4M!|\|3

A nerd doing some AI and feeling homesick…

[4M!|\|3 @_minus__]

Our 1st scientific paper at @instadeepai is on arXiv! We introduce a new algorithm, Ranked Rewards (R2) to enable self-play benefits for single-player games of combinatorial nature. On the 2D bin packing, R2 outperforms MCTS as well as RL algorithms. #AI https://t.co/0tRnQphjQB

[Patrick Heizer @PatrickHeizer]

Sorry to be the downer because this is an impressive story in some senses.

But it is ~trivially easy to make a single mRNA vaccine. It's not hard.

I cure mice of various cancers with various therapeutics all the time. I've made mice lose more weight in a month than tirzepatide does in a year.

What is hard and expensive is proving its BOTH safe AND effective **in a randomized and controlled study in humans** while ALSO manufacturing it at clinical scale and grade.

I am happy for this man and his dog. It is impressive.

But y'all are overhyping it.

[DeepPCB @DeepPCB]

https://t.co/C2b7XjOuHo

[Robin Cole @robmarkcole]

🎙️New Podcast & YouTube: Democratising access to GeoAI with InstaGeo with Ibrahim Salihu Yusuf

[InstaDeep @instadeepai]

🚀 Building on InstaNovo and shaped by our team, collaborators and community, we’re excited to introduce Winnow: a new method for estimating false discovery rate in de novo peptide sequencing. 🧵⬇️

[Andrej Karpathy @karpathy]

+1 for "context engineering" over "prompt engineering".

People associate prompts with short task descriptions you'd give an LLM in your day-to-day use. When in every industrial-strength LLM app, context engineering is the delicate art and science of filling the context window with just the right information for the next step. Science because doing this right involves task descriptions and explanations, few shot examples, RAG, related (possibly multimodal) data, tools, state and history, compacting... Too little or of the wrong form and the LLM doesn't have the right context for optimal performance. Too much or too irrelevant and the LLM costs might go up and performance might come down. Doing this well is highly non-trivial. And art because of the guiding intuition around LLM psychology of people spirits.

On top of context engineering itself, an LLM app has to:
- break up problems just right into control flows
- pack the context windows just right
- dispatch calls to LLMs of the right kind and capability
- handle generation-verification UIUX flows
- a lot more - guardrails, security, evals, parallelism, prefetching, ...

So context engineering is just one small piece of an emerging thick layer of non-trivial software that coordinates individual LLM calls (and a lot more) into full LLM apps. The term "ChatGPT wrapper" is tired and really, really wrong.

[Gergely Orosz @GergelyOrosz]

Give it time for more people to understand this

[Gergely Orosz @GergelyOrosz]

“AI Engineering is just software engineering with AI models thrown into the stack”

- @abacaj in the book AI Engineering by @chipro

Totally agree - and it’s why learning how to integrate LLMs is such a big win for devs!

[DeepPCB @DeepPCB]

We’ve talked routing—now let’s dive into placement! Discover how DeepPCB Placement elevates your PCB design process 🚀

Real-world use-cases showing why placement + routing = winning combo🏅

👉 https://t.co/zcsuPqmzqJ

[Nando de Freitas @NandoDF]

Two European startups @recraftai and @bfl_ml lead image generation in the world. The third place model, Imagen 3, was developed in London but under a Californian company. https://t.co/p7c0sa87An

Europe could lead in other AI products by supporting more entrepreneurship and competition. One issue is that American companies enforce 6 month to 1 year notice periods and non-competes in Europe, but don’t do it in California. A good example of this is @GoogleDeepMind. They force employees to sign these contracts or retaliate by preventing them from getting merit promotions. This is wrong - it’s time for the UK government ( @matthewclifford ) not to be seduced by Google and develop its own AI industry. @xai Grok 3 came to be because a few deepminders left Google under the protection of musk. People leaving DeepMind also led to the creation of @MistralAI - if it was easier to leave, imagine how much more competitive Europe could be.

Eliminate notice periods and non-competes in Europe. It has become a question of national security.

[Nathan Benaich @nathanbenaich]

it’s funny - famous ai researchers said during the deep learning wave that radiologists would be obsolete

turns out, its market research professionals that might be obsolete first

radiology is still pretty hard.

[Andrej Karpathy @karpathy]

Agency > Intelligence

I had this intuitively wrong for decades, I think due to a pervasive cultural veneration of intelligence, various entertainment/media, obsession with IQ etc. Agency is significantly more powerful and significantly more scarce. Are you hiring for agency? Are we educating for agency? Are you acting as if you had 10X agency?

Grok explanation is ~close:

“Agency, as a personality trait, refers to an individual's capacity to take initiative, make decisions, and exert control over their actions and environment. It’s about being proactive rather than reactive—someone with high agency doesn’t just let life happen to them; they shape it. Think of it as a blend of self-efficacy, determination, and a sense of ownership over one’s path.

People with strong agency tend to set goals and pursue them with confidence, even in the face of obstacles. They’re the type to say, “I’ll figure it out,” and then actually do it. On the flip side, someone low in agency might feel more like a passenger in their own life, waiting for external forces—like luck, other people, or circumstances—to dictate what happens next.

It’s not quite the same as assertiveness or ambition, though it can overlap. Agency is quieter, more internal—it’s the belief that you *can* act, paired with the will to follow through. Psychologists often tie it to concepts like locus of control: high-agency folks lean toward an internal locus, feeling they steer their fate, while low-agency folks might lean external, seeing life as something that happens *to* them.”

[Lotus Cars @lotuscars]

The first Lotus F1 car, Type 12, now on display at Lotus London 🏁

[Deedy @deedydas]

New junior developers can't actually code.

Fantastic blog where the author argues that AI is preventing devs from understanding anything

Unlike the internet which only replaced fact lookups but adds explanations, AI replaces all reasoning.

[4M!|\|3 @_minus__]

@MW_A Chiefs

[Marc Andreessen 🇺🇸 @pmarca]

This is playing out in AI right now.

[InstaDeep @instadeepai]

🎊 We’re proud that #InstaGeo has been selected by the @ParisPeaceForum as one of just 50 projects from 770 submissions across 111 countries for the #AIActionSummit 🇫🇷

InstaGeo uses AI and geospatial data to tackle global challenges. From monitoring desert locust swarms to creating scalable solutions for crises, we’re excited to showcase how #AIForGood is making an impact 🛰️.
Learn more: https://t.co/kbLtoWitHP

[Gergely Orosz @GergelyOrosz]

"We've reduced our support load by 99.9% using AI"

In reality... your customer can't get ANY support now

I'm getting tired of these terrible AI bot rollouts...

Example:

[Karim Beguir @kbeguir]

Thank you, Jensen, for an engaging discussion about @instadeepai’s R&D journey and story! It was a pleasure attending this incredible @nvidia event!

[Mario Nawfal @MarioNawfal]

🇦🇷 ECONOMISTS WHO PREDICTED MILEI'S 'DEVASTATION' NOW AWKWARDLY QUIET AS ARGENTINA REBOUNDS

Turns out the "crazy" guy with a chainsaw knew what he was doing.

After experts warned Milei would destroy Argentina, his 30% spending cuts and mass bureaucrat firings led to the first budget surplus since 2008.

Even more shocking?

His 50% approval rating suggests Argentinians prefer smaller government and 2.4% inflation over socialism's 118% interest rates.

Who knew?

Source: NY Post

[Karim Beguir @kbeguir]

Component placement is a game changer! Now available on https://t.co/JxGH9F97Ar 🤗