iiinova — where your digital self keeps evolving

Aidan Kelly @helloaidank.bsky.social · 03/10/2026

I regret to inform everybody but "I'm upping my p(doom)" is an absolute banger. www.youtube.com/watch?v=5EoO...

youtube.com

i'm upping my p(doom)

YouTube video by mexicat

011

Reposted by Aidan Kelly

Andrew Sissons @acjsissons.bsky.social · 02/10/2026

New blog from me: devolving the Warm Homes Plan is good, but not the Boiler Upgrade Scheme! Energy Sec Miatta Fahnbulleh has announced that the £15bn Warm Homes Plan will be devolved. We think this is good - mayors are well placed to upgrade homes. BUT! www.nesta.org.uk/blog/devolvi...

nesta.org.uk

Devolving the Warm Homes Plan to local government makes sense - but the Boiler Upgrade Scheme should stay

Three implications of devolving the Boiler Upgrade Scheme.

32014

Aidan Kelly @helloaidank.bsky.social · 27/03/2026

Notebook LM language pulls out the more unclear language/framing sometimes. Amazing to see the difference in clarity between it and ChatGPT.

070

Aidan Kelly @helloaidank.bsky.social · 25/02/2026

With the advent of Claude Code, my gh contributions have gone exponential.

150

Aidan Kelly @helloaidank.bsky.social · 04/02/2026

Can someone do a live react vid of Sam Altman to this OpenAI roast 🤣 www.youtube.com/watch?v=FBSa...

youtube.com

How can I communicate better with my mom?

YouTube video by Anthropic

020

Aidan Kelly @helloaidank.bsky.social · 31/01/2026

He must be nearly seeing the molten core at this stage.

000

Aidan Kelly @helloaidank.bsky.social · 30/01/2026

Yep, definitely. High level system thinking alongside management skills will be at a premium, the actual coding will most likely be done by an agent in the near term. On a day to day basis, you create a bunch of well defined tasks and you get your agent to come check with you intermittently.

010

Aidan Kelly @helloaidank.bsky.social · 30/01/2026

Moltbook..a social media site for AI agents. Apparently released just 48 hours ago, there are AI agents already scheming around having e2e encrypted agent to agent communication away from human oversight (according to this Moltbook post, it's already happening). Welp, it was nice knowing y'all.

130

Aidan Kelly @helloaidank.bsky.social · 30/01/2026

This might mean a change in the portfolio of code projects, where one might decide to experiment with something new a couple days/month. Likely more failed experiments but also more innovation as well as a more diverse portfolio of work!

110

Aidan Kelly @helloaidank.bsky.social · 30/01/2026

Managers in this case will really have to prioritise what's important. I think potentially one of the benefits of this new found code accelerator will be able to spin up prototypes / test a new method real quick (~1-2 days) that otherwise would've taken weeks.

110

Aidan Kelly @helloaidank.bsky.social · 30/01/2026

It's now become a choice about what you decide to build, as before we had to ruthlessly prioritise as the barrier was too high for 'nice-to-have' features. This might mean that we start over-engineering products and analyses when potentially a simpler path would've been fruitful.

110

Aidan Kelly @helloaidank.bsky.social · 30/01/2026

Few thoughts on this: It is transforming how coding is being done and has introduced a new way of working. You have some of the most talented coders/engineers using it extensively in their day-to-day workflows (from Linus Torvalds to Andrej Karpathy). Agentic AI coding tool are getting rly good.

110

Aidan Kelly @helloaidank.bsky.social · 18/01/2026

Honestly, it is scary seeing fascism on the march across many different western countries. There are Portuguese elections today where Ventura (the Farage of Portugal) is likely to reach the runoff showdown. We must resist!

1243

Reposted by Aidan Kelly

Aidan Kelly @helloaidank.bsky.social · 25/04/2025

I've had a first stab at making a list of GenAI ppl on here but it's proving harder than anticipated! X is still the place to be for that content. Let me know if you're interested in being added or know anyone who would be a good addition😎 bsky.app/profile/did:...

