krumeto.bsky.social
@krumeto.bsky.social
Senior Data Scientist at Financial Times.
Python/R, ML, NLP, but also the occasional local politics.
Dad of two girls.
Opinions are my own.
Reposted
Had never read Jane Goodall's original 1963 article in Nat Geo on the wild chimpanzees in Tanzania until now. It's a wonderful blend of science and journalism, and well worth your time.

www.nationalgeographic.com/pdf/jane-goo...
October 1, 2025 at 8:15 PM
Reposted
EU–INC is the single best thing Europe could do to catch-up in the AI race

A simple unified pan-European startup structure, with modern employee ownership and simple access to capital, able to tap into Europe’s full talent pool.

‼️ but it’s at high risk of not seeing the light of day. You can help👇
September 27, 2025 at 10:52 AM
Reposted
A cool use of AI in newsgathering, this. US companies are talking more and more about the risks rather than the benefits of AI in their SEC filings (while still being super-optimistic in earnings calls). By @melissahei.bsky.social @chriscook.news & @claradoodle.bsky.social www.ft.com/content/e93e...
September 23, 2025 at 11:45 AM
Reposted
Andor would want you to cancel your Disney+ subscription.

It's the least you can do for the Rebellion.
September 20, 2025 at 4:15 AM
Reposted
September 16, 2025 at 2:40 PM
Reposted
An update on this: I've confirmed that "gpt-realtime has a mix of data specific enough to itself that its not really 4o or 5" - see quote from OpenAI at bottom of simonwillison.net/2025/Sep/1/i...
Introducing gpt-realtime
Released a few days ago (August 28th), gpt-realtime is OpenAI's new "most advanced speech-to-speech model". It looks like this is a replacement for the older gpt-4o-realtime-preview model that was rel...
simonwillison.net
September 2, 2025 at 4:46 PM
Reposted
I was interviewed in the Financial Times about "the perils of vibe coding", and it made it into the print edition! simonwillison.net/2025/Aug/29/...
The perils of vibe coding
I was interviewed by Elaine Moore for this opinion piece in the Financial Times, which ended up in the print edition of the paper too! I picked up a copy …
simonwillison.net
August 29, 2025 at 6:04 PM
Reposted
Big news - Anthropic agrees to settle with authors in copyright lawsuit.

As tech bros like to say, ‘we’re only just getting started’.

www.reuters.com/sustainabili...
August 26, 2025 at 7:01 PM
Reposted
“It’s perfectly consistent to advocate for a policy of ‘No one is allowed to do x,’ but then if that policy fails and everyone else does X, to reluctantly do x ourselves.” 🤔

X=help dictators

www.wired.com/story/anthro...
Leaked Memo: Anthropic CEO Says the Company Will Pursue Gulf State Investments After All
“Unfortunately, I think ‘No bad person should ever benefit from our success’ is a pretty difficult principle to run a business on,” wrote Anthropic CEO Dario Amodei in a note to staff obtained by WIRE...
www.wired.com
July 22, 2025 at 1:34 AM
A link to the Bulgarian model, in case somebody is interested - huggingface.co/HPLT/hplt_be...
July 18, 2025 at 12:21 PM
Reposted
That was my theory going into the piece, but it turned out to be not only false but seemingly the opposite: the biggest employment gains young male grads have made in the past year have been in software and engineering jobs
July 18, 2025 at 11:31 AM
Reposted
Hah, missed this that other say

Elon: "It is surprisingly hard to avoid both woke libtard cuck and mechahitler!"

(Narrator: this was not surprising at all)
"looking back at the audience for approval like a contestant on the price is right"
July 15, 2025 at 6:38 PM
That's fun!
posit.co Posit @posit.co · Jul 14
Announcing Orbital for Python! For Scikit-learn users, this tool transforms your ML pipelines into SQL queries, letting predictions run directly in your database without a #Python environment.

Learn more: posit.co/blog/introdu...
July 14, 2025 at 3:25 PM
Reposted
Advertising, with guardrails and healthy metrics, funds a healthy media ecosystem.
When people run into a paywall, 99% don't end up paying. Half go elsewhere. A third give up.

I get why people paywall and do not judge your payday, but this in a nutshell is why I’m uncomfortable with it.

This is from Pew:

www.pewresearch.org/short-reads/...
July 13, 2025 at 6:20 PM
Reposted
A very good, and rather sad, Big Read on children and reading from @emmavj.bsky.social

on.ft.com/4krOKZA How to get children reading again
July 4, 2025 at 7:20 AM
Reposted
ChatGPT referrals to news sites are growing, but not enough to offset search declines
ChatGPT referrals to news sites are growing, but not enough to offset search declines | TechCrunch
Not surprisingly, organic traffic has also declined, dropping from over 2.3 billion visits at its peak in mid-2024 to now under 1.7 billion.
techcrunch.com
July 3, 2025 at 10:46 AM
Reposted
‼️Sentence Transformers v5.0 is out! The biggest update yet introduces Sparse Embedding models, encode methods improvements, Router module for asymmetric models & much more. Sparse + Dense = 🔥 hybrid search performance!

Details in 🧵
July 1, 2025 at 2:00 PM
Reposted
Wild.
June 30, 2025 at 12:29 AM
Reposted
Yesterday I used this "AI -Free Statement" in a conference talk for the first time. Still trying to figure out the exact language.

Feel free to borrow, modify, etc.
June 27, 2025 at 12:12 PM
Reposted
Honestly, copyright never seemed like the right approach to this to me, so I'm not surprised.

Copyright law was designed in the era of books, and only slightly updated for the internet. It was meant to stop people literally *copying* a work. It has no appropriate tools to deal with generative AI.
A judge ruled Meta’s AI training on copyrighted books is fair use, calling it transformative since the AI didn’t reproduce the books. The authors also failed to show it harmed their market, leading the court to side with Meta.

Another win for the AI industry after Anthropic’s win in a similar case.
Federal judge sides with Meta in lawsuit over training AI models on copyrighted books | TechCrunch
A federal judge sided with Meta in a lawsuit that alleged the company had illegally trained its AI models on copyrighted works.
techcrunch.com
June 26, 2025 at 2:12 PM
Reposted
Once again, it must be said as loudly as possible: AI can't do the things described in this paragraph, and there is no pathway to it ever being able to do them.
June 25, 2025 at 11:41 AM
Reposted
The copyright war between the AI industry and creatives https://on.ft.com/3T450Vm | opinion
The copyright war between the AI industry and creatives
We have surely gone beyond being able to give the tech sector the benefit of the doubt
on.ft.com
June 23, 2025 at 4:22 AM
Reposted
It’s fascinating to see Microsoft and Apple go in such different directions when it comes to AI.

Microsoft went all-in and wrote “Sparks of AGI” www.microsoft.com/en-us/resear...

Apple has done the bare minimum and just wrote “The Illusion of Thinking” ml-site.cdn-apple.com/papers/the-i...
ml-site.cdn-apple.com
June 8, 2025 at 12:40 PM
Reposted
South Korea gender divide update 😲

Young men lean right by 50 points
(74% conservative vs 24% centre-left)

Young women lean left by 22 points
(58% centre-left vs 36% cons)
June 4, 2025 at 10:01 AM