Tom Westgarth

@tomwestgarth.bsky.social

Head of Growth at AI hardware startup @Fractile-ai.bsky.social. Ex policy in Sovereign AI Unit and TBI. Bridging tech and policy worlds at TxP.

Kanishka Narayan's appointment is a really positive development - he is fantastically capable and credible on AI. Key questions: what levers he will have in his new role, as well as whether HMG keep excellent teams in AISI, No 10 etc, well-empowered in the new structure

This is a notable break from orthodoxy that guides thinking within the British state. But it is necessary in order for the UK to compete globally and restore sovereignty There are a few challenges that this poses. What are they, and how can govt respond? www.theguardian.com/business/202...

‘I’m picking winners’: UK business secretary takes activist approach to economic growth

AI evangelist Peter Kyle wants to scale up businesses, attract overseas investors and look out for UK’s poorer regions

theguardian.com

50 papers published at Neurips 2025 (one of the three major global ML conferences) have AI hallucinations. Another example of how institutions need to evolve the tooling they have to spot pollution of the knowledge commons

Bild

Extremely exciting day for open science. Another reason to be bullish on UK AI. Boltz, a startup that emerged out of MIT to be at the frontier of small molecule and protein design, has moved to London. I hear cases like this from a few founders, often a mix of fundamentals (talent)+personal.

Gabriele Corso@gcorso.bsky.social · 7mo ago

Big news from Boltz today: we’re launching Boltz Lab, a new platform with new small-molecule + protein design agents, announcing Boltz PBC and a $28M seed round, and sharing a multi-year partnership with Pfizer. More below!

Decentralised training of AI, where thousands of small datacenters, or individual consumers, pool their spare compute over the internet, is 20xing in scale a year. But this is nowhere near the scale of leading AI infra projects or non-AI decentralised projects. Great piece by @epochai.bsky.social

BildBild

V good thread. IMV what’s double damning about the Ovenden piece is that you could write a thousand pieces about how Whitehall could run better, but absolutely none of them would run through “why should the diplomatic service care about a British citizen in an authoritarian jail?”

Tom Westgarth@tomwestgarth.bsky.social · 7mo ago

I have a few thoughts on the Stakeholder State and the piece by Paul Ovenden. I disagree with the extent of emphasis on stakeholders, but the more dismissive engagement on this app misses that a Strong Stakeholderism does exist, and has placed particular constraints on a govt that lacks direction 🧵

I have a few thoughts on the Stakeholder State and the piece by Paul Ovenden. I disagree with the extent of emphasis on stakeholders, but the more dismissive engagement on this app misses that a Strong Stakeholderism does exist, and has placed particular constraints on a govt that lacks direction 🧵

Duncan Robinson@duncanrobinson.bsky.social · 7mo ago

The Blob, the Groups and now The Stakeholder state Alaa Abd el-Fattah has shown supremacy of the Stakeholder State www.thetimes.com/article/3720...

Brilliant appointment. Jade has done exceptional work in government through AISI, and with her background in VC and frontier AI from the very beginning, she’s exactly the sort of person you want helping the UK prepare for the AI era

Bild

There is a growing community dedicated to unlocking U.K. progress and abundance. Come to the next TxP to hear from organisations such as ARIA, Centre for British Progress, Nucleate, Inference Magazine, Digital Frontier, and Project Europe! lu.ma/fuf590zb?loc...

TxP: The UK Progress Idea Machine · Luma

Join us at Wayve for a showcase of organisations working to unlock UK progress! Momentum is growing amongst a community of ambitious young folks in the UK.…

lu.ma

This prompts an interesting discussion about how these tools may soon be cultural artifacts. Things like BBC historical video archives aren’t just great for understanding evolving values, but changing production techniques etc. Will compute costs eat these cultural phenomena?

Bild

And ultimately much of the work we wrote about is attempting to break away from a model of R&D that is hierarchical and target-driven. First sc from NNP biotech, second sc from the first NNP

BildBild

So Chinese company Deepseek released their latest frontier grade LLM. Some astounding context worth repeating: -They only used 2000 GPUs on a budget of around $6m. This level of capability requires normally around a 16k GPU cluster -Deepseek is literally a project ran by a Chinese hedge fund!

Bild

Excellent post about the recent OpenAI o3 results on ARC (& other benchmarks). I don't know how @natolambert.bsky.social manages to write these so quickly! I highly recommend his newsletter. www.interconnects.ai/p/openais-o3... I am (more slowly) writing my own take on all this, coming soon.

o3: The grand finale of AI in 2024

A step change as influential as the release of GPT-4. Reasoning language models are the current big thing.

interconnects.ai

Why is there a lot of disagreement amongst economists about AI productivity forecasts? This article suggests differences in assumptions around speed of adoption, and if this adoption is concentrated to a small number of sectors.

Bild