I've always just seen it as a floating, continually soft rebooting timeline more akin to The Simpsons.

But if you submit AI-generated output that is broken but doesn't have obvious tells, it may sail through. Meanwhile a paper that does have a smoothing line from an LLM left in by accident despite checking and re-checking, but is actually excellent overall, may get sanctioned. This doesn't seem to solve the issue, just remove some of the most obvious symptoms.

Depends if NASA actually approached them for a tailored solution or whether they just took it upon themselves to deploy an unmodified Windows 10/11 on something up there.

In the UK I have a Cineworld Unlimited card so I don't pay per movie - just a flat rate of 12.99 GBP ($17.72). In practice it probably works out around the same as the tickets in Egypt given I go > 5 times a month. It's very cheap leisure compared to most things *if* you have the card.

I just add half an hour to the start time.

This happened at the Glasgow Film Festival in 2013. We had a screening of 'Cloud Atlas' (which iirc was a very recent movie then) and the hard drive had been delivered and plugged in, but the distributor had given the unlock key for the wrong time. It took about an hour to track down the person in Los Angeles (mid-afternoon in Scotland, early morning there) who could issue a new key.

I typically go three times in a week (often the same day) so actually that was just me. I have a card for unlimited viewings. Sorry for messing up the stats.

If PornHub's user base skews male, and tech fields where Linux is more common are still dominated by men then that alone could account for a higher incidence of Linux usage among PornHub visitors. If anything after thinking about it for a minute I'd almost come to expect it.

They should just measure global performance metrics and if that clearly stratifies based on AI use they'll have some picture. But a major problem I foresee is regardless of what you think of AI, not all coders operate the same way. Some may benefit from extensive usage, others will be hobbled, even on the same tasks. There is unlikely to be a single way to use it that simply translates to more/less is better.

Even if it were to turn out to be true that white collar work *could* be largely done by AI in that time, there's still the barrier of actual adoption. Does he really think that it's going to be rolled out at that pace? When simply getting Microsoft's user base from Windows 10 to 11 is a massive ordeal?

Maybe a third party kernel like Xanmod rather than the Ubuntu one would help with nordrand?

I wonder if it might be time to make a choice between Ubuntu LTS and Debian as the base, rather than maintaining both. I don't know how much that could free up resource-wise. Though it would also alienate users on whichever base lost out, which would have a complicated transition process.

This is already the direction of travel or current reality with laws like the Online Safety Act in the UK. It seems like platforms may now choosing to pre-empt governments on this rather than question it, which is worrying.

I suspect a lot wouldn't know precisely but a larger proportion than that could reason it being mid to late 20th century when prompted as to what it was. I know off-hand it was during the Kennedy presidency which narrows it to '61-'63 with my guess being '62 since it feels for some reason like there was more than a year between the crisis and the Kennedy assassination.

This is why I always export my data every couple of weeks. It's actually allowed me to do some interesting things trying to see how much of the functionality I can replicate locally, putting my conversations into a vector DB and seeing if a local LLM will do anything like 'Memories' for example. But then the platform does nothing to encourage this behaviour when it really should. I'm also not aware of a way to import the zip file they give you *back* into ChatGPT.