英国人工智能创业路线图

Thank you to every startup, scale up, venture capital firm, founder, policy analyst, academic, and investor whocontributed to this report, whether through larger discussions, case studies, or in one-on-one interviews.Additionally, we would like to extend our thanks toandfor the data and data analysis that madeparts of this report possible. We are optimistic for the future of the UK’s AI leadership – thanks in no short AI is likely to be the definitive technology of the century, disrupting and reforming whole economies. And thisis a make or break moment for AI development in the UK. Supporting, enabling, and turbocharging the UK’s If the UK Government succeeds in harnessing our AI sector, the benefits will be huge. Successfully unleashingAI could help us,, and– potentially addingin economic value by 2030. But if the Government’s approach One thing is clear: we will not succeed without startups. That is why Startup Coalition, Onward, and the TonyBlair Institute for Global Change (TBI) have come together to deliver the AI Project: to better understand whatthe UK’s world-beating AI startups need to succeed in the new AI age. We have spent the last three months speaking to startup founders to build an in-depth understanding of thechallenges they face and what is needed to fix them. We did this through moderated group discussions,one-on-one interviews, and policy dinners. While we covered many topics and our discussions were Four key challenges emerged from our discussions. First, access to capital. While the UK is far up the rankingsin global tech investment, European nations are making significant gains. Paris-based Mistral’s recent $initial funding round was theseed round in Europe. Raising funds remains AI founders’top concern. Untapped potential funding from the pension market would be a vital means of increasing thecapital pools, underlining the importance of reforming pension funds as committed to in the . Cuts to R&D tax credits in 2022 have also Second, access to talent. AI startups struggle to compete with the US for top tech talent due to issues likesalary levels and visa challenges. Some founders argued that the pull factors that used to make the UK anenticing place to home their startups have lessened due to broader economic issues. Startups also told usthat they fear an overfocus on AI safety by the Government could, in the long run, hinder our ability to attract andschemes are both held back due to lack of clarity over the endorsement systemsunderpinning them. Third, access to compute. Overwhelmingly, startups wanted one thing more than anything else when it cameto compute: access to cost-effective and readily available compute capacity. But our compute offer isn’t competitive. By the government’s own: the UK went from third in global compute capacity in 2005,to 10th by 2022. Initiatives like the nationalin Bristol are welcome but will struggle to Fourth, regulatory compliance. Despite welcoming the sector-specific approach to AI regulation set in theGovernment’searlier this year, founders fear that the increased focus on AI safety concernscould create an impetus for over-regulation that would stifle the sector. They also fear that the tradeoffbetweenand accuracy of models may not be appreciated – an overly burdensome focus onexplainability could cause models to be “dumbed down” andcompanies’ innovative edge. Uncertainty The overarching lesson is that as the Government develops its approach to AI we have to keep thetried-and-tested basics in mind. Across capital, talent, compute, and compliance, we have yet to really nailthe fundamentals: accessing capital throughout the lifecycle, hiring and bringing the best and brightest talent To address these challenges, we have compiled “The UK’s AI Startup Roadmap” for the Government.Implementing these steps as quickly as possible will put the UK in prime position to succeed and cement its 1.Deliver on the implementation of Mansion House Compact – and go further2.Renew EIS beyond 2025, review the Financial Health Requirement, and increase HMRC’s EIS/SEIS capacity 4.Expand the HPI Visa to better target AI and startup talent, through a focus on bespoke sub-categories or 5.Negotiate a reciprocal agreement with the US to add them to Youth Mobility Schemes 6.Urgently clarify plans for the Global Talent Visa fix and simplify the endorsement system for the Innovator 7.Simplify the spinout system by reforming how TTOs operate and create a two-tiered model for university equity 8.Grow the compute menu by increasing national compute capacity, getting better compute partnerships, 9.Create new R&D initiatives for compute alternatives and develop new acceleration programmes for quantum 10. Successfully implement the sectoral approach to AI by scaling up regulator capacity as quickly as possible 11. Work with regulators to issue AI-specific guidance for intermediary liability, digital competition, copyright, and 12. Work with like-minded global part