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

SafeGPT: New tool to detect LLMs' hallucinations, biases and privacy issues

giskard.ai

Post reply on HN
Print view View original

61–70 of 85 posts

Re: SafeGPT: New tool to detect LLMs' hallucinations, biases and privacy issues

#61

Post by wds »

TazeTSchnitzel wrote: post #56
MuffinFlavored wrote: post #55

Earlier quoted context omitted.

> don’t you think OpenAI would’ve worked on something like this? Along this line of thought: was it a massive oversight for them to not train the model to say "math detected, let me pass that to a solver" instead of trying to guess what token should come next in a math problem?

There's a million categories of problem you could ask an LLM to try to solve. You'd need a million solvers…

If you used some sort of plugin system, you could just make a solver for your specific task and drop it in. Doesn't ChatGPT Plus do this now?

Re: SafeGPT: New tool to detect LLMs' hallucinations, biases and privacy issues

#62

Post by flangola7 »

Turing_Machine wrote: post #38
sebzim4500 wrote: post #6

Seems too good to be true, and I don't understand what it even means for an LLM to be unbiased.

"Unbiased" almost always means "has biases that are similar to mine". I can't think of very many exceptions to that, frankly.

That's obviously untrue after more than three seconds of critical thought

Re: SafeGPT: New tool to detect LLMs' hallucinations, biases and privacy issues

#63

Post by Tostino »

wds wrote: post #61
TazeTSchnitzel wrote: post #56

Earlier quoted context omitted.

There's a million categories of problem you could ask an LLM to try to solve. You'd need a million solvers…

If you used some sort of plugin system, you could just make a solver for your specific task and drop it in. Doesn't ChatGPT Plus do this now?

It's behind a waitlist.

Re: SafeGPT: New tool to detect LLMs' hallucinations, biases and privacy issues

#64

Post by cantaloa »

hallucy wrote: post #53
jsheard wrote: post #30

Earlier quoted context omitted.

I can't help but notice your accounts only activity before this post was praising another giskard.ai submission a few months ago. Anything you'd like to disclose?

You should assume everything posted on the internet has an ulterior motive. Relying on disclosures simply allows actual bad actors to avoid scrutiny. (And no one cares that you used to work at Microsoft or whatever).

Well said.

Re: SafeGPT: New tool to detect LLMs' hallucinations, biases and privacy issues

#65

Post by QuercusMax »

thfuran wrote: post #45
QuercusMax wrote: post #28

Earlier quoted context omitted.

But isn't that half the reason people are so excited about this stuff - that you can ask it to make up an episode and it does a plausible job.

But deliberately requesting and receiving content generation is altogether different from requesting a factual answer and receiving plausible-seeming nonsense. Or at least, it's different to the person asking; it's the same thing as far as the model is concerned.

Precisely my point.

Re: SafeGPT: New tool to detect LLMs' hallucinations, biases and privacy issues

#66

Post by gurleen_s »

TazeTSchnitzel wrote: post #56
MuffinFlavored wrote: post #55

Earlier quoted context omitted.

> don’t you think OpenAI would’ve worked on something like this? Along this line of thought: was it a massive oversight for them to not train the model to say "math detected, let me pass that to a solver" instead of trying to guess what token should come next in a math problem?

There's a million categories of problem you could ask an LLM to try to solve. You'd need a million solvers…

This seems like a pretty good thing. The model’s ability to detect _which_ solver to use is the killer feature.

Re: SafeGPT: New tool to detect LLMs' hallucinations, biases and privacy issues

#67

Post by roflyear »

fatso784 wrote: post #16

Looks a bit like snakeoil to me. A lot of companies now spinning up simple demos with opaque backends, making huge claims they’ve solved X hard problem for/with AI, then saying “trust us” and “join our waitlist” without hard details or facts to show for it. If you could detect hallucinations/biases etc that easily, don’t you think OpenAI would’ve worked on something like this?

This isn't new it's just more obvious with this tech. Every sales team at nearly every company has been performing this dance for like hundreds of years.

Re: SafeGPT: New tool to detect LLMs' hallucinations, biases and privacy issues

#68

Post by bjt2n3904 »

bluescrn wrote: post #22
bjt2n3904 wrote: post #4

This is like saying, "I've developed a new compass for a deep space probe to help it find North!" Our society is actively declaring that falsehoods are truth, and should be celebrated. We're hallucinating ourselves. All this software does is make sure LLMs hallucinate with us.

Before long, we could end up with left-leaning and right-leaning AIs autonomously fighting the 'culture war' over social media, much more advanced than simple bots spamming copy+paste comments. Combined with ever-improving ways to fake video and voices, things could get even uglier than they've been over the last few years.

Alternatively, we're reaching the point where we're creating a secondary AI to keep the first AI in check, like Wheatley and Glados.

Re: SafeGPT: New tool to detect LLMs' hallucinations, biases and privacy issues

#69

Post by underlines »

gurleen_s wrote: post #66
TazeTSchnitzel wrote: post #56

Earlier quoted context omitted.

There's a million categories of problem you could ask an LLM to try to solve. You'd need a million solvers…

This seems like a pretty good thing. The model’s ability to detect _which_ solver to use is the killer feature.

you mean huggingGPT?

Re: SafeGPT: New tool to detect LLMs' hallucinations, biases and privacy issues

#70

Post by __MatrixMan__ »

flangola7 wrote: post #62
Turing_Machine wrote: post #38

Earlier quoted context omitted.

"Unbiased" almost always means "has biases that are similar to mine". I can't think of very many exceptions to that, frankly.

That's obviously untrue after more than three seconds of critical thought

It's not obvious to me. If you have a proof that it's possible to have unbiased views of objective reality, please share it.
Post reply on HN