proper targeted on the safety of AI innovations (Pinsent Masons , 2021; Deamer, 2022)?

Might amendments be made to present copyright laws to deal with obstacles particular to present software program and AI developments?

The UKIPO launched a session on the subject of copyright and patents in AI in October 2021 (UKIPO, 2021).

Such questions have been raised in a number of studies and research. A current 2020 examine by the USPTO sought to reply a few of these questions on patenting AI innovations (USPTO, 2020). On this examine, a number of respondents indicated that present IP rights have been adequate, with any gaps being crammed with appropriate confidentiality agreements. Others indicated {that a} new type of IP proper is likely to be wanted to guard knowledge related to ML in bigger corporations. In addition they beneficial to make this knowledge obtainable to others to permit different market entrants to coach their very own AI.

Entry to exterior knowledge has been raised by a big company interviewee. This firm makes use of commercially offered buyer knowledge from different events and labels for coaching functions. As this knowledge is steadily handed by means of a series, labelled and due to this fact steadily modified (in some instances by quite a few events), an necessary query was raised about: “the purpose at which it now not resembles the unique knowledge offered and as such, the place quite a few events may have modified that knowledge by means of the chain, who truly owns the rights to that knowledge?”

Conclusions from in-person interviews

This part highlights the primary conclusions derived from the interviews with stakeholders undertaken as detailed above in part 9. The principle insights from the interviews are summarised beneath and damaged down into particular subjects: IP, human expertise capital and staffing, funding, and AI software. Capturing views from focused and stratified face-to-face interviews has offered many insights that might not have been obtained by means of responses to written questionnaires.

We be aware that having undertaken greater than 20 interviews, the identical points and subjects have been repeated by completely different interviewees in numerous sectors.

Mental property

  • nearly all of micro and SME corporations interviewed perceived that they might not get patents for his or her software program developments. Know-how sector individuals had a usually perceived false impression that ‘core AI software program shouldn’t be patentable’. The same remark was made in a lately revealed DCMS and OAI commissioned report on AI (Westgarth, et al., 2022). This report additionally notes that the worth of AI is derived from its software to unravel an present downside. But, patent safety for AI-related innovations is obtainable throughout all expertise areas. Patent innovations embodying algorithms or software program are allowable in the event that they make a technical contribution. The UKIPO, just like the EPO, grants many hundreds of patents for pc and software program carried out innovations together with within the subject of AI (UKIPO, 2019). Thus, they do present safety for innovators on this subject. The UKIPO’s Handbook of Patent Observe gives in depth steerage on the patenting of software program and pc carried out innovations

  • corporations that have been purely builders for third events working on the pre-application stage felt restricted of their capability to guard their innovations by patents. Conversely, these growing their very own AI system to unravel a selected downside felt they demonstrated technical results, which might be extra simply patented

  • amongst micro and SME corporations we noticed an imbalance within the understanding of the patent course of, copyright and commerce secrets and techniques (see Determine 10.1). Patents have been probably the most readily understood with copyright and commerce secrets and techniques being a lot decrease down in each understanding and perceived worth. While in early software program improvement levels, corporations relied on copyright and commerce secrets and techniques and not using a full understanding of the rights or authorized safety they offered

Determine 10.1 Perceived imbalance in understanding of relative significance of various IP rights for various actors. Scales picture beneath licence from

Figure 10.1 - An image of a balance, weighted down to the left due to patents, which are viewed as more important by investors, versus the others side representing copyright and trade secrets, often viewed as more appropriate by AI SMEs
  • start-ups in early (pre-application) AI improvement levels have a dilemma since their predominant property are software program which isn’t patentable as such (within the UK). With out patents, it’s tougher for buyers to acquire an goal measure of the inventiveness of software program code resulting in a patent bias for buyers. It’s then more durable for some early-stage software program and AI corporations to boost cash and results in ‘patent something or make filings’ to fulfill buyers. But, in an early-stage firm the finances required for patent submitting is troublesome to justify

  • some corporations really feel that their US counterparts have a bonus in capability to patent software program, together with AI

  • the view from some SME’s, particularly early-stage corporations, is that they should depend on copyright and commerce secrets and techniques. Nonetheless, their understanding of these authorized rights is low and the expertise investor group doesn’t view these rights as very safe. This contrasts with corporations within the media sector that view copyright as their predominant asset, which can also be nicely understood by media buyers

