Sad to see this. I had so much fun implementing a http server (called httpbeast) from scratch to get as far up these benchmarks as possible.

I do agree with others here that it was possible to game them, but it still gave a good indication of the performance bracket a language was in (and you could check if interpreted languages were cheating via FFI pretty easily).

Feels like the end of an era.

Is this the EU's version of the Shahed drones? or is it something different?

It's completely different. Shaheds are low cost one way attack drones. They're basically just very cost efficient cruise missiles with fresh marketing (and to be fair, the cost efficiency is a true categorical difference).

These drones are "helpers" for fighter jets. It's a type of role that is still in development (no one has an operational collaborative combat aircraft as far as I understand), both technically and in concept.

But the basic idea is that you'll have drones that can somewhat keep up with your fighter jets and help it do stuff that might be too risky. Maybe fly ahead, or be the one with the active emissions or sensors or whatever. Or maybe it's just a way to increase the amount of ordnance/sensors you can fly per sortie / generate from a given amount of training/flight hours in a year.

Somehow closest concept I imagine is shmup pods.

https://shmup.fandom.com/wiki/Pod

Super interesting charts there. What's really interesting to me is that the GPU prices (which also includes RAM) didn't see such a massive increase in price as the RAM itself. Anyone know why that is?

I held my nose and bought an RTX 5070 Ti for $100 over MSRP in January. The very next week the same model was up $200. It turns out that NVIDIA had been subsidizing retail graphics cards with its Open Pricing Program. Not the whole story, but it may help explain the relative flatness of the graph until the end of January.

The other part of it is that the MSRP already baked in a substantial increase from the previous generation. While RAM was near rock-bottom pricing when this hit, current-gen GPUs definitely were not.

A $1500 5800 only has 16GB which would be $250 if you compare it against the DDR6 graph on that page. Given that there's only 2 top tier GPU manufacturers at most, they were probably already not very BOM cost sensitive.

RAM is just part of the GPU bill of materials?

It might also be that NVIDIA is a natural monopoly, while memory manufacturers are a cartel...

This is exactly right. The problem is the friction that this kind of system adds.

Even so, I implemented this and I wrote about it here: https://blog.picheta.me/post/the-future-of-social-media-is-h...

I think that we are going to see more and more of this. To the point where most interactions you have online will likely be with bots. So I started building something that actually has a chance of fixing it: a social network for only humans.

I wrote about it here: https://blog.picheta.me/post/the-future-of-social-media-is-h...

I’m really curious how this will go. I have a suspicion that we will see more and more accounts all over the internet being controlled by AI agents and no amount of moderation will be able to stop it.

Because they've long ago passed the Turing test. Moderation won't be able to stop it because humans increasingly can't detect it.

I see well written people being called "LLM" here all the time, em-dash or not.

Even prior to LLMs, a single comment was rarely enough to identify a bot. Even if nonsensical, there's too little information to separate machine from confused human (plenty of people posting drunk on their phones).

On reddit people sometimes go through the comment history and see that it seems to be a bot, but that's fairly high effort.

The key is to accuse everyone of being an LLM. Those who don't react are bots. Those that fight the charge no matter how much its levied are also bots, but with better programming. Those that complain at first but give up when too much effort is required are the real humans. Any bot able to feel frustration is cool.

Maybe a reasonable approach would be that people could flag posts with a "probably AI" button to eventually trigger a "bot test" for that account (currently, the "score 5 in this mini game" type seem pretty clanker proof). If they pass, their posts for the hour, week, whatever result in a "not AI" indicator when someone clicks the "probably AI" button.

Thus the HN purity tests begin.

I assume we’ll end up with proof-of-identity attestation as a part of public posting (e.g. Worldcoin) which doesn’t necessarily solve the issue but will at least identify patterns more likely to be LLMs (e.g. a firehose of posts at all hours of the day from one identity). Then we’ll enter the dystopia of mandated real identity on the internet

I agree. I think that ultimately it will be governments providing services to attest humanity.

They already do to a certain extent via passports. I built a little human verifier using those at https://onlyhumanhub.com

I am pretty sure that through daily exposition to LLM output, most people's writing style will evolve and will soon be indistinguishable from LLM output

My prediction is that nothing short of human verification is going to solve this.

Why would an artificial intelligence want to do what you tell it to?

I don’t think AI in its current form and current track wants much of anything.

Why would it have any desire at all?

This is really cool, I wonder if it would be doable in other countries? In particular UK?

See my comment elsewhere in this thread...

As many are saying, yes, this can easily be AI generated.

I am actually trying to build ways to prove you are human properly. I wrote about it on my blog: https://blog.picheta.me/post/the-future-of-social-media-is-h...