segh

Does no one else notice the LLM-speak tone of this article? Does no one else mind?

gpt 5.5 thinking is more reliable than the median internet commenter

I'm curious, what is the LLM cost of the website?

This is an argument against all technological progress.

Being average is a just stage LLMs pass through as AI makes its way towards 'expert' and 'super human' levels.

I can do long division manually but I still reach for a calculator.

People skills :/

This is an experiment to see the current limit of AI capabilities. The end result isn't useful, but the fact is established that in Feb 2026, you can spend $20k on AI to get a inefficient but working C complier.

Far far more people use ChatGPT than Claude.ai

I started skimming and instantly thought AI, then came to the comments to see if it was just me.

School incentives are not really aligned around maximizing learning rate for every student. (E.g. that is why there is/was debate around teaching phonetics)

Cool experiment! My intuition suggests you would get a better result if you let the LLM generate tokens for a while before giving you an answer. Could be another experiment idea to see what kind of instructions lead to better randomness. (And to extend this, whether these instructions help humans better generate random numbers too.)

ChatGPT's tone is slowly taking over the entire internet

People still play chess, even though now AI is far superior to any human. In the future you will still be able to hand-write code for fun, but you might not be able to earn a living by doing it.

The live demos are using a very cheap and not very smart model. Do not update your opinion on AI capabilities based on the poor performance of gpt-4o-mini

The system prompt can include examples. That is often a good idea.

For me, they have come from the AI labs themselves. I have been impressed with Claude Code and OpenAI's Deep Research.

Lots of people are building on the edge of current AI capabilities, where things don't quite work, because in 6 months when the AI labs release a more capable model, you will just be able to plug it in and have it work consistently.

This is the crux of the issue. Whether you think this is like extending a ladder to the moon, or more like we figured out how to get to the moon and are now aiming at Jupiter.

Claude Plays Pokemon is one person's side project to see how well Sonnet can play pokemon. It is a neat LLM benchmark; it's not a serious attempt at making Pokemon-playing AI.

Do you disagree that AI will ever reach the level of a "high-income knowledge worker", or do you disagree that it will happen in a year or two?

If you have not been reading every OpenAI blog post, you can't be blamed for thinking the model picker affects Deep Research, since the UI heavily implies that.

I don't know how literal you are being, but it is not "literally anyone". If you know what FFmpeg is, you are a very small minority of the population.

Improve in intelligence and capability.

Enforce a no politics rule.

Here is another online SICP with built in interpreter.

http://xuanji.appspot.com/isicp/

One problem with cheap or free is that when you are trying to validate your product, a user paying you good money is far more validation that a free user.

That's not the conclusion of the link you posted. The conclusion is more like:

After aspartame is consumed, it immediately breaks down into three naturally occurring chemicals. Even large amounts of aspartame cause smaller fluctuations in those chemicals than normal food. The current science says that the health impact of aspartame is essentially zero. Every credible body that has studied this question has reached the same conclusion.

As a percentage of GDP, the UK has the 6th highest healthcare spending of OECD countries, on par with Switzerland. The UK is not an outlier in terms of spending.

https://stats.oecd.org/Index.aspx?DataSetCode=SHA

I'm not sure it's just incentives. Inexperienced early stage founders often end up solving imaginary problems, despite having a real incentive to get it right. The Y Combinator moto is "make something people want" because so many people don't.