HUGE thanks to the amazing team behind this report, incl. Jo Puddick, @dralanwager.bsky.social, Naman Goel, Rhydderch, Daniel Cameron, @Benjamin Roff, and @helenmargetts.bsky.social!

6/6

This matters. Without public support, the government will struggle to deliver on the AI action plan and its wider growth agenda

To build trust in AI, the government must ensure that AI systems are beneficial and safe, as well as center its messaging around AI on how it improves social outcomes

5/6

🛟 38% of UK adults cite lack of trust in AI content as the main barrier to adoption

4/6

📉 More UK adults view AI as a risk for the economy (39%) than an opportunity (20%)

3/6

New polling by TBI and @ipsosintheuk.bsky.social show that:

👫 While 25% of UK adults use generative AI weekly, nearly half have never used it

2/6

Policymakers should build on existing AI policies and plug gaps related to:

● Mandatory safety certification pre-deployment
● Clear liability frameworks for autonomous systems
● Investment in EAI safety research
● Economic and social transition policies

Pre-print: all feedback is welcome!

4/4

With these 4 embodied AI risk categories in mind, we analyzed existing policies (in US, EU, and UK) and found critical gaps

While a good starting point, current frameworks for industrial robots and autonomous vehicles are insufficient to address the full range of risks EAI systems pose

3/4

The core problem: Alongside excitement and opportunity, Embodied AI pose severe risks

EAI inherits traditional AI risks (privacy, bias, security etc) - and pose new ones, e.g. related to physical harm, mass surveillance and displacement of manual labor

We identify 4 key EAI risk categories

2/4

Succeeding in this quest for sovereignty, security & scale requires urgent policy action:

1) Create a clean, robust energy system for the AI era

2) Reform planning systems to make it easier to build

3) Use AI growth zones to overcome market failures

4) Align incentives between UK gov departments

Talk about sovereign AI is often unclear what it means for data, models and compute

The UK doesn’t need to build everything, but it must build enough infrastructure to deploy AI where it matters, to ensure resilience, and to anchor a domestic ecosystem that delivers for the public and the economy

The climate transition and AI revolution are intrinsically linked. To succeed in one we need to succeed in both

institute.global/insights/cli...

We live in the climate paradox. Awareness of the crisis has never been higher but meaningful action is in decline

institute.global/insights/cli...

Panel 4: Using AI to Enhance Democracy is our last panel of the day and will start at 3:40pm ET. Don't miss it.
Panelists: @hahrie.bsky.social, @lukethorburn.com, Spencer Overton, and MH Tessler.
Moderator: @jakobmokander.bsky.social.
#AIDemocraticFreedoms

Huge thanks all experts who have contributed w/ input and feedback!

@rory.bio @areeq.bsky.social @leecronin.bsky.social @erika-alden.bsky.social @econormist.bsky.social @saakohl.bsky.social @mariokrenn.bsky.social @stianwestlake.bsky.social @harrisbio.bsky.social @richardaljones.bsky.social

11/11

Second, at the heart of science is a quest to find patterns in nature and use those to improve the human condition. AI-driven science is science supercharging itself. This means that the normal limitations of – reproducibility, validity, ethics – will not be ‘solved’ by AI

10/11

Two final reflections:

First, ‘AI’ should be understood broadly. Focusing narrowly on LLM-adoption would be shortsighted. Key to success is equipping scientists with the skills and resources to continuously adapt their research methods amid rapid technological change

9/11