  • it seems that the expertise /science sector (patent-centric) and media sector (copyright-centric) are actually merging with the arrival of AI. The assist infrastructure by way of service suppliers and experience has developed individually over time to assist every of the sectors with little overlap. Thus, the expertise/science sector must entry service and experience historically targeted on the media sector

  • one respondent indicated that slightly than growing a brand new proper for AI, it is likely to be acceptable to easily create a parallel model of copyright legislation making use of to AI. The present copyright legislation would shield the written phrase, imagery and many others whereas a brand new various copyright would shield software program, AI and related expertise property

  • corporations’ reliance on completely different components of the UK IP framework adjustments as they modify in measurement, as illustrated in Determine 10.2

  • corporations need consistency in international IP frameworks.

  • AI corporations have been typically based by means of college spinouts or by PhD college students. One respondent talked about that the UK is favoured because of the decrease prices of patent filings and excessive stage of experience amongst UK patent attorneys. Nonetheless, the overall view amongst UK SMEs interviewed was that the route of submitting for patent safety is usually prohibitively costly and gradual. Consequently, most SMEs have chosen to maintain filings to a minimal to cut back prices

  • we be aware, from chatting with quite a few interviewees, that it isn’t clear the place the UK sits on commerce secret safety. There have been questions on whether or not the present Commerce Secrets and techniques (Enforcement) Rules 2018 (Crown Copyright, 2018) will stay as such following departure of the UK from the EU. Different questions associated to commerce secret safety within the UK compares to the USA (US Authorities Publishing Workplace, 2016) or to the EU (European Parliament, 2016). One interviewee indicated that the existence of this safety must be higher highlighted

Determine 10.2 Notion of worth and good thing about IP altering relying on the actor and stage of enterprise

Figure 10.2 - A table indicating the changing importance of different types of IP and the different purpose of IP across different actor types and across different development stages of a company, from Small-Medium Business through to large corporate

Human capital and staffing

  • the UK ecosystem is usually seen positively with good supply of grants, funding, and expertise

  • human assets and expertise are basic to all sectors, firm sizes and buyers. The interviewees have been unanimous that resourcing was a very powerful facet for AI

  • London and Cambridge have been cited by worldwide corporations as AI hubs. Ecosystems have been cited as being necessary in attracting additional expertise, by interviewees each inside and outdoors the UK. Ecosystem are outlined as co-location of corporations, universities and analysis

  • in lots of instances, certified AI workers are troublesome to seek out and, in most companies, it isn’t adequate to solely have AI expertise. Most companies additionally want somebody with expertise of specialist fields by which AI could also be utilized, reminiscent of medical imaging or drug discovery. This multi-disciplinary strategy permits for probably the most novel makes use of of AI

  • many bigger SMEs commonly current at conferences, publish articles and take part to hackathons to showcase their actions and talents alongside massive multinationals reminiscent of Google. It is a requirement for a lot of AI researchers searching for to ascertain their names within the subject and eager to share their work outdoors the corporate. This will not all the time sit simply alongside firm IP coverage requiring a excessive diploma of secrecy safety round commerce secrets and techniques

  • many companies interviewed had AI groups comprising individuals within the UK and in different international locations. Visa points after the UK left the EU could have made journey tougher to and from Europe. Nonetheless, the convenience at which individuals have tailored to working remotely through the COVID-19 pandemic has enabled improvement groups to be situated around the globe.

Funding and funding

  • most VC buyers wish to see corporations with patents, while company buyers are extra focused on expertise and the group

  • many SMEs view patents as necessary of their capability to draw and fulfill buyers. Nonetheless, different rights reminiscent of copyrights or commerce secrets and techniques are much less recognised by buyers, particularly VCs

  • the UK grant funding system, together with Innovate UK, Innovate UK Edge grants and UKIPO grants, was seen as very useful at early improvement stage

  • the nice majority of start-ups plan to exit as quickly as attainable. The UK shouldn’t be seen as having the market measurement or capital availability to scale up a enterprise. The USA is seen as having each market measurement, capital and enormous corporations seeking to purchase start-ups. Challenges in scaling up for UK progressive companies can also be highlighted within the Proof paper of the UK Innovation Technique (BEIS, 2021)

  • quite a few corporations utilizing AI for drug design and binding web site identification have been interviewed and this seems to be a rising space. That is mirrored in an AI Index Report which states that the Drug Discovery subject was the realm receiving most non-public AI funding in 2020 (Zhang, et al., 2021).