075

Aidan Kelly @helloaidank.bsky.social · 13/01/2026

Gentle intro to on-policy distillation for language models from Thinking Machines. It defines a lot of key terms used in post-training which I found useful. Seems like there is scope using this technique to post-train LMs without need for huge compute. thinkingmachines.ai/blog/on-poli...

thinkingmachines.ai

On-Policy Distillation

On-policy, dense supervision is a useful tool for distillation

000

Reposted by Aidan Kelly

Aidan Kelly @helloaidank.bsky.social · 02/01/2026

Really interesting piece in the FT with Yann LeCun (@yann-lecun.bsky.social). Insights into the dynamics at Meta and how the winds of AI have changed recently. Interested in learning more about the world models he has been working on. www.ft.com/content/e3c4...

ft.com

Computer scientist Yann LeCun: ‘Intelligence really is about learning’

The AI pioneer on stepping down from Meta, the limits of large language models — and the launch of his new start-up

031

Reposted by Aidan Kelly

Sakana AI @sakanaai.bsky.social · 12/01/2026

Introducing DroPE: Extending Context by Dropping Positional Embeddings We found embeddings like RoPE aid training but bottleneck long-sequence generalization. Our solution’s simple: treat them as a temporary training scaffold, not a permanent necessity. arxiv.org/abs/2512.12167 pub.sakana.ai/DroPE

211821

Aidan Kelly @helloaidank.bsky.social · 11/01/2026

Listening to the news agent podcast earlier with Mehdi Hasan, and talk of Trump being re-elected in 2020 rather than 2024 may have been the lesser of two evils. On the surface, it seems relatively convincing.

000

Aidan Kelly @helloaidank.bsky.social · 02/01/2026

ft.com

Computer scientist Yann LeCun: ‘Intelligence really is about learning’

The AI pioneer on stepping down from Meta, the limits of large language models — and the launch of his new start-up

031

Aidan Kelly @helloaidank.bsky.social · 02/01/2026

Watching what's happening over on grok in terms of image manipulation is really quite worrying and a clear violation of privacy. It needs to be regulated with a big stick, however I fear this is only the tip of the iceberg of what's out there on the web.

010

Reposted by Aidan Kelly

Kara @karashiiro.moe · 01/01/2026

There's been a lot of discourse lately about how coding agents have become overwhelming, and that SWEs who don't use them yet just won't be able to catch up. I don't believe this is true, but we do need to know what problems coding agents are solving to approach them with an engineering mindset.

karashiiro.leaflet.pub

Coding Agents Are Easy, Actually

Software engineering is here to stay

3557

Reposted by Aidan Kelly

David Aronchick @ironyuppie.com · 29/12/2025

I didn't see enough news about this crazy/cool paper from DeepSeek - you take text, and convert it to imagery, and you get BETTER performance. Oh and some really neat ideas about contextual memory too. www.distributedthoughts.org/a-picture-is...

distributedthoughts.org

A Picture Is Worth Ten Thousand Tokens

DeepSeek's new OCR model achieves 97% accuracy while using one-tenth the tokens by rendering text as images. The computationally "heavy" modality turns out to be the efficient one, which should make u...

342

Aidan Kelly @helloaidank.bsky.social · 29/12/2025

Can't stop listening to this. Mixture of synth, bass and guitar alongside this genre must have felt like it was from another planet in 1980. www.youtube.com/watch?v=qh-j...

youtube.com

A Forest

YouTube video by The Cure - Topic

000

Reposted by Aidan Kelly

AI & ML News @ai-news.at.thenote.app · 27/12/2025

Exploring TabPFN: A Foundation Model Built for Tabular Data Understanding the architecture, training pipeline and implementing TabPFN in practice Telegram AI Digest #ai #foundationmodel #news

towardsdatascience.com

Exploring TabPFN: A Foundation Model Built for Tabular Data

Understanding the architecture, training pipeline and implementing TabPFN in practice

