MP_1729

This is Eliezer Yudkowskyto to blame LMAO, alongside with all the EAs extremists that shifted the Overton Window to think that images can be unsafe.

Because tradition is incredibly important.

Really funny people complaining about classical chess, I don't know which games they have been watching.

Nepo Magnus game 6, Nepo Ding many many games, Nepo Caruana draw on round 14 of candidates. ALL OF THEM WERE TERRIFIC GAMES.

I don't understand what people mean by stagnation

I'm in a low-cost country, haha! So the intern is even cheaper.

25 prompts is the daily limit on o1-preview. And I wrote that script in just one day.

My 3rd day intern still couldn't do a script o1-preview could do in less than 25 prompts.

OBVIOUSLY a smart OAI employee wouldn't want the public to think they are already replacing high-level humans.

And OBVIOUSLY OAI senior management will want to try to convince AI engineers that might have 2nd-guessings about their work that they aren't developing a replacement for human beings.

But they are.

That's not how pricing works.

If o1-pro is 10% better than Claude, but you are a guy who makes $300,000 per year, but now can make $330,000 because o1-pro makes you more productive, then it makes sense to give Sam $2,400.

My new internet is in his 3rd day in the job and he's still behind o1-preview with less than 25 prompts.

Good. Effective Austruism is trying to destroy democratic institutions and it's likely a bigger threat to society than fascism. The pro-progress people needs to organize themselves to stop types like SBF from giving AI to China.

Satya Nadella's Microsoft is such a weird company. It's like there's one side of it that is running with Zuckerberg's "move fast and break things" and the other side is saying "wait, we're the most important software company in the world! Things can't break!"

“You want to know how to paint a perfect painting? It's easy. Make yourself perfect and then just paint naturally.” - Robert M. Pirsig

The Musk reasoning here is stupid, but smart. If he makes a superhuman intelligence, he can only ask it "What is dark matter?" and it might figure out.

I have some big problems with this idea, but it isn't 100% stupid. Just 98% stupid.

It's more like: don't invest $100B in capital when there's still orders of magnitude improvements in software to be made.

Ask stuff like "Check whether there's some correlation between the major economies fiscal primary deficit and GDP growth in the post-pandemic era" and get an answer.

Obviously, I know these timetables.

But there's a light and day difference post-Nov22 than before. Both in the AI race it sparkled, but also in the funding all AI labs have.

If you're expecting GPT-5 by 2026, that's ok. Just very weird to me.

Nope, I'm not

Was there ever a time when betting that computer scientists would not make better algorithms was a good idea?

I'm curious now. Can you give color on what you're doing that you keep hitting boundaries? I suppose it isn't limited by human-attention.

they literally launched TODAY a GPT-4 model!

Two pushbacks.

1- The mania only started post Nov 22. And the huge investments since then didn't meant substantial progress since GPT-4 launch in March 22. 2- We are running out of high quality tokens in 2024. (per Epoch AI)

I am sure that people at OpenAI, particularly former YC CEO Sam Altman, will be fine, even if they read the bad stuff MP_1729 says around here.

All of them, without exception. Just recently, Sprint sold their fiber business for $1 lmfao. Or WorldCom. Or NetRail, Allied Riser, PSINet, FNSI, Firstmark, Carrier 1, UFO Group, Global Access, Aleron Broadband, Verio...

All fiber went bust because despite internet's huge increase in traffic, the amount of packets per fiber increased a handful of magnitudes.

That's the opposite of what I am saying.

What I am saying is that computers are SO GOOD that AI is getting VERY CHEAP and the amount of computing capex being done is excessive.

It's more like you are in 1999, people are spending $100B in fiber, while a lot of computer scientists are working in compression, multiplexing, etc.

I am just talking about scaling laws and the level of capex that big tech companies are doing. One hundred billion dollars are being invested this year to pursue AI scaling laws.

You can be excited, as I am, while also being bearish, as I am.

I read Simon Sigh during high schools and that was such a beautiful story of perseveration that I decided to do Math.

This thing continues to stress my skepticism for AI scaling laws and the broad AI semiconductor capex spending.

1- OpenAI is still working in GPT-4-level models. More than 14 months after the launch of GPT-4 and after more than $10B in capital raised. 2- The rhythm that token prices are collapsing is bizarre. Now a (bit) better model for 50% of the price. How people seriously expect these foundational model companies to make substantial revenue? Token volume needs to double just for revenue to stand still. Since GPT-4 launch, token prices are falling 84% per year!! Good for mankind, but crazy for these companies. 3- Maybe I am an asshole, but where are my agents? I mean, good for the consumer use case. Let's hope the rumors that Apple is deploying ChatGPT with Siri are true, these features will help a lot. But I wanted agents! 4- These drop in costs are good for the environment! No reason to expect them to stop here.

Simmons is one of the greatest people and a true inspiration as a mathematician, even though my career drifted from academia. He and Andrew Wiles are the reason why I always say I am a mathematician, even though I work elsewhere.

RIP

Beware Goodheart's Law: "when a measure becomes a target, it ceases to be a good measure". If your goal is stopping to waste time solving bugs, I'm sure you're going to be able to do that.

You should have an important counter-metric to see if you're not messing with software. It could be number of reported bugs, crashs in production, etc.

Financial market applications of "transformers", not LLMs

It seems hard for people to take a nuanced approach that GPT-4 level models have in the present the potential to improve many people's lives and corporations' bottom lines while still being cautiously pessimists on the next generation of models.

GPT-4, Claude 3 & Co. are simply too useful for certain coding tasks or to review a contract. Obviously, you need to understand you're dealing with a probabilistic being, and for many tasks it isn't the correct tool, but I use ChatGPT ~5 a day and the $20 are super well spent. Now, there's an ocean apart of me liking to use ChatGPT and the promises from Silicon Valley and Microsoft.

Mark said in a podcast they are currently at MMLU 85, but it's still improving.