  • AI is being extensively utilized within the MedTech sector, particularly in drug-discovery, doubtlessly spurred by the COVID-19 pandemic


  • entry to knowledge is crucial for all sectors. But, bigger established corporations have entry to extra of their very own knowledge whereas start-ups typically have restricted entry to knowledge, which might gradual improvement. Bettering widespread entry to knowledge or knowledge sharing would assist start-ups speed up developments and create extra of a stage enjoying subject. Massive corporations have a possible monopoly on predominant knowledge assets.

  • most SMEs interviewed famous that they’d adequate entry to AI coaching knowledge by means of commercially obtainable datasets in addition to in-house synthesised datasets. In addition they famous {that a} authorities supported, or business supported, pooling of anonymised knowledge for coaching can be strongly welcomed. However, they recognise that it could be troublesome to influence business to take part. In that respect, China has a transparent benefit in its capability to assemble and use massive quantities of coaching knowledge relative to western nations (WIPO, 2019). This suggestion is consistent with suggestions made in an impartial report Commissioned by DCMS and BEIS (Corridor & Pesenti, 2017). This report recommends that coaching knowledge must be extra available and authorities and business ought to goal to develop “knowledge trusts” to facilitate knowledge sharing

  • many corporations use a combination of commercially obtainable (bought) knowledge units and synthetically created knowledge units for coaching their AI system. Bigger company companies could have entry to massive quantities of knowledge in-house which they personal, thereby giving them a bonus. That is notably the case within the pharma and biotech industries

  • in sectors the place private data is extensively used, as within the medical sector, a selected knowledge possession construction is most popular. On this case, people would retain possession of their private knowledge however permit to be used in sure permitted circumstances

  • an interviewee within the client product and medical software subject indicated they’d entry to massive quantities of knowledge by means of the units their purchasers use. Nonetheless, massive quantities of knowledge require cleansing and labelling to be exploited in helpful methods. In some instances, regulatory controls additionally restrict entry to doubtlessly obtainable knowledge

Relevance of being within the UK

  • the UK ecosystem and availability of funding is usually appeared upon favourably in comparison with elsewhere in Europe. Nonetheless, corporations would usually search funding outdoors the UK for scaling a enterprise

  • distant working implies that many group members don’t have to be within the UK with nearly all of respondents having worldwide groups (Determine 10.3)

  • a big company investor for an Asian multinational famous that they understand little variations between the UK and Europe (the UK is seen as a part of Europe) by way of funding. The determinant is, usually, an excellent and well-developed expertise with a business software

  • one other massive company indicated that any IP safety change for AI-related innovations and creations would want to happen at a bigger worldwide or European stage. A singular “particular” AI proper offered within the UK alone can be of little curiosity or use to many internationally working companies

Determine 10.3 Typical UK start-up AI assets throughout Europe with a number of group members and groups scattered round Europe. European map picture provided beneath licence from

Figure 10.3 - A map of Europe showing a typical AI start-up being headquartered in UK but with co-founders, individual team members and teams located in other countries throughout Europe

Total headline conclusions

  • the UK IP Framework shouldn’t be a basic consideration for funding by the individuals on this examine

  • different components presently dominate funding selections. The supply of human capital notably PhD graduates, AI know-how throughout the UK workforce and the popularity of expertise hubs and universities are notably necessary. Authorities grants and entry to early-stage funding are additionally important

  • the individuals regarded the UK as an excellent location to put money into a start-up AI enterprise for a number of causes. The UK has a relative energy of the above components in relation to different areas. Particularly, nearly all of UK individuals had graduated from UK universities with AI {qualifications}

  • the UK IP framework was broadly considered trusted and dependable and due to this fact not of concern concerning funding selections. Nonetheless, it must be famous that early stage start-ups didn’t have a lot ‘in-house’ expertise of utilizing the IP framework

  • individuals from the expertise sector famous a propensity for buyers to require patents to assist an funding spherical. But, their predominant asset, AI-related software program, is seen as not patentable. This latter level is a standard false impression, because the presence of a technical impact does present potential patentability

