This is from January 2026

14:00 “KEEP IT AWAY FROM AUTOCRACIES”
smiles in palantir

Great interview and excellent questions

If all that they say ever come true, by now we all should be travelling around in flying cars. No more roads and no more bridges that's a huge saving for the world. It's always super hype to pump up the price of the stocks of the corresponding companies. They only think of money day and night no real meaning in life how life used to be in 1900s. Life was bitter sweet, we had family, friends, communit and meaningful life. Today we are all living in a pressured cooker.

We’re at Davos

Interesting until Opus 4.5 is mentioned as ”state of the art”… Should have been published 3-4 months ago.

This is an old interview from January.

Posted a day ago but already feels a year old. He mentions just launching a cloud version that’s now 2 versions behind current.

So far Im lucky to have Dario and his team managing one of the most powerful AI in the world. Not sure about the future though...

Needs timestamps

AI technology was supposedly used by researchers but released it to the public? that's like opening a Pandora's box

One mistake: The cost are rising exponential, not the outcome.

The proposed legislation seeks to increase the Currency Transaction Report (CTR) limit from $10,000 to $30,000.

Do they really think? Maybe that is the problem. Deeming the illusion of thinking (predicting the next token/word segment) as thinking rather than a ghost in the machine represented by the data it was trained on.

Clear and concise briefing❤️, but gdp's may shrink due to lower cost applied to brain intensive tasks. isnt🤔

What was that piece of paper all about?  ( 12:15 )

Key issue that was not mentioned: frontier models are no longer available to the common user. Dario is creating a new class system where a small portion of society has access to full powered AI and everybody else gets AI that is slightly stoopid. Mechanistic interpretability cannot solve this problem -- at the end of the day "nerfed" versions of models have less accurate world models and will always be dumber than the frontier. The "enterprise" framing helps Anthropic gloss over this issue for now, but it is one of the most important social issues of our time.

For enterprise rollouts of Claude Code context window management: what fits and what does not., start with permissions and data boundaries, not feature demos. I shared a checklist on my profile.

This guy says using AI for pennies to build apps, but they have the most expensive API in the market

I use AI extensively and have been able to build fairly sophisticated electronic devices despite having little background in electronics or coding. So the increase in productivity is indeed revolutionary and indeed will lead to higher GDP. After that the questions seem to be essentially political. Who gets the wealth and who does AI serve? These are solvable, but it means making Democracy stronger, which is on all of us. The Industrial Revolution may provide a model: probably 50% of all jobs were wiped out and  people’s standard of living initially declined. But ultimately politics kicked in and we got shorter working weeks, higher pay and a vast expansion of “white collar” jobs. This time we know in advance this trajectory, let’s hope we can navigate it better.