They want everyone to use their AI tools.

Certain government, health care, and financial organizations are legally restricted from sharing data. Having worked in such a place before, they'll blanket ban AI rather than risk the unauthorized transmission of protected data. This functionality offers an alternative that they can accept.

If Microsoft wants their AI features to be adopted widely, they basically had no choice but to implement a local model. The fact that it's cheaper to use the customer's electricity is probably meaningless to them; they want to capture the userbase.

Compared to what was available before, it is quite impressive.

The negative feedback is prompted by the fact that AI is constantly being shoved into every one of our orifices 24/7 by every vaguely tech-related company as if it was the second coming of Jesus. To justify that amount of social pressure, it would indeed have to be quite a bit better than it actually is, and that's why people aren't impressed.

People who want to use AI are already doing it. They will use the AI that meets their needs the best.

For casual users, that's GPT, often ChatGPT, because it's ubiquitous. They can access it anywhere and get the same capabilities. That's what they care about: ease of use, ease of access, and consistency.

It's the same reason people still prefer Windows on the desktop. Except this time, Microsoft didn't get there first. So now they're the latecomer or the afterthought, and they don't like it. Too bad; innovate faster next time instead of jumping on a bandwagon.

You mean, why does Linus keep agreeing to be interviewed, and then reply to straightforward questions with the obvious answers?

What would you rather he do? Refuse to be interviewed, or maybe make up unexpected answers just to be edgy?

Let's put idiots in charge of formulating medical guidance.

We'll have the greatest measles epidemic in the world if we stay the course.

Sorry, still can't parse; who or what is supposed to be jewish again?

I found GP2.5 to be great at academic-style research and writing; it was absolutely awful at writing code. So; I would tell it to plan some thing for me and write it in a way that could be used by another agent (Claude Code) to build the code to do the thing. In this way, it has been great! I haven't yet attempted it with 3.

That said, I found GP3.0's page to be hilarious:

It demonstrates PhD-level reasoning with top scores on Humanityâ(TM)s Last Exam (37.5% without the usage of any tools) and GPQA Diamond (91.9%). It also sets a new standard for frontier models in mathematics, achieving a new state-of-the-art of 23.4% on MathArena Apex.

It then proceeds to show, lower down on the page, an example of what it can do, by showing off 'Our Family Recipes". If there's anything that touts PhD-level reasoning and writing, it's a recipe book.

... which, to be clear, is a good thing. Working around the compiler is dangerous and a code smell, so it shouldn't be something that is easy to do. It usually indicates that either the compiler's capabilities aren't sufficient to meet your needs (in which case, a better solution would be either a better compiler, or to re-evaluate the wisdom of your approach), or that you are doing something the wrong way and should find a way to do it that works with the compiler, rather than around it, so that you get the benefits of the compiler's co-operation.

They weren't just wrong about consumption/demand. I've said before that the peak oil wikipedia article has been one of the more amusing ones to follow for a while:
https://en.wikipedia.org/wiki/...

Figure 1: What we propagandized to you for decades and hoped to influence public policy with.
Figure 2: How pants-shittingly wrong we were.

Gave me flashbacks of Scott Hall's infamous night:

https://www.youtube.com/watch?...

https://www.youtube.com/watch?...