  • buyers in early-stage AI corporations seen patents as providing a third-party opinion on the distinctiveness of the corporate providing. In some instances, it was thought of an ‘asset insurance coverage’ which might be bought within the occasion the corporate ceased buying and selling

  • throughout the early-stage expertise sector corporations, there was an excellent understanding of the patent system, however a a lot weaker understanding of copyright

  • all of the individuals had a comparatively weak understanding of the IP framework associated to commerce secrets and techniques

  • expertise sector corporations have been usually served by patent legal professional corporations, with a concentrate on acquiring patents

  • individuals from the media sector have been very accustomed to the copyright framework and would usually use the companies of a copyright solicitor

  • there seem like two communities of curiosity utilizing and/or growing AI expertise and the UK IP framework. The expertise sector group tends to concentrate on patents and depends on patent legal professional corporations whereas the media sector is served by copyright solicitors

  • presently there’s a historic hole (See Determine 10.4) between these communities with little mixing of experience. Bringing collectively mixed patent and copyright experience to concentrate on the AI sector can be notably useful for early-stage corporations. This fashion, they might develop business methods to grasp the total potential of their property each by way of safety, fundraising and subsequent exploitation

Figure 10.4 - Venn diagram showing overlaps between software, patents and copyright areas
  • the UK has established a powerful popularity as a spot to incubate new applied sciences, together with AI. The UK is seen as too small a market to scale-up and too exhausting to get the bigger rounds of funding wanted for main long-term development. Many younger corporations select early exit or shifting to the USA. It’s a far bigger market and has a popularity for threat and because the place to get massive funding. This poses the danger of the UK shedding out within the larger-scale financial advantages of AI enterprise development to different a lot bigger economies.


Aashind, A., 2021. Curious Synthetic Intelligence Statistics UK [2021]. [Online]
[Accessed 26 January 2022].

ABI Research, 2017. Synthetic Intelligence Funding Monitor 2017. [Online]

Accenture, 2021. Rewire for Sucess – Boosting Indias AIQ, Bangalore: Accenture.

Avin, S., 2019. Exploring synthetic intelligence futures.. Journal of AI Humanities.

BEIS and DCMS, 2019. Coverage Paper – AI sector Deal. [Online]
[Accessed 06 December 2021].

British Enterprise Financial institution, 2021. Small Enterprise Fairness Tracker 2021, Sheffield: British Enterprise Financial institution.

Bughin, J. et al., 2019. Synthetic intelligence in the UK: Prospects and challenges, London: Mckinsey & Firm.

Business Insider, 2017. How 9 industries are being reworked by UK innovation around the globe. [Online]
[Accessed 29 November 2021].

CB Insights, 2021. Synthetic Intelligence In Numbers Q2 2021. [Online]
[Accessed 26 January 2022].

Chinese State Council , 2017. A Subsequent Technology Synthetic Intelligence Growth Plan. [Online]
[Accessed 30 November 2021].

Cornell College, INSEAD & WIPO, 2020. The World Innovation Index 2020: Who will finance innovation?, Ithaca, Fontainebleu & Geneva: Cornell College, INSEAD & WIPO.

Council of Europe, n.d. What’s AI? [Online]
[Accessed 13 10 2021].

Cowan, P. & Hinton, J., 2018. Mental Property and synthetic intelligence: What does the long run maintain? , London: Regulation Enterprise Analysis .

Crown Copyright, 1997. The Copyright and Rights in Databases Rules 1997. [Online]
[Accessed 20 January 2022].

Crown Copyright, 2018. The Commerce Secrets and techniques (Enforcement, and many others.) Rules 2018. [Online]
[Accessed 06 December 2021].

DCMS & OAI, 2022. New UK initiative to form international requirements for Synthetic Intelligence. [Online]
[Accessed 20 January 2022].

Deamer, L., 2022. New report claims fixing copyright would enhance AI startup scene. [Online]
[Accessed 27 January 2022].

Deloitte , 2021. State of AI in India, Mumbai: Deloitte Touche Tohmatsu India LLP.

Deloitte Development LLC, 2021. Can medtech start-ups present us the place the business is headed? Insights from Medtech Innovator and business leaders. [Online]
[Accessed 11 10 2021].

Die Bundesregierung, 2020. Synthetic Intelligence Technique of the German Federal Authorities, Berlin: Die Bundesregierung.

