Plenty of systems are set lime this. It is quite common in image generation for live streams.

The thing I have realized is that eventually, someone will figure out the "right" prompt that gemerates an image that bypassss the filter.

In my experience, it takes chat about 3 hours to get there.

Don't threaten me with good times!

I was struggling doing video editting on linux until I ran into DaVince Resolve Studio. Bought for $120 at the time. Best $120 I spent on software ever!

AI generated summaries maybe be an improvement on slashdot

I suspect tbag is true, or will be true.

The problem I see is tbat viewed from the outside, excellent code look like the dog shit that comes out of "please make xyz work with abc".

And since you can't tell the difference, you kind of have to treat everything AI generated the same.

I think there is only one good reason to build a space datacenter at this point. Avoiding regulations. Is it good enough? I doubt it.

But I agree the math doesn't seem to make sense. Cooling any sizeable deployment that would matter is impractical. The cost of putting it up is really high. It all come burn in the atmosphere is about 5 to 10 years, so you can't reuse anything for the next hardware cycle.

Emacs!

Even AI can't figure out how to count the parentheses in lisp!

At least part of the drop in users is because AI is being used to moderate posts and then use that "moderation" to temporarily and then permanently ban users, in the process completely deleting ALL posts on all pages. There is little to no human/personal moderation now - just AI makes the decisions and users have to suck it up. How do I know this for a fact? It's happened to me and I was given no opportunity to delete or argue for the "offending" post. This has happened to several friends as well for various reasons. You'd have to pay me to have anything to do with meta ever again.

Anecdotically, it seems harder to get jobs as fresh graduates now, at least in Computer Science.
Been teaching undergrad and grad students for about 15 years. seniors used to send 15 applications and get 3 interviews. Nowadays, they send hundreds and might not get a single call back.

If you look at the numbers you provided (thanks for that), the definition of recent graduates is "22 to 27 years old who hold college degrees". So maybe the first 3 month after graduation got much tougher, but within 5 years it evens out.

It is clearly not meant to address the fundamental issue. Remember that arXiv was never meant to be a peer reviewed repository. There are still tons of bad papers in there.
I am guessing it is meant to cut on the volume and avoiding polluting the system with too much crap.

Think of it as a spam filter. You don't need to filter all the spam. Just filtering the "this obviously is spam" already helps a lot.

Working for a chosenite and expecting ethics? LOL

Every time they tried to recruit me I deliberately wasted as much of their time and as little of mine as possible.

I did some video where I had a script. So it was easy to extract subtitle and compare to script. Yeah, its pretty bad. Probably every other sentence had some obviously incorrect subtitling. And these videos were on clean audio inputs: no background noise, good mic.
I ended up having to correct the subtitles using the transcript.

I have seen a study recently on noisy audio in a group setting, so the speakers maybe not be perfectly mic-ed and all. The speech-to-text was done with a couple of Whisper models. The error rate was something like 30% of the transcription was bogus. (yeah, 3 words in 10.) I imagine that YT would do some preprocessing. But overall, I find the subtitles not terribly good. It's useful and better than nothing.

One of the counter arguments - that I don't agree with by the way - is that if people are allowed to even try to repair stuff that they will "wreck it". My response is that if they do, that's on them as the "right to repair" means that it is repairable, not that anyone and everyone can do it.