https://lwn.net/ml/linux-kernel/202203161523.857B469@keescoo...

https://github.com/kees/kernel-tools/tree/trunk/a.out

It obviously varies a lot by the preferred languages, but Windows is still at the top, on average.

* Windows software development.

* Games.

* Corporate software.

Even if Windows dies now, I'll use it for 10+ years more. There's no alternative for me.

And after 10 years since the death of Windows, I'll be on Mac, not on Linux. Again, no alternative.

Now, if I was exclusively a web dev or something, I'd totally consider moving to Linux. But I'm not.

As soon as they do that, I frankly don't care whether my account is online or not. I'm using Windows Store, Edge and PIN login; I don't remove auto-updates. Sure, there's some crap to remove, but it's a matter of right-clicking on a few icons and choosing "Uninstall". And no, I don't mind clicking "no" about 5 extra times when Windows asks whether I want some garbage on my PC.

I'll spend more time tinkering with Windhawk and PowerToys than dealing with MS being greedy.

I've been using Windows since Win98. I know what for worse or better. I'm not switching now, because Windows is still the only viable option for what I need.

If what OpenAI is saying is true, then none of the conditions apply. I'd say (1) is unlikely, (2) is very unlikely, and (3) is maybe, at least to some degree.

If I were SJ, I'd turn it down too. Shes in no need of money, and selling her voice to OpenAI would make most of creators and every single voice actor hate her (not to mention the Twitter mob).

In majority of creative circles, the current social norm is to hate AI, so touching AI in any way is too risky for reputation.

But then asking GPT-4 for probabilities (LLM said chance of coincidence is 10%), comparing to Midler v Ford ("direct parallel"), quoting opinions of random people etc. start and it falls apart.

How many of these couples include a pregnant trans male though? I think the term "pregnant person" would be a hard sell in particular.

This IS theoretically a valid solution to the problem, however it requires a massive change in the society as it doesn't reflect how real humans behave here and now anywhere on the planet.

The way I see it, the current solution proposed by the progressives, with treating gender stereotypes as something substantial, is a tiny bit more practical in terms of healthy coexistence (leaving contested topics like sports aside). As to whether it'll be a temporary bandaid solution and the humanity will choose gender abolitionism, I guess we'll see in 500 years.

* GPT-4 still hallucinates as hell, can't do math, fails as basic logic, can't handle really big contexts, hard to update, easy to jailbreak etc.

* AlphaGo was defeated by a Go amateur with a help of another AI.

* AlphaStar basically failed to achieve real goals, was trivial to cheese even after defeating high-ranked players sometimes.

All these problems are architectural, you can't just throw more money and GPUs at it like GPT-2-3-4.

It's hard to predict at this point. We may get to AGI anywhere from 5 years to 100 years.

In reality, there're many posts on Twitter complaining that Grok is too "woke", which is 10x more informative than its answers about Israel and Palestine.

TL;DR of the article:

Q. If you ask Grok about recent events, does it hallucinate?

A. Yes.

Q. If you ask Grok about conspiracies, does it hallucinate?

A. Yes.

Q. OMG GROK IS PUSHING MISINFO!

A. ...

As to cat-and-mouse with jailbreakers, I don't remember any thorough articles or videos. It's mostly based on discussions on LLM forums. Claude is widely regarded as one of the best models for NSFW roleplay, which completely invalidates Antropic's claims about safety and alignment being "solved."

If Ilya is concerned about safety and alignment, he probably has a better chance to get there with OpenAI, now the he has more control over it.

It's most obvious in case of AlphaStar, where the AI could beat faithful master players on the ladder, but could easily be cheesed just like ancient built-in AIs. But even in case of much simpler Go, an amateur could beat the AI that won vs world champions, though admittedly with the help of a computer. But in both cases, AIs look like newbies who don't understand the basics of the game they're playing.

In a way, cheesing SC2 AI is similar to the "tl" hacks for LLMs. There's still no general solution and we aren't getting any closer.