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

Gorilla: Large Language Model connected with massive APIs

gorilla.cs.berkeley.edu

Post reply on HN
Print view View original

21–30 of 123 posts

Re: Gorilla: Large Language Model connected with massive APIs

#21

Post by moritzwarhier »

andrewmutz wrote: post #16
ethanbond wrote: post #14

Don't worry folks, the years and years of "of course we'd never connect experimental AI systems to the open internet" assurances were never necessary for safe development and deployment. This reassessment was founded on the last several months that showed us these systems are close to 100% dependable, they never deceive humans, they are entirely aligned with human morals (you know, all those morals we all agree on),…

Many of us have not been persuaded by the pervasive, kneejerk fear of AI that floods social media. If social media had been around in the past, we'd have never developed trains or electricity or cars because of the crippling fear that something bad might happen along the way.

I don't see cars as being comparable to the universal usage of electrical energy as an intermediate of converting physical forms of energy. Not even close.

And also, the "deployment" of cars if you will, has brought and still brings malignant effecfs on society, not only positives. If by cars you mean cars for transporting individual people in non-critical situations and the corresponding industrial and political developments, especially around urban planning.

Nor arguing about the significance of the effects, but the discovery of electricity and how to use it seem more useful to me than the discovery of using fossil fuel cumbustion for transporting people.

Road transit in total is a different thing of course.

Re: Gorilla: Large Language Model connected with massive APIs

#22

Post by sebastiansm »

behnamoh wrote: post #17
anaganisk wrote: post #13

Earlier quoted context omitted.

Why should they, and it seems very silly to flag a post for their preference of name. Are users on Twitter birds? Or is Mastodon filled with elephants? Name is merely a symbol in today's world Google is synonymous to search but Google is not just search anymore. I'm honestly curious about your gripe with it.

Some people confuse research with production. Imagine if Einstein named his theory the Theory of Chimps, just to get attention.

Imagine working in the Manhattan project outside Manhattan.

Re: Gorilla: Large Language Model connected with massive APIs

#23

Post by andrewmutz »

moritzwarhier wrote: post #21
andrewmutz wrote: post #16

Earlier quoted context omitted.

Many of us have not been persuaded by the pervasive, kneejerk fear of AI that floods social media. If social media had been around in the past, we'd have never developed trains or electricity or cars because of the crippling fear that something bad might happen along the way.

I don't see cars as being comparable to the universal usage of electrical energy as an intermediate of converting physical forms of energy. Not even close. And also, the "deployment" of cars if you will, has brought and still brings malignant effecfs on society, not only positives. If by cars you mean cars for transporting individual people in non-critical situations and the corresponding industrial and political dev…

If a technology has positives and negatives consequences for the world, but the positives outweigh the negatives, then we want those technologies to be developed. For cars, the positives outweigh the negatives.

Re: Gorilla: Large Language Model connected with massive APIs

#24

Post by sebzim4500 »

behnamoh wrote: post #17
anaganisk wrote: post #13

Earlier quoted context omitted.

Why should they, and it seems very silly to flag a post for their preference of name. Are users on Twitter birds? Or is Mastodon filled with elephants? Name is merely a symbol in today's world Google is synonymous to search but Google is not just search anymore. I'm honestly curious about your gripe with it.

Some people confuse research with production. Imagine if Einstein named his theory the Theory of Chimps, just to get attention.

There's loads of stuff in physics with silly sounding names. quarks, anyons, etc.

Re: Gorilla: Large Language Model connected with massive APIs

#25

Post by sebzim4500 »

andrewmutz wrote: post #16
ethanbond wrote: post #14

Don't worry folks, the years and years of "of course we'd never connect experimental AI systems to the open internet" assurances were never necessary for safe development and deployment. This reassessment was founded on the last several months that showed us these systems are close to 100% dependable, they never deceive humans, they are entirely aligned with human morals (you know, all those morals we all agree on),…

Many of us have not been persuaded by the pervasive, kneejerk fear of AI that floods social media. If social media had been around in the past, we'd have never developed trains or electricity or cars because of the crippling fear that something bad might happen along the way.

When cars were first entering the market they were only legally allowed to go at 4mph and a guy had to walk in front of it with a big flag.

I don't think it's the example you want to use.

Re: Gorilla: Large Language Model connected with massive APIs

#26

Post by moritzwarhier »

andrewmutz wrote: post #23
moritzwarhier wrote: post #21

Earlier quoted context omitted.

I don't see cars as being comparable to the universal usage of electrical energy as an intermediate of converting physical forms of energy. Not even close. And also, the "deployment" of cars if you will, has brought and still brings malignant effecfs on society, not only positives. If by cars you mean cars for transporting individual people in non-critical situations and the corresponding industrial and political dev…

If a technology has positives and negatives consequences for the world, but the positives outweigh the negatives, then we want those technologies to be developed. For cars, the positives outweigh the negatives.

Regarding the comparison: it's just that there are infinitely many modes of transporting people different from cars, that's why I argued against it.

I would not argue against the point that cars in themselves are very useful.

If the point however is how we are using them and how many of them we build or how we arrange our living around the necessary public infrastructure to use cars for anythint, I'm not sure if I agree.

IMO, as of now, this is a case of "market failure", if you will. But I know that the majority of people don't agree with this.

As I see it, one of the main usages of cars seems to be to use them for not having to be around car traffic: e.g. living in suburbs, "escaping" into nature on the weekends.

Re: Gorilla: Large Language Model connected with massive APIs

#27

Post by soperj »

bioemerl wrote: post #20
ethanbond wrote: post #14

Don't worry folks, the years and years of "of course we'd never connect experimental AI systems to the open internet" assurances were never necessary for safe development and deployment. This reassessment was founded on the last several months that showed us these systems are close to 100% dependable, they never deceive humans, they are entirely aligned with human morals (you know, all those morals we all agree on),…

If AI becomes harmful we will likely see it in motion and be able to respond and adapt to it. We will live in a world where the humans on the internet are far far far more dangerous than any of these machines. The harms here are relatively small. The only thing AI has is scale, but what's running at a higher scale than humanity? Every site already runs scam filters and handles misinformation. AI won't likely move the…

> If AI becomes harmful we will likely see it in motion and be able to respond and adapt to it.

What are you basing this on?

> AI won't likely move the needle much because it's going to be less powerful than humans.

Won't it just be used to augment the amount of harm that humans can do?

Re: Gorilla: Large Language Model connected with massive APIs

#28

Post by sebzim4500 »

andrewmutz wrote: post #23
moritzwarhier wrote: post #21

Earlier quoted context omitted.

I don't see cars as being comparable to the universal usage of electrical energy as an intermediate of converting physical forms of energy. Not even close. And also, the "deployment" of cars if you will, has brought and still brings malignant effecfs on society, not only positives. If by cars you mean cars for transporting individual people in non-critical situations and the corresponding industrial and political dev…

If a technology has positives and negatives consequences for the world, but the positives outweigh the negatives, then we want those technologies to be developed. For cars, the positives outweigh the negatives.

I don't think anyone (well maybe the Amish) are arguing that all technology is bad. Only that some technology might be bad, and it is worth trying to figure out the implications of releasing increasingly capable models to the public.

Re: Gorilla: Large Language Model connected with massive APIs

#29

Post by EGreg »

ethanbond wrote: post #14

Don't worry folks, the years and years of "of course we'd never connect experimental AI systems to the open internet" assurances were never necessary for safe development and deployment. This reassessment was founded on the last several months that showed us these systems are close to 100% dependable, they never deceive humans, they are entirely aligned with human morals (you know, all those morals we all agree on),…

I totally agree!

I for one welcome our AI overlords. Scan my text. I am of course in favor of the good things and good morality and let’s stay away from the bad things.

Re: Gorilla: Large Language Model connected with massive APIs

#30

Post by EGreg »

bioemerl wrote: post #20
ethanbond wrote: post #14

Don't worry folks, the years and years of "of course we'd never connect experimental AI systems to the open internet" assurances were never necessary for safe development and deployment. This reassessment was founded on the last several months that showed us these systems are close to 100% dependable, they never deceive humans, they are entirely aligned with human morals (you know, all those morals we all agree on),…

If AI becomes harmful we will likely see it in motion and be able to respond and adapt to it. We will live in a world where the humans on the internet are far far far more dangerous than any of these machines. The harms here are relatively small. The only thing AI has is scale, but what's running at a higher scale than humanity? Every site already runs scam filters and handles misinformation. AI won't likely move the…

How exactly do you propose we adapt to swarms of AI agents that quickly outnumber humans in quantity and quality of content produced?
Post reply on HN