Błażej Osiński

ML Researcher.

[Paul Graham @paulg]

There's never been an investment like the investment in railroads. (This graph has a log scale!)

[Błażej Osiński @blazejosinski]

How many breakthroughs away are we from this? @karpathy @lxbrun @ilyasut

[Błażej Osiński @blazejosinski]

In the future, finding an error in AI-generated code will be like finding a compiler bug: rare enough to earn years of bragging rights.

[Błażej Osiński @blazejosinski]

@MarioNawfal Do you still think that "the war will end soon"?

[Błażej Osiński @blazejosinski]

Sounds cool! Congratulations Latent Labs and @SeBayesian

[Simon Kohl @saakohl]

Today we're launching Latent-Y: the world's first autonomous agent for drug design, lab-validated end to end.

Give it a research goal. Latent-Y reasons, designs, iterates, and delivers lab-ready antibodies, autonomously or collaboratively, with the biological reasoning of a PhD protein design expert.

Technical report: https://t.co/E7IHfkvvD3
Blog post: https://t.co/GfJAfzj0Qx
Apply for access: https://t.co/E0SR9znZiP

[Błażej Osiński @blazejosinski]

Hear that? That's the collective sigh of relief from the world's professional Brainfuck developers. AI still can't replace them.

[Lossfunk @lossfunk]

2/ Our method: test them on esoteric programming languages.

Brainfuck. Befunge-98. Whitespace. Unlambda. Shakespeare.

All Turing-complete. All requiring identical reasoning to Python. All with 1,000-100,000x fewer GitHub repos than mainstream languages.

Same problems. Radically less training data.

[Błażej Osiński @blazejosinski]

I have not realised how much of a polymath I am until recently. I wonder why it took me so long to figure this out.

[François Chollet @fchollet]

There are two categories of people: those who quickly figure out that chatbots give you the answer you expect when you ask questions in a biased way, and the ascended polymaths currently out-thinking every expert on Earth

[Błażej Osiński @blazejosinski]

@jakozaur Thanks for sharing.
I'm sure that this will end up being only the first chapter of a great success story.

[Błażej Osiński @blazejosinski]

@hubermanlab In other words: not enough challenge is literally killing you!

[Błażej Osiński @blazejosinski]

@jakozaur Could you please elaborate what exactly eval do you run and why is it that crucial? You want to know in the real-time, which LLM is the best?

[Piotr Sankowski @piotrsankowski]

Instytut Ideas może zacząć działać - wczoraj dokonany został wpis do KRSu. Jednakże wśród dobrych informacji są też takie dużo bardziej przejmujące o chorobie jednego z liderów zespołów, który miał niedługo zacząć pracę. @LukeKucinski walczy z glejakiem i szuka wsparcia na: https://t.co/3z4P7hl5D2.

[Błażej Osiński @blazejosinski]

@MgShelby A czemu ten DAX ma all-time-high? Wydaje się, że problemów i ryzyk co niemiara, choćby groźby taryf.

[Błażej Osiński @blazejosinski]

@jakozaur And most of its big customers (Google, Amazon, Meta, Apple) are building their own chips for AI.
And there might be better chips, not just cheaper; e.g. https://t.co/8vXmDlyZxU looks promising.

[Błażej Osiński @blazejosinski]

Jevons Paradox: When a resource becomes more efficient, its consumption increases as new use cases emerge.
Today’s selloff in NVIDIA suggests the market forgot this paradox.
DeepSeek’s models can very well lead to more usage of GPUs, not less.

[Błażej Osiński @blazejosinski]

I think this is a very good take on AI - overhyped as hell, but the progress is real.
@filippie509, don't you agree? You seem to post a lot about the BS, but I can't figure out if you see the substance behind it.

[François Chollet @fchollet]

For about as long as I've been in AI, it has always been overhyped in some way -- shrouded in a cloud of bullshit. No matter the year, many commentators have assumed that it had incredible capabilities that it clearly didn't, and that it would soon pose risks that were clearly not real.

The trap is to think that, just because dim-witted people hype it up, AI isn't real. It's extremely real, and it will be a much bigger thing in 5 years than it is today, and then an even bigger thing in 10 years. It was already real and well on its way in, say, 2016 -- back when everyone assumed that no one would be driving their own car by the year 2020, back when some VCs were claiming that LSTM text generators were already good enough to replace journalists, back when doomers were already panicking about Bostromian fantasies.

The claims may be BS, but the tech is real, it's moving fast, and it drives an increasing fraction of the global GDP every year. That overarching trend is showing no sign of slowing down.

[Błażej Osiński @blazejosinski]

Sure, there are valid reasons to be cautious about nVidia's valuation (can they maintain a stable monopoly?).
But in my opinion, better models like DeepSeek's are not one of them.

[Błażej Osiński @blazejosinski]

@gkossakowski I think the most spectacular charismatic prototype to date is autonomous driving.

Already in 1995 there were demonstrations of driving on "public Autobahns from Munich to Denmark and back, up to 158 km without human intervention" citing @SchmidhuberAI https://t.co/IzgZ2IMI6Z

[Błażej Osiński @blazejosinski]

@m_gdula @IDEAS_NCBR Zapewne Pan i Rada Nadzorcza też sugerowaliby Google w 1999 nowy kurs w oparciu o ich niskie przychody?
https://t.co/WuTvzWjngB

[Jon Erlichman @JonErlichman]

In 1999, Google’s revenue was $200,000.

Today, it’s $200,000 every 24 seconds.

[Błażej Osiński @blazejosinski]

@m_gdula @DominikWro @IDEAS_NCBR Dobrze, że osoby zarządzające Google w 1999 miały oczekiwania.
https://t.co/WuTvzWjngB

[Jon Erlichman @JonErlichman]

In 1999, Google’s revenue was $200,000.

Today, it’s $200,000 every 24 seconds.

[Marek Cygan @marek_a_cygan]

Im więcej wiem na temat procesu wyboru nowego prezesa IDEAS NCBR, tym bardziej uważam, że w interesie publicznym jest transparentne wyjaśnienie kryteriów wyboru nowego prezesa. Poniższy list wysłałem do Premiera Donalda Tuska @donaldtuskEPP