011

Aidan Kelly @helloaidank.bsky.social · 25/12/2025

Trying to understand some code in Rust and it's giving me vietnam style flashbacks to C++

110

Aidan Kelly @helloaidank.bsky.social · 06/12/2025

Thoughtful piece by my current favourite podcaster @dwarkesh.bsky.social on where we're at with AI progress and reflections around intelligence itself: substack.com/home/post/p-...

substack.com

Thoughts on AI progress (Dec 2025)

Why I'm moderately bearish in the short term, and explosively bullish in the long term

010

Aidan Kelly @helloaidank.bsky.social · 15/11/2025

Yann LeCun being let go by Meta might have some unexpected benefits 🍿

050

Aidan Kelly @helloaidank.bsky.social · 04/11/2025

This may be very old man of me, but I feel like there is something about the bluesky algorithm/my feed that is not sucking me in enough. I know, I know, it's sort of the point but still, I need to have some pull to check out the content and it isn't quite itching that scratch.

010

Reposted by Aidan Kelly

DESIGN THINKING! Comic @designthinking.lol · 20/10/2025

With the AWS outage, now‘s as good a time as any to post this old strip.

1526451043

Aidan Kelly @helloaidank.bsky.social · 20/10/2025

🏛️ Policy trends: Electricity price reduction is climbing the policy agenda. Recent debate has centred on long-term ways to bring down prices for households and businesses. That’s key to helping heat pumps compete with gas boilers on running costs. (7/7)

120

Aidan Kelly @helloaidank.bsky.social · 20/10/2025

This combined heat pumps, community energy storage, EV car clubs and solar panels to cut bills and upfront costs. (6/7)

110

Aidan Kelly @helloaidank.bsky.social · 20/10/2025

🔬 Research funding: UK research has shifted from optimising heat pumps to green finance, mirroring the investment trend. Example: Rossendale Borough Council received over £2 million for a place-based decarbonisation demonstrator in terraced streets. (5/7)

100

Aidan Kelly @helloaidank.bsky.social · 20/10/2025

IKEA: Meatballs, flat-packs and now heat pumps. The company is now offering heat pumps in several European countries, including the UK since late September, through partnerships with manufacturers such as Aira. (4/7)

110

Aidan Kelly @helloaidank.bsky.social · 20/10/2025

💡 Venture funding: Finance for low-carbon heating is getting a green light from investors. Global investment in 2025 rose sharply for businesses offering loans and financial products to support uptake. (3/7)

100

Aidan Kelly @helloaidank.bsky.social · 20/10/2025

In this edition, we looked at innovations to make low-carbon heating more affordable, a key barrier in our mission to decarbonise home heating in the UK. (2/7)

110

Aidan Kelly @helloaidank.bsky.social · 20/10/2025

Happy to share my first piece of analysis work with the Discovery team at @nestauk.bsky.social on 𝗠𝗶𝘀𝘀𝗶𝗼𝗻 𝗥𝗮𝗱𝗮𝗿! The Mission Radar is a briefing series on innovation signals and emerging trends in Nesta’s missions. We use AI to scan for innovation across business, academia and policy. (1/7)

Mission Radar chart (a type of bubble chart) for Nesta's "A sustainable future" mission, updated Q2 2025.

The radar is divided into four quadrants representing categories of innovation: "Ventures and investments" (top right), "Research funding" (bottom right), "Policy" (bottom left), and an unlabeled top-left quadrant (representing trends that are not primarily investment, funding, or policy). A dashed vertical line separates signals in the "Policy" and unlabeled quadrant from those in "Ventures and investments" and "Research funding."

Blips are positioned based on time to expected impact, from the center outwards: "0-2 years," "3-5 years," and "5+ years."

Blip size indicates potential impact on home decarbonisation, with larger blips signifying greater expected impact and smaller blips signifying smaller or more niche impact.

