A fundamental lesson of modern AI is that scale is essential: training bigger models on bigger datasets unlocks new capabilities. A fundamental lesson of AI engineering is that scaling up isn't trivial: it is not just a matter of spending more money and resources.

Strong Platonic Representation Hypothesis: the universal latent structure of text representations not only exists, but can be learned and, furthermore, harnessed to translate representations from one space to another without any paired data or encoders.

Got recommend this substack from Leash bio by a friend. I think this is a masterclass in how to correctly split the data if there ever was one. Respect your chemistry folks! open.substack.com/pub/leashbio...

Not sure who came up with "Manhattan Plot", but in 2014 I coined the alternative term "Nijmegen Plot" (inspired by the Dutch town where I live) to describe underwhelming results from our earliest genome-wide association scans of language/reading traits.

Love these maps of "street-text sightings" in the Pudding's latest piece pudding.cool/2025/07/stre...

Great blog post on rotary position embeddings (RoPE) in more than one dimension, with interactive visualisations, a bunch of experimental results, and code!

Can an AI model predict perfectly and still have a terrible world model? What would that even mean? Our new ICML paper (poster tomorrow!) formalizes these questions. One result tells the story: A transformer trained on 10M solar systems nails planetary orbits. But it botches gravitational laws 🧵

Today's #RDKit blog post is a heartfelt plea for clearer communication. greglandrum.github.io/rdkit-blog/p...

There is a new startup from China called Moonshot. The original “moonshot” was the Apollo Program. An AI based moonshot could be referred to as an “AI pollo” program. “ai pollo” in Italian means something like “to the chicken”.

I was recently on a flight with free Wi-Fi for texting but nothing else. Jokes on them: I can use Llama through WhatsApp now …

The new #RDKit blog post, inspired by a question from @valencekjell.com, looks at the impact of molecular size on similarity thresholds. greglandrum.github.io/rdkit-blog/p...

Yay for @pschwllr.bsky.social and @mlederbauer.bsky.social (and all your co-authors who aren't on BlueSky yet) 🥳 This #dataset is a prime example of #GoodData, and it ties nicely with what @clarakirkvold.bsky.social and @grynova.bsky.social were talking about a few weeks ago in their #JournalClub

I've got a joke about Osysseus. I got lost on the way to the punchline...

change my mind: bruot RIS to Bibtex converter is the best website ever built. www.bruot.org/ris2bib/

If anybody out there working on antimicrobial resistance (AMR) and needs some motivation on this gloomy New England Monday.

I think the ranking of things which are hard to predict goes: 1. The stock market. 2. LaTeX figure placement. 3. The meaning of life.

Cheminformatics family businesses be like

One of the surprising things about working in a microbiology lab is that you become more worried about washing your hands before using the restroom rather than after.

I think the thing I'm most excited to see over the next ~10 years of #dataviz is web-based content that interweaves long-form text and modular interactives. Not as heavy as scrollytelling and not as aimless as a dashboard, but something in between. This is what I was going for with the QR project!

change my mind: bruot RIS to Bibtex converter is the best website ever built. www.bruot.org/ris2bib/

You know volatility is going crazy when sitting down to write a PAC proof about the sampling efficiency of an active-learning algorithm feels like a therapy session. At least math hasn't changed over the past 12 months ...

Not me accidentally typing `squeue` into the Facebook chat.

Thought experiments that you have down the pub are in vino

Has anybody ever paused to think that the glass that is metonymically meant by “in vitro” is actually also just made of silicon and could just as well be called “in silico”?

Do I really write "... reckons Deutsche Bank, a German bank"?

It’s always “We launched these tardigrades into space!” “We froze them to absolute zero!” “We threw them into an erupting volcano!” Never “We took these tardigrades on a holiday. They’re getting tiny massages and sipping little umbrella drinks on the beach.”

One day ChatGPT is going to learn to use MorganGenerator ... not today though.

Very true.

If a protein unfolds and there is no one to measure the CD spectrum, did it really unfold?

It's such a privilege to live in a time where I can fully pretend like the climate crisis isn't happening simply by living in constant existential fear about the replacement of all meaningful work by more powerful, intelligent systems. Oh and somebody said something about a recession?

I know Manus AI didn't hit the mark but with everybody rushing into agentic AI, I'm genuinely at the point where I'm wondering whether I need to get PhD applications out pre-singularity. Who still hires PhDs if they can have a 24/7 researcher which inputs a PI's command and outputs results/data?

New blog post - State of the lab 12, yearly series on being a group leader in academia. This one reflects back on the transition from junior PI to a more established scientist www.evocellnet.com/2025/03/stat...

Ah, yes, more top quality chemistry content.

The New Yorker is reäppraising the internet age

This is horrible to post, but I may as well post it. We are essentially shutting down research operations in my group, which is focused on treatments for pediatric brain cancer. I’m a well funded investigator, and there’s no choice. Science can’t function without the stability of NIH

In 20-30 years time, I wonder whether my children will look back on 2024 or 2025 the same way that I look back on 1968.

87% Of Loud Crashing Noises Are Nothing, Report Top Experts From Other Room theonion.com/87-of-l...

You have never felt relief like a homeless hermit crab the moment they’re safe inside a new shell

Helpful table to determine how much your University or health system is set to lose if the NIH indirect cap is suddenly lowered to 15%. For SJSU it is $374k which is on the smaller end but devastating considering our other budget issues. datawrapper.dwcdn.net/l0ZqA/8/

Ask an LLM to summarize your paper because some reviewer will for sure read your paper this way

The title of the month award goes to...

The hottest new programming language is E̶n̶g̶l̶i̶s̶h̶ Mandarin #DeepSeek

Has anybody’s ChatGPT also been starting to use an uncommon amount of emojis? Is this meant to lure me back from Deepseek?

Chemical Formulas xkcd.com/3040

Why do I have to rent software? Why can't I just buy it?

In 1980, the journal The Behavioral and Brain Sciences published John Searle's "Minds, brains, and programs". The origin of the Chinese Room thought experiment. Uniquely, it also published a response from 27 "qualified peers" to criticize Searle. That's how all papers get published there.

Autocorrect just changed “dear” at the start of an email to someone I don’t know to “dearest” and I didn’t notice before clicking send. How is everyone else’s day going?

and (20+25)² = 2025