Digis, 2021. State of Synthetic Intelligence in Europe: Statistics 2021. [Online]
[Accessed 10 February 2022].

DMCS and OAI, 2022. New UK initiative to form international requirements for Synthetic Intelligence. [Online]
[Accessed 20 January 2022].

European Fee, 2020. White Paper On Synthetic Intelligence – A European strategy to excellence and belief, Brussels: European Fee.

European Commission, 2021. Europe match for the Digital Age: Fee proposes new guidelines and actions for excellence and belief in Synthetic Intelligence. [Online]
[Accessed 09 February 2022].

European Parliament, 2016. DIRECTIVE (EU) 2016/943 OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL. [Online]
[Accessed 02 12 2021].

Exscientia, 2021. Bristol Myers Squibb workouts choice to develop Exscientia’s AI-designed, immune-modulating drug candidate. [Online]
[Accessed 13 January 2022].

Exscientia, 2022 . Exscientia and Sanofi Set up Strategic Analysis Collaboration to Develop AI-driven Pipeline of Precision-Engineered Medicines. [Online]
[Accessed 13 January 2022].

Gartner, 2019. High developments on the Gartner Hype cycle for Synthetic Intelligence, 2019. [Online]
[Accessed 15 10 2021].

Government of Ireland, 2021. AI – Right here for Good: Nationwide Synthetic Intelligence Technique for Eire. [Online]
[Accessed 05 December 2021].

Grace, Okay. et al., 2018. When will AI exceed human efficiency? Proof from AI consultants. Journal of Synthetic Intelligence Analysis, Quantity 62, pp. 729-754.

GrantTree, 2021. UK Begin-up funding information. [Online]
[Accessed 27 January 2022].

Gruetzemacher, R., Paradice, D. & Lee, Okay. B., 2019. Forecasting Transformative AI: An Professional Survey. arXiv preprint, p. arXiv:1901.08579..

Corridor, W. & Pesenti, J., 2017. Rising the substitute intelligence business within the UK , London: Crown Copyright.

Haskel, J. & Westlake, S., 2018. Capitalism with out Capital – The rise of the intangible financial system. 1 ed. New Jersey: Princeton College Press.

Hayes, J., 2021. UK’s AI masterplan vs actuality. [Online]
[Accessed 5 December 2021].

Hilty, R., Hoffmann, J. & Scheuerer, S., 2020. Mental Property Justification for Synthetic Intelligence. Mental Property Justification for Synthetic Intelligence. Draft chapter. Forthcoming in: J.-A. Lee, Okay.-C. Liu Synthetic Intelligence & Mental Property, Oxford, Oxford College Press, 2020, p. 20.

HM Authorities, 2017. Industrial Technique – Constructing a Britian match for the long run , London: Crown Copyright.

HM Authorities, 2021. Nationwide AI Technique , London: Crown Copyright.

Home of Lords, 2018. Choose Committee on Synthetic Intelligence – AI within the UK: prepared, keen and in a position?, London: Home of Lords.

IFC, 2019. Synthetic Intelligence: Funding Tendencies and Chosen Trade makes use of, Washington DC: Worldwide Finance Company.

IFC, 2020. Synthetic Intelligence in Rising Markets, Washington DC: IFC.

Worldwide Finance Company, 2019. Synthetic Intelligence and 5G Cellular Know-how can drive funding alternatives in rising markets , Washington DC: IFC .

Judah, N., 2018. The worldwide affect of AI throughout industries. [Online]
[Accessed 29 November 2021].

Kearney, n.d. Investing in AI-driven industrial automation. [Online]
[Accessed 29 November 2021].

Lexology, 2021. What are digital twins and what are the authorized points with them?. [Online]
[Accessed 15 February 2022].

Lord Clement Jones and Christian Gordon-Pullar, 2022. Synthetic Intelligence and Mental Property: incentivize human innovation and creation. [Online]
[Accessed 20 January 2022].

Lord Clement Jones, 2022. Synthetic Intelligence and Mental Property: incentivize human innovation and creation. [Online]
[Accessed 20 January 2022].

Mathys & Squire LLP, 2021. IPO launches IP Entry grant funding scheme to assist progressive UK-based SMEs. [Online]
[Accessed 27 January 2022].

McCarthy, J., 2011. What’s AI? / Primary Questions. [Online]
[Accessed 14 February 2022].

