Live data from Hacker News

GPT-4

openai.com

501–510 of 1001 posts

Re: GPT-4

#501

Genuinely surprised by the positive reaction about how exciting this all is. You ever had to phone a large business to try and sort something out, like maybe a banking error, and been stuck going through some nonsense voice recognition menu tree that doesn't work? Well imagine chat GPT with a real time voice and maybe a fake, photorealistic 3D avatar and having to speak to that anytime you want to speak to a doctor,…

We're going to see resurrection of p2p service, what's not to like?

Re: GPT-4

#502

Genuinely surprised by the positive reaction about how exciting this all is. You ever had to phone a large business to try and sort something out, like maybe a banking error, and been stuck going through some nonsense voice recognition menu tree that doesn't work? Well imagine chat GPT with a real time voice and maybe a fake, photorealistic 3D avatar and having to speak to that anytime you want to speak to a doctor,…

Agreed. AI systems should be required to identify as such when interacting with a human or we are going quickly to a strange place. Like you get a warning when your conversation is being recorded. Write your representatives today.

Re: GPT-4

#503

Genuinely surprised by the positive reaction about how exciting this all is. You ever had to phone a large business to try and sort something out, like maybe a banking error, and been stuck going through some nonsense voice recognition menu tree that doesn't work? Well imagine chat GPT with a real time voice and maybe a fake, photorealistic 3D avatar and having to speak to that anytime you want to speak to a doctor,…

I think the dread you may be feeling is "facts without agency" which is to say that a system which can answer any question on a topic but doesn't have the agency to understand can be really bad. The whole "best way to hide a body" stuff when Siri was released, now backed up by facts is what? possible? The example (no I don't know how real it was) of an alleged 13 year old girl asking how to make sex with a 31 year old male she met on the internet "special" is the kind of thing where a human in the loop starts with "Wait, this is the wrong question." Similarly with questions about how to successfully crime.

Having run a search engine for a bit it quickly became clear how criminals use search engines (mostly to search out unpatched web sites with shopping carts or wordpress blogs they could exploit at the time). I don't doubt that many malicious actors are exploring ways to use this technology to further their aims. Because the system doesn't "understand" it cannot (or at least has not been shown to) detect problems and bad actors.

FWIW, the first application I thought of for this tech is what the parent comment fears, basically having people who can follow a script running a "Front end" that presents to an end user a person who looks familiar and speaks their language in a similar accent (so accent free as far as the caller is concerned) about a topic such as support or sales. Off shore call centers become even more cost effective with on-the-fly translation because you don't even need native language speakers. That isn't a "bad thing" in that there is nominally a human in the loop but their interests are not aligned with the callers (minimize phone time, costs, boost satisfaction).

And of course the whole "you trained it on what?" question where you wonder just what is used as source material and without knowing that what sort of trust can you put in the answer?

Re: GPT-4

#504

Genuinely surprised by the positive reaction about how exciting this all is. You ever had to phone a large business to try and sort something out, like maybe a banking error, and been stuck going through some nonsense voice recognition menu tree that doesn't work? Well imagine chat GPT with a real time voice and maybe a fake, photorealistic 3D avatar and having to speak to that anytime you want to speak to a doctor,…

I can't articulate this well for now, but in all of the mayhem you asked to imagine, I must say I also see a possibility to freedom. Freedom from news because they are all garbage anyway, already now. Freedom from stupid mortgage application processes because you just buy/rent what you can afford. And so on. Of course, it is likely most people do not choose this freedom - maybe not even me - but it is there.

Re: GPT-4

#505
post #218

A class of problem that GPT-4 appears to still really struggle with is variants of common puzzles. For example: >Suppose I have a cabbage, a goat and a lion, and I need to get them across a river. I have a boat that can only carry myself and a single other item. I am not allowed to leave the cabbage and lion alone together, and I am not allowed to leave the lion and goat alone together. How can I safely get all three…

> It's not clear to me if the lesson here is that GPT's reasoning capabilities are being masked by an incorrect prior (having memorized the standard version of this puzzle) or if the lesson is that GPT'S reasoning capabilities are always a bit of smoke and mirrors that passes off memorization for logic. It's a lot closer to the latter. GPT doesn't have "reasoning capabilities", any more than any other computer progra…

When a parrot says something, I ask it "what does that mean" and it is stumped. When a human or GPT says something, I ask them "what does it mean" and they both reply in a way that shows they understand.

In order to show lack of understanding you need to provide a clear test that ChatGPT would fail but a human would pass.

Re: GPT-4

#508

Genuinely surprised by the positive reaction about how exciting this all is. You ever had to phone a large business to try and sort something out, like maybe a banking error, and been stuck going through some nonsense voice recognition menu tree that doesn't work? Well imagine chat GPT with a real time voice and maybe a fake, photorealistic 3D avatar and having to speak to that anytime you want to speak to a doctor,…

I mentioned the same thing to my wife. Today, if you get stuck in some corner case of software, you can eventually reach a human who will observe the buggy behavior and get you unstuck. With this stuff… may we all never get caught in a buggy corner or edge case…

Re: GPT-4

#509

I'll be finishing my interventional radiology fellowship this year. I remember in 2016 when Geoffrey Hinton said, "We should stop training radiologists now," the radiology community was aghast and in-denial. My undergrad and masters were in computer science, and I felt, "yes, that's about right." If you were starting a diagnostic radiology residency, including intern year and fellowship, you'd just be finishing now.…

What is your take then on how this affect your field? And your occupation? Do you think you will incorporate such technology into your day-to-day?

Re: GPT-4

#510
post #218

A class of problem that GPT-4 appears to still really struggle with is variants of common puzzles. For example: >Suppose I have a cabbage, a goat and a lion, and I need to get them across a river. I have a boat that can only carry myself and a single other item. I am not allowed to leave the cabbage and lion alone together, and I am not allowed to leave the lion and goat alone together. How can I safely get all three…

Honest question: why would you bother expecting it to solve puzzles? It's not a use case for GPT.

Designing a new product can often be thought of like one of these puzzles. E.g. designing an "integrated" traffic system for a city that balances use by pedestrians, cyclists, cars, and trolleys.
Post reply on HN