On January 13, the government unveiled its AI Opportunities Action Plan, a comprehensive document crafted over several months by Matt Clifford, who will continue as an advisor during its execution. This plan contains 50 recommendations, almost all of which the government fully accepted. Let’s dive into the details.
1. Within six months, create a long-term strategy addressing the UK’s AI infrastructure, supported by a 10-year funding commitment.
2. Increase the AI Research Resource (AIRR) capacity by at least 20 times by 2030, starting now.
3. Designate mission-driven “AIRR programme directors” to efficiently manage sovereign compute resources.
4. Set up “AI Growth Zones” to expedite the establishment of AI data centers.
5. Address sustainability and security risks linked to AI infrastructure while seizing new solution opportunities.
6. Form international compute partnerships with countries that share similar goals to enhance computing capabilities for researchers.
7. Quickly identify and release at least five significant public datasets for AI researchers and innovators.
8. Intentionally shape data collection strategies instead of relying on existing datasets.
9. Develop clear guidelines for releasing open government datasets suitable for AI use, focusing on effective data structures and sharing methods.
10. Link compute allocation with access to proprietary datasets to attract researchers and startups to the UK.
11. Build necessary infrastructure for public sector data collection and finance the creation of valuable new datasets.
12. Encourage and reward the curation and unlocking of private datasets by researchers and industry.
13. Create a copyright-cleared British media asset training dataset for international licensing.
14. Assess the skills gap accurately.
15. Help higher education institutions increase the number of AI graduates and teach industry-relevant skills.
16. Promote diversity within the talent pool.
17. Develop educational pathways into AI careers.
18. Launch a major undergraduate and master’s AI scholarship program, comparable to Rhodes or Fulbright, for students studying in the UK.
19. Prepare a lifelong skills program focused on AI.
20. Build internal talent recruitment capabilities to attract elite professionals to the UK.
21. Optimize the existing immigration system to draw graduates from top AI programs worldwide.
22. Expand the Turing AI Fellowship to attract more talent.
23. Continue developing the AI Safety Institute to enhance research on model evaluations and safety.
24. Update the UK’s text and data mining regulations to match or exceed EU standards.
25. Ensure adequate funding for regulators to enhance their AI capabilities swiftly.
26. Include safe AI innovation goals in regulatory guidance across all departments.
27. Collaborate with regulators to accelerate innovation in key sectors, utilizing tools like regulatory sandboxes.
28. Require regulators to report annually on how AI has spurred innovation and growth in their sectors.
29. Strengthen the AI assurance ecosystem to boost trust and usage.
30. Assess the broader institutional landscape to realize the full potential of the Alan Turing Institute.
31. Assign an AI lead for each government mission to explore AI solutions aligned with user needs.
32. Establish a cross-government technical team for horizon scanning and market intelligence in AI.
33. Foster partnerships with AI vendors and startups to anticipate developments and express public sector needs.
34. Use a consistent framework for sourcing AI solutions—whether to build, buy, or innovate through challenges.
35. Implement rapid prototyping for key projects needing immediate attention.
36. Provide specific support for hiring external AI talent.
37. Create a data-rich experimentation environment with easy access to datasets and necessary infrastructure.
38. Streamline the AI procurement process to enable fast funding access for pilot projects.
39. Offer a scaling service for successful pilots backed by senior support and central funding.
40. Launch mission-focused national AI tenders for accelerated adoption in decentralized systems.
41. Develop or acquire a scalable AI tech stack capable of handling millions of citizen interactions.
42. Require interoperability, code reusability, and open-source development in infrastructure.
43. Make smart procurement decisions within the AI ecosystem while shaping the market.
44. Utilize digital government infrastructure to foster new innovation opportunities.
45. Create a centralized “AI Knowledge Hub” for best practices and case studies accessible to all users.
46. Identify quick wins for adopting the “scan, pilot, scale” method within three months.
47. Leverage the new Industrial Strategy for AI growth.
48. Appoint AI Sector Champions in key industries to collaborate with government on AI adoption strategies.
49. Promote AI adoption nationwide.
50. Establish a new unit called UK Sovereign AI to maximize the UK’s stake in frontier AI.
The action plan was initially introduced with a foreword from the Prime Minister, which was later swapped for one by the Secretary of State for Technology. Who knows what happened there? Maybe AI had a hand in it, or perhaps we just misread it.
In summary, Matt Clifford emphasizes we can’t continue as usual. The government must be ready to embrace some uncertainty and take risks. What do you think about this plan? Does it take the right steps toward harnessing AI opportunities? What’s effective, and what might be missing? I’d love to hear your thoughts on how the government should navigate regulation in an uncertain landscape.