How would you argue that ChatGPT is not a big deal?

It can answer any question but the answer won't necessarily be right.

Lazy people will think it is a simple matter of adding some little piece to it that can classify its right and wrong answers but that is a problem like trying to get a bubble out from under a rug.

The book

https://en.wikipedia.org/wiki/G%C3%B6del,_Escher,_Bach

has a set of dialogues about some people struggling furiously to build an impossible machine who take a long time to understand what they're trying to do is impossible. I've talked to people who see that book as a rebuttal to the old symbolic A.I. which was fundamentally based on logic, but the constraints that it talks about apply to any technology because they are about the very concept of logical truth, which is something that you have have to master to correctly answer a wide range of questions. Neural networks do not escape the results of Turing, Godel, Tarski and such, unless they don't promise to get the right answers.

Back when I was a Chuunibyou and really into things like Bigfoot and E.S.P. I remember reading something by Edgar Cayce about how there were two pathways to the answers you get doing channeling or pendulum dowsing (which I liked to do back then) There was one for the true answers and one for the false and you could become a good psychic if you could tell the difference. Of course you can't. Cayce was wrong about many things like much of the U.S. sinking beneath the waves, the U.S. government reinventing the death ray that sank Atlantis, etc.

GPT is not new and ChatGPT's implementation of it really just adds basic Markdown support. AI bots still suffer from the same problems as they always have, seeing as they're synthesizing text that doesn't necessarily reflect the true nature of reality. Our misconception of what ChatGPT is will be the eventual downfall of it's usability.

You can't rely on stuff it puts out. You really can't.

Unless you already have preexisting knowledge about the topic of questions, you might find there is a lot of subtle problems and inaccuracies in the confidently put-out output. Copy-pasting code will introduce bugs, using it for knowledge will leave you worse off, but the poems it can create are nice though.

I expect it may lead to deterioration of content online. We've seen this with sites scraping information from stack overflow, now they will be using these bots to create incorrect answers to most of the problems people face. Just feed into ChatGPT top 1000 searched questions from google for the query "How to .* in Python", put it on a site and watch the numbers pile up. Will most of it be correct? Maybe, maybe not, you're not checking. All this leads to is that internet will just become unusable. We've had a good run anyways.

It's just a language model. It is easily confused by things as simple as three-digit arithmetic. If your question mistakenly suggests that something false is true, it will often hallucinate impressive-sounding reasons why this is so, rather than tell you that your assumptions were wrong. It is not going to be possible to trust any answer it gives you, especially if you're not expert in the area and the answer seems correct.

> it must be replicated in the open domain as soon as possible.

I think it's GPT-3 with some training on how to follow a conversation, right? I think there are many mature open source GPT-3 competitors to choose from, many of which perform better than GPT-3, so the replication may already be close by.

I cannot.

Earlier replies to this thread note ChatGPT's lack of reliability, but fