Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

artisana.ai

Post reply on HN
Print view View original

141–150 of 151 posts

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#141

Post by rozal »

RC_ITR wrote: post #129
rozal wrote: post #125

Earlier quoted context omitted.

I used to agree with you. I used to think the military was kinda dumb. But after doing a deeper dive into past military technology, and present - I've come to realize this is just an intelligence ruse. They made some mistakes in the 40's and 50's to where they had nuclear secrets stolen by the Russians. And ever since has been hyper compartmentalized. It would not surprise me if in the 90's or 00's they had an intern…

I mean, I think you’re again missing the point here. Jet engines can be solved with money. The actual steps of making an LLM require complex math and you can’t just pay people to make better math. And if you could, wouldn’t those people decamp for industry and become literal trillionares?

That's not how military intellectual property works. I don't doubt your technical credentials, but I think you lack deeper insight to what the military does, and has done.

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#143

Post by sph »

biohax2015 wrote: post #73
courseofaction wrote: post #24

We need new political arrangements to distribute the gains of AI or things are going to get very bad very quickly.

We are fucked. I have no hope in humanity managing this technology responsibly and no hope in my future. The months since ChatGPT's release have been some of the worst in my life, mental-health-wise.

Dude let me share my plan: go build something for yourself, a modest bootstrapped business. If you want to keep being an employee, you'll see your job and other people jobs getting more and more replaced by AIs. Bosses hyped to put AI in the coffee machine. Clients asking to add ChatGPT to their wordpress site.

Either you go build something you want, an island where code doesn't talk back, or leave tech altogether.

I am saddened that I don't even recognise this place anymore. It's not Hacker News anymore, it's AI news. It's starry eyed engineers jumping over each other ready to sell their metaphorical soul. Even I, the Luddite, can't seem to talk about anything else than this bloody thing.

My current plan is to build a small business and retire in the middle of the woods somewhere. Do some Lisp coding while the rest of the world is dancing around their new idol.

/rant, send me an email if you wanna rant about it as well, and discuss your concerns.

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#145

Post by RC_ITR »

rozal wrote: post #141
RC_ITR wrote: post #129

Earlier quoted context omitted.

I mean, I think you’re again missing the point here. Jet engines can be solved with money. The actual steps of making an LLM require complex math and you can’t just pay people to make better math. And if you could, wouldn’t those people decamp for industry and become literal trillionares?

That's not how military intellectual property works. I don't doubt your technical credentials, but I think you lack deeper insight to what the military does, and has done.

>That's not how military intellectual property works.

So your implication is that the military is full of unnamed linear algebra, systems engineering, and linguistics super geniuses and these people never leave, never talk about their work publicly, never publish anything ever, and they're all cool with their huge innovations being kept away from the public forever? All because military IP regulation?

And none of these effective state prisoners ever defect to China (where they could live like royalty) because...

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#146

Post by troops_h8r »

NumberWangMan wrote: post #97
brian_spiering wrote: post #6

Earlier quoted context omitted.

Parts of NLP have made great progress. There are still parts of NLP that could still be improved, such as the truthiness of generated answers.

Sorry to get heavy here: truth is not an NLP problem, it's an alignment problem. We want truth, but we don't have a reliable way to train an AI to provide the truth, only to provide things that are either true, or sound true enough that they fool the reward function. And even then, that may not be exactly what the AI learns to do, because of there's another level of alignment problem, the "inner alignment" or "mesa-o…

Hard agree. I'm really trying to figure out how to inject this idea into my friends' heads effectively. The main struggle I'm facing is how to convey the danger behind it. Why can it be deadly exactly? What can a program actually do to harm people, to the level where it's a risk of extinction or societal collapse?

Personally I didn't need to imagine a specific scenario to understand that there's risk, but I think it would help me convince other folks if I did.

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#147

Post by alex7734 »

troops_h8r wrote: post #146
NumberWangMan wrote: post #97

Earlier quoted context omitted.

Sorry to get heavy here: truth is not an NLP problem, it's an alignment problem. We want truth, but we don't have a reliable way to train an AI to provide the truth, only to provide things that are either true, or sound true enough that they fool the reward function. And even then, that may not be exactly what the AI learns to do, because of there's another level of alignment problem, the "inner alignment" or "mesa-o…

Hard agree. I'm really trying to figure out how to inject this idea into my friends' heads effectively. The main struggle I'm facing is how to convey the danger behind it. Why can it be deadly exactly? What can a program actually do to harm people, to the level where it's a risk of extinction or societal collapse? Personally I didn't need to imagine a specific scenario to understand that there's risk, but I think it…

> risk of [...] or societal collapse

If you want society to collapse all you need to do is succeed in having AI automate all jobs.

Every single country where money comes from somewhere other than people (oil, diamonds...) is an authoritarian nightmare simply because keeping people happy is not necessary.

Once AI can do everything and robots that can do any physical labor are developed the population will shrink dramatically as people with killer robots kill each other for resources. There is no need for AI rebellion or AI failure to get there.

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#149

Post by rozal »

RC_ITR wrote: post #145
rozal wrote: post #141

Earlier quoted context omitted.

That's not how military intellectual property works. I don't doubt your technical credentials, but I think you lack deeper insight to what the military does, and has done.

>That's not how military intellectual property works. So your implication is that the military is full of unnamed linear algebra, systems engineering, and linguistics super geniuses and these people never leave, never talk about their work publicly, never publish anything ever, and they're all cool with their huge innovations being kept away from the public forever? All because military IP regulation? And none of the…

Yes. That’s exactly how it works. Here is a brilliant one admitting to JUST that: https://youtu.be/8rHTff55fq4 (at the end)

Re: GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500k and 20k hours

#150

Post by stevenhuang »

rozal wrote: post #149
RC_ITR wrote: post #145

Earlier quoted context omitted.

>That's not how military intellectual property works. So your implication is that the military is full of unnamed linear algebra, systems engineering, and linguistics super geniuses and these people never leave, never talk about their work publicly, never publish anything ever, and they're all cool with their huge innovations being kept away from the public forever? All because military IP regulation? And none of the…

Yes. That’s exactly how it works. Here is a brilliant one admitting to JUST that: https://youtu.be/8rHTff55fq4 (at the end)

Ah the reverse engineering of crashed ET craft.

If that is your angle, I very much agree it's possible some deep black project exists that (once) looked into this. In the UFO lore, there are many stories about these advances (tr3b etc).

But all of it is orthogonal to LLMs though. Picking one exceptional area that the military is great at (aerospace) does not suddenly make the military exceptional in other areas like AI.

Post reply on HN