McKinsey & Firm, 2019. Notes from the AI Frontier – Tackling Europe’s Hole in Digital and AI, Brussels: McKinsey & Firm.

Mckinsey Digital, 2020. World Survey: The state of AI in 2020 , s.l.: McKinsey & Firm .

MMC Ventures, 2019. The State of AI 2019: Divergence.” MMC Ventures Report in Partnership with Barclays. [Online]
[Accessed 30 09 2021].

Mobihealthnews, 2021. UK males’s well being start-up Numan acquires Swedish well being tech firm Vi-Well being. [Online]
[Accessed 13 January 2022].

Müller , V. C. & Bostrom, N., 2016. Future progress in synthetic intelligence: A survey of professional opinion. Elementary problems with synthetic intelligenc, pp. 555-572.

Nesta, 2017. New Technology of Synthetic Intelligence Growth Plan. [Online]
[Accessed 30 November 2021].

New America, 2017. China’s Plan to ‘Lead’ in AI: Objective, Prospects, and Issues. [Online]
[Accessed 30 November 2021].

OECD, 2019. Manmade Intelligence in society, Paris: OECD Publishing.

OECD, 2021. Authorities assist for enterprise analysis and innovation in a world in disaster. In: OECD Science, Know-how and Innovation Outlook 2021: Instances of Disaster and Alternative. Paris: OECD.

OECD, 2021. OECD Enterprise and Finance Outlook 2021 : AI in Enterprise and Finance, Paris: OECD.

ONS, 2019. Funding in intangible property within the UK by business: 2019, London: Open Authorities Licence v3.0.

ONS, 2019. Funding in intangible property within the UK by business: 2019, London: Open Authorities Licence v3.0.

ONS, 2019. UK innovation survey 2019: predominant report. [Online]
[Accessed 26 January 2022].

Oxford Insights, 2019. Authorities Synthetic Intelligence Readiness. [Online]
[Accessed 26 January 2022].

Pierides, M. & Roxon, C., 2021. Legislative Approaches to AI: European Union v. United Kingdom. [Online]
[Accessed 05 December 2021].

Pinsent Masons, 2021. UK considers new IP proper for AI. [Online]
[Accessed 27 January 2022].

Rijksoverheid, 2019. Strategic Motion plan for Synthetic Intelligence , Den Haag: Ministerie van Economische Zaken en Klimaat.

Searle, J. R., 1980. Minds, brains, and applications. Behavioral and Mind Sciences 3, Quantity 3, pp. 417-457.

South China Morning Post, 2018. China plans to be a world chief in Synthetic Intelligence by 2030. [Online]
[Accessed 15 October 2021].

Statista, 2021. Worth of investments into AI in China 2016-2020. Sourced from: Deloitte, Intel; varied sources (Shenzhen AI Trade Affiliation); ITJuZi. [Online][Accessed 25 January 2022].

Statista, 2022. Variety of exits made by start-ups after taking part in chosen accelerators in the UK as of July 2021. [Online]
[Accessed 21 January 2022].

TechMonitor, 2021. AI semiconductor funding rockets as demand for superior chips grows. [Online]
[Accessed 26 January 2022].

TechUK, 2021. Cloud Computing and Knowledge prices to be coated beneath expanded R&D tax credit score. [Online]
[Accessed 06 December 2021].

Tejaswi, M., 2021. India anticipated to take a position $1 billion in AI by 2023. The Hindu, 26 April, pp.

The Financial Instances, 2021. India Inc. intends to extend spending on AI: Deloitte report. The Financial Instances, 16 December.

The Manufacturer, 2022. Monolith AI and Imperial School London to make use of £500k Innovate UK grant to create AI that may assess metallic half manufacturability. [Online]
[Accessed 21 Jan 2022].

Turing, A. M., 1950. I. – Computing Equipment and Intelligence,. Thoughts, LIX(236), pp. 433-460.

UK AI Council, 2021. AI Roadmap. [Online]
[Accessed 14 10 2021].

UKIPO, 2019. Synthetic Intelligence A worldwide overview of AI patents and patenting by the UK AI sector, Newport: UKIPO.

UKIPO, 2021. Synthetic Intelligence and IP: copyright and patents. [Online]
[Accessed 27 January 2022].