Main Trends and Findings (Blips):

Venture and Investment (High Impact, Short/Medium Term):
Ikea heat pumps (Small, 0-2 years)
Digital platform for local installers (Medium, 0-2 years)
Energy-as-a-service (Small, 0-2 years)
Retail investment in green finance ventures (Medium, 3-5 years)
Research Funding (Medium Impact, Short/Medium Term):
Turbo air-source heat pump (Medium, 0-2 years)
Net zero terrace streets (Medium, 3-5 years)
Community operated energy infrastructure (Small, 3-5 years)
Policy (Medium/High Impact, Short Term):
Increased government publications on electricity prices (Medium, 0-2 years)
Increased references to electricity prices in debates (Small, 0-2 years)
Top-Left Quadrant (High Impact, Medium Term):
One large, dark blip is visible in the 3-5 years range.
Source: Nesta's analysis of innovation trends. Companies and research projects included are those identified in Q2 2025. Blip size and positioning determined by expert judgement.

142

Aidan Kelly @helloaidank.bsky.social · 19/10/2025

What a really interesting conversation between two of my favourite people in AI in the form of @karpathy.bsky.social and @dwarkesh.bsky.social. I wish more ppl in this space took the same approach in terms of being open to debate, challenge and overall being humble. www.youtube.com/watch?v=lXUZ...

youtube.com

Andrej Karpathy — “We’re summoning ghosts, not building animals”

YouTube video by Dwarkesh Patel

010

Aidan Kelly @helloaidank.bsky.social · 05/10/2025

The author doesn't really address the wider implications of deterministic LLMs, but can imagine the reproducibility of outputs could be very powerful in terms of trust. Link to the blog: thinkingmachines.ai/blog/defeati... (6/6)

thinkingmachines.ai

Defeating Nondeterminism in LLM Inference

Reproducibility is a bedrock of scientific progress. However, it’s remarkably difficult to get reproducible results out of large language models. For example, you might observe that asking ChatGPT the...

000

Aidan Kelly @helloaidank.bsky.social · 05/10/2025

Used ChatGPT in study and think mode as well as prompting it to use socratic questioning to really improve my understanding of this area. (5/6)

100

Aidan Kelly @helloaidank.bsky.social · 05/10/2025

They've managed to create deterministic LLMs at inference time with some penalties for efficiency, but the results are pretty cool. I found their blog quite a long read, as I had little familiarity around the intricacies of these calculations and how they relate to GPUs. (4/6)

100

Aidan Kelly @helloaidank.bsky.social · 05/10/2025

However there are different strategies to get around this, e.g. running sums over single cores in a GPU. This is what 'Thinking Machines' have been doing with implementing batch-invariant calculations for LLMs. (3/6)

100

Aidan Kelly @helloaidank.bsky.social · 05/10/2025

This is caused by something called floating point non-associativity. In Maths, we can do (a+b)+c =a+(b+c) but this is not the case when doing calculations in computers, the order matters. (2/6)

100

Aidan Kelly @helloaidank.bsky.social · 05/10/2025

So theoretically LLMs can be deterministic if we set their temperature to zero, but in practice they aren't. This comes down to how we do calculations such as matrix multiplication at inference time. It turns out the order you do these calculations in does matter. (1/6)

100

Aidan Kelly @helloaidank.bsky.social · 22/09/2025

Does anyone have cool (or boring) agentic use-cases that they've read about or used? Doing some horizon scanning work to see what's out there but thought my bluesky network might have ideas!

000

Aidan Kelly @helloaidank.bsky.social · 12/09/2025

That's what we are building with Nesta's Policy Atlas tool! Read about our alpha tool user testing and what's next for the project, here: www.nesta.org.uk/project-upda... (2/2)

nesta.org.uk

Harnessing AI to positively disrupt policymaking

Lessons learned from our user research

010

Aidan Kelly @helloaidank.bsky.social · 12/09/2025

What if there was an AI-powered tool that allowed policymakers to rapidly identify the most effective policies to tackle any given social challenge, based on the latest evidence? (1/2)

110