make3

The most concerning is the use of these tools to detect cheaters in schools.

are the other devices waterproof

Not if it's going to keep dropping, which I think is fair to assume as people wake up to the fundamentals after the hype is done (not financial advice)

wow that's kind of crazy impressive that it can do that honestly, VLMs have gone so far, can't imagine the crazy amount of annotations they had to create to get to that level

Second group essentially is how you have to think of it

They need to sell like 300+ per single Nvidia B100 GPU lol I'm not sure they'll quite fund themselves with this

no, last stage is removing you completely from the equation

I'm sure 99% don't? Unless your abuser is a CS major

I don't expect that Mamdani will cave in to real estate lobbyists lol. What you're describing is exactly why Establishment Democrats are losing to Mamdani and his ilk (DSA)

Yes, I really don't think that such a high fraction of people get purchased for fucking 15M after one year lol

Yes, exactly this. Migrant workers, under the threat of being deported if they don't work enough, obviously can be forced to accept much worse work conditions and work-life balance.

Every drug indeed has a huge burden of proof before it is distributed to humans, including the 3 levels of FDA clinical drug trials that cost billions of dollars, as well as all the internal testing we do beforehand.

there's no benefit to anyone for being so reductive, from whichever side you're on.

As a progressive person, the constructive thing is to try to see how to harness it for good or how to fight it, it's not to pretend that it's weak. It's not weak.

I professionally use it to design pharmaceutical drugs, essentially I do my best to harness the good sides of it

"the code is not quite detail oriented enough to be AI", times are changing

Efficiency is an angle of research that's quite active, see quantization, MoE, speculative decoding, QKV cache optimizations, sparse attention, etc

the thing went from the transformer paper in 2017 to basically doing your job for you in 2026

I don't think that people relying on the tools too much is the first sign I would identify to mean the bubble is popping

launches suck in the US too. See: every single AAA game launch ever

it's not, the genome is treated locally by tools called by the LLM, the LLM itself can't do much with the raw DNA sequence

it's also a lot faster I would assume

it might be worth it if speed is an issue

AI cost companies billions of dollars to make the time of hundreds of expert researchers. You're saying that my tool became too good so I can't make money from it anymore?

What about protein language models that create new protein drugs, can pharmaceutical companies not protect them too? That's straight up the same time of model as LLMs and image models.

Claude is a computer program, so is Suno. Someone has to pay Anthropic & to run Claude. AI does not have special moral grounding in our society.

That's honestly so dumb, if I use a non AI computerized tool to generate orders of notes or orders of characters, I own the output. AI is just that. It's a fancy computer program that cost billions to build.

This is giving weird independent moral grounding to AI as more than a computer that has never existed before. And what kind of AI does it count for ? Does it also count for image classifiers? For image quality improvers? etc

toupée fallacy

People for sure have always cheated in these take home exams. This has to have been to protect rich kids with parents who give money to the university. It's insane to learn how many fancy universities have garbage blatantly unmeritocratic evaluation systems like this

you can give people laptops with Internet disabled, that's a good solution

A more spikey distribution exactly makes the distribution closer to deterministic. That's not the point though. Even in greedy (deterministic) decoding, it is still a black box though that reacts in ways ways that are unpredictable to the inputs. Switching one word around might lead to different scores for example.

This is just a pointer on how not to sound like Sheldon from TBBT or like Claude Opus 4.8 with thinking set to Max in casual conversation.

Not saying it's better though.