I am “GI” without the A last time I checked, and I don’t require much compute at all!
Just a couple million years of training!
OpenAI and Elon Musk
391–400 of 1001 posts
Re: OpenAI and Elon Musk
#392All of these people are so bizarre. It's like Howard Hughes versus some potential acquisition turned competitor and the yellow journalism that followed. Time is a flat circle.
Re: OpenAI and Elon Musk
#393Earlier quoted context omitted.
It just means that LLMs are an interpolation of everything on the internet. They would seem less like they have a point of view or an opinion on things.
They would have the average point of view of the Internet, which is far from truthful or even useful.
They would have whatever you prompted them to have, minus the guardrails.
Re: OpenAI and Elon Musk
#394The timestamps of the emails are interesting. apparently they work at 3:40 AM but also at midday and at 8 AM
Re: OpenAI and Elon Musk
#395> For example, Albania is using OpenAI’s tools to accelerate its EU accession by as much as 5.5 years What an insane thing to say so matter of factly. This is like a character in an airport bookstore political thriller who is poorly written to be smart.
Why is that such an insane thing for them to say?
Last time I checked, the decision to admit a new member state requires the unanimous approval of the EU's current member states. As such unless OpenAI can literally influence world politics the claim is 100% bogus. All it takes is one rogue member and Albania will never join the EU.
Re: OpenAI and Elon Musk
#396Earlier quoted context omitted.
Well, the Manhattan project springs to mind. They truly thought they were laboring for the public good, and even if the government let them wouldn’t have wanted to publish their progress. Personally I find the comparison of this whole saga (deepmind -> google —> openai —> anthropic —-> mistral —-> ?) to the Manhattan project very enlightening, both of this project and our society. Instead of a centralized government…
They literally created weapons of mass destruction. Do you think they thought they were good guys because you watched a Hollywood movie?
I think they thought it would be far better that America developed the bomb than Nazis Germany, and the Allies needed to do whatever it too to stop Hitler, even if that did mean using nuclear bombs.
Japan and the Soviet Union were more complicated issues for some of the scientists. But that's what happens with warfare. You develop new weapons, and they aren't just used for one enemy.
Re: OpenAI and Elon Musk
#397> For example, Albania is using OpenAI’s tools to accelerate its EU accession by as much as 5.5 years What an insane thing to say so matter of factly. This is like a character in an airport bookstore political thriller who is poorly written to be smart.
Indeed. I found the claim about Icelandic even more telling. Here's a small language that basically exists in its present form since the past thousand years (i.e: it has changed little since Old Norse). It also is notoriously conservative in its preservation of native vocabulary through avoidance of loanwords.
Iceland/Icelandic don't need gee whiz computer things to "preserve" itself.
Re: OpenAI and Elon Musk
#398Earlier quoted context omitted.
> Collectivism involves deciding for others. That may be true, but that's not sufficient to define collectivism. There are many other forms of societal structures where "deciding for others" exists as well. Unless you mean to lump all these together, and say that companies and tyranny for example are the same as collectivism? > If it’s fully voluntary it’s individualism. If I _voluntarily_ decide to join a "collectiv…
> If I _voluntarily_ decide to join a "collective", am I individualist or collectivist? An individualist, if you are free to leave it at any time. There's nothing wrong with forming a collective in the US, I think like 20,000 of them have been formed over the last 240 years. You don't hear about them much because they all failed. You're free to start a collective anytime in the US and try to make it work. Isn't freed…
They haven't all failed. I hear about REI quite a lot. Rainbow Grocery is quite popular in SF. I hear good things about Organic Valley. Equal Exchange is in Massachusetts. It's popular to bank at a credit union instead of a bank.
The NCBA maintains a list of several thousand collective/coop businesses.
Re: OpenAI and Elon Musk
#399Earlier quoted context omitted.
I mean, you have accurately summarized the exact thing that safety advocates want. :) > legally halt ai progress everywhere in the world and carefully monitor large compute clusters This is in fact the thing they're working on. That's the whole point of the flops-based training run reporting requirements.
Reporting requirements are not going to save you from Chinese, North Korean, Iranian or Russian programmers just doing it. Or some US/EU based hackers that don't care or actively go against the law. You can rent large botnets or various pieces of cloud for few dollars today, doesn't even have to be a DC that you could monitor.
Sure, but China is already honestly more careful than America: the CCP really doesn't want competitors to power. They're very open to slowdown agreements. And NK, Iran and Russia honestly have nothing. The day we have to worry about NK ASI takeoff, it'll already long have happened in some American basement.
So we just need active monitoring for US/EU data centers. That's a big ask to be sure, and definitely an invasion of privacy, but it's hardly unviable, either technologically or politically. The corporatized structure of big LLMs helps us out here: the states involved already have lots of experience in investigating and curtailing corporate behavior.
And sure, ultimately there's no stopping it. The whole point is to play for time in the hopes that somebody comes up with a good idea for safety and we manage an actually aligned takeoff, at which point it's out of our hands anyways.
Re: OpenAI and Elon Musk
#400Earlier quoted context omitted.
“ To some extent, they may be right that open sourcing AGI would lead to too much danger.” I would argue the opposite. Keeping AGI behind a walled corporate garden could be the most dangerous situation imaginable.
There is no clear advantage to multiple corporations or nation states each with the potential to bootstrap and control AGI vs a single corporation with a monopoly. The risk comes from the unknowable ethics of the company's direction. Adding more entities to that equation only increases the number of unknown variables. There are bound to be similarities to gun-ownership or countries with nuclear arsenals in working th…
You're talking about it as if it was a weapon. An LLM is closer to an interactive book. Millennia ago humanity could only pass on information through oral traditions. Then scholars invented elaborate writing systems and information could be passed down from generation to generation, but it had to be curated and read, before that knowledge was available in the short term memory of a human. LLMs break this dependency. Now you don't need to read the book, you can just ask the book for the parts you need.
The present entirely depends on books and equivalent electronic media. The future will depend on AI. So anyone who has a monopoly is going to be able to extract massive monopoly rents from its customers and be a net negative to the society instead of the positive they were supposed to be.