UKIPO, 2021. Financial Evaluation of the position of Mental Property (IP) on investments in Synthetic Intelligence (AI), Newport: UKIPO.

US Government Publishing Office, 2016. S.1890 – Defend Commerce Secrets and techniques Act of 2016. [Online]
[Accessed 02 12 2021].

USPTO, 2020. Public Views on Synthetic Intelligence and mental property coverage , Alexandria : USPTO .

van der Merwe, B., 2021. Weekly knowledge: How China turned an AI powerhouse. InvestmentMonitor, 9 November, pp.

Verdict, 2021. UK makes an attempt to affix the worldwide AI arms race – however at a price to regulation. [Online]
[Accessed 05 December 2021].

Wikipedia, 2021. [Online]

WIPO, 2019. WIPO Know-how Pattern 2019 – Synthetic Intelligence, Geneva: WIPO.

WIPO, 2019. World Mental Property Report; Intangible Capital in World Worth Chains – Chosen AI Classes and phrases. [Online]
[Accessed 15th October October 2021].

Wise, J., 2017. Balderton Capital (UK) LLP – Written Proof (ALC0232). [Online]
[Accessed 07 October 2021].

World Bank, 2021. Harnessing synthetic intelligence for improvement within the put up Covid-19 period – A overview of nationwide AI methods and insurance policies. [Online]
[Accessed 8 October 2021].

Zhang, D. et al., 2021. The AI Index 2021 Annual Report, Stanford: Stanford College.


Appendix 1: Actors interviewed

Senior representatives of the next stakeholder organisations, companies and buyers have been interviewed in the midst of this examine:

IP Places of work


Firms by measurement, AI sort, subject and software

Firm Sort AI Sort Area Utility
Aistetic Micro Pc Imaginative and prescient Medtech Retail / Digital Well being Micro Neural Networks Leisure Film & gaming business
Peptone Micro Machine Studying Medtech Drug Discovery
Agile Analog Small Machine Studying Chip manufacturing Circuit design
Brainomix Small Machine Studying Medtech Medical Imaging Small Machine Studying Medtech AI primarily based digital remedy
Arctic Shores Medium Machine Studying Recruitment Tech Recruitment
Oxbotica Medium Pc Imaginative and prescient AV Autonomous Automobiles
Benevolent AI Massive Machine Studying Medtech Drug Discovery
Onfido Massive Machine Studying Fintech Safety in Monetary sector
Amadeus Massive (multinational) Machine Studying IT Journey business techniques
GSK Massive (multinational) Machine Studying Medtech Drug discovery
Nvidia Massive (multinational) Machine Studying Chip manufacturing Pc techniques
Philips Massive (multinational) Machine Studying Medtech Shopper merchandise


  • Samsung Ventures – Company investor

  • Positive Valley Ventures – VC

  • Capital Enterprise – Seed funding

Universities and analysis organisations

Miscellaneous organisations (think-tanks, incubators, accelerators)

  • Alliance for IP (media/copyright)

  • Naomi Korn Associates (copyright consultancy)

  • (start-up/accelerator)

  • Tech Nation (accelerator)

Determine A1.1 beneath exhibits a breakdown for the businesses’ respondents by sort of AI used. The bulk makes use of ML, whereas others use Pc Imaginative and prescient or Neural Networks

Determine A1.1 Chart displaying the kind of AI utilized by interviewed corporations

Figure A1.1 - A pie chart showing the split in the types of AI used by companies’ respondents, and indicating that Machine Learning is most common at 79%, followed by Computer Vision at 14% and Neural Networks at 7%

Determine A1.2 signifies the kind of AI being utilized by corporations’ respondents and the fields of AI software. One of many predominant areas is in MedTech. Current studies point out the numerous development of this subject lately, particularly in drug discovery (Zhang, et al., 2021).

Determine A1.2 Overview of corporations’ respondents by sort of AI used, subject of software and firm measurement

Figure A1.2 - A barplot indicating the split by type of AI used and field of application across Micro, SME and large companies interviewed

Appendix 2: Inquiries to assist interviews

Questions for SMEs/Corporates

The next questions have been used to assist the face-to-face interviews with SMEs and Corporates.

A. Private data

a. Title of Firm:
b. Title of individual interviewed:
c. Position of individual interviewed (must be an IP Supervisor/CTO or related)
d. Title of interviewer:
e. Date and time:
f. Gender
g. Ethnicity

B. Basic data

a. In what subject is what you are promoting lively?
b. Is what you are promoting concerned in innovation?
c. What are the large developments/adjustments taking place in your market in the meanwhile?
d. What do you see taking place sooner or later, say 5 years from now?
e. Do you have got workers primarily based within the UK, if that’s the case, roughly what number of?
f. Why do you have got individuals within the UK?
g. What are some great benefits of being within the UK?
1. Are there disadvantages to being within the UK?

C. AI use/funding and determinants

a. How would you outline AI in what you are promoting context?

b. What advantages does AI present, now or future?

c. Do you employ / put money into AI in what you are promoting?

i. If sure,

1. For what function (analysis actions, advertising and marketing, …)?

2. What components have acted as a barrier to undertake AI?

3. Do you have got AI experience inside your organization?
a. Are they primarily based within the UK?
i. if sure, what are the benefits and downsides of getting workers within the UK?

4. Do you purchase/licence AI functions? And if that’s the case, what sort of rights?

d. Are there any financial components which cease you from investing in AI now?

e. Do you’re employed with companions analysis or growing AI capabilities?

f. Do you fund any exterior AI developments, if that’s the case, by which nation is the event being carried out?

D. IP use and position in enterprise selections together with AI funding

a. Do you personal any registered IP rights reminiscent of patents, commerce marks or designs? Specify which rights you personal.
i. If sure,
1. How necessary are patents to what you are promoting?

2. Do patents play a task in what you are promoting selections? [All the time, Some of the time, Never]
a. If no, why not?

3. How necessary are designs to what you are promoting?

4. Do designs play a task in what you are promoting selections? [All the time, Some of the time, Never]
a. If no, why not?

5. How necessary are commerce marks to what you are promoting?

6. Do commerce marks play a task in what you are promoting selections? [All the time, Some of the time, Never]
a. If no, why not?

b. Do you have got unregistered rights reminiscent of copyright or commerce secrets and techniques?
i. If sure,
1. What kinds of rights?

2. How necessary are copyrights to what you are promoting?

3. Do copyrights play a task in what you are promoting selections? [All the time, Some of the time, Never]
a. If no, why not?

4. How necessary are commerce secrets and techniques to what you are promoting?

5. Do commerce secrets and techniques play a task in what you are promoting selections? [All the time, Some of the time, Never]
a. If no, why not?

c. Does the UK IP framework incentivise you to put money into AI or AI initiatives?
1. If sure, how and in what approach is the IP framework an incentive so that you can put money into IP?
2. If no, why is the IP framework not an incentive to take a position?

d. Is there something concerning the UK IP framework that you simply wish to see modified?

Questions for inventors

The next questions have been used to assist the face-to-face interviews with Buyers:

Title of Firm
Title of individual interviewed
Position of individual interviewed
Interviewed by

  • what sort of investor are you? – e.g. angel, VC, company

  • in what areas to you make investments?

  • what are the large developments/adjustments taking place in your market in the meanwhile?
    What do you see taking place sooner or later, say 5 years from now?

  • do you put money into AI? What sort?

  • the place are your predominant investments/ portfolio corporations?

  • what quantity of portfolio corporations relate to AI or ML primarily based applied sciences?

  • do you have got AI experience in your funding group? What number of are UK primarily based?

  • what are some great benefits of being within the UK?

  • are there disadvantages to being within the UK?

  • what are the primary components you take a look at when contemplating investing in a enterprise?

  • what kind of due diligence do you conduct earlier than investing? What are the primary necessities?

  • do you contemplate registered IP rights reminiscent of patents, commerce marks or designs when doing all your due diligence of potential funding?

  • how necessary are registered rights to you in your funding choice?

  • what do you see as the primary advantages of registered rights in business phrases?

  • do you contemplate unregistered rights reminiscent of copyright or commerce secrets and techniques?

  • what’s your view on know-how and expertise retention?

  • does the UK IP framework incentivise you to put money into AI or AI initiatives?

  • if sure, how and in what approach is the IP framework an incentive so that you can put money into IP?

  • if no, why is the IP framework not an incentive to take a position?

  • is there something concerning the UK IP framework that you simply wish to see modified?

  • if this was modified, would it not incentivise you to take a position extra money in AI in what you are promoting?

  • are there any financial components which cease you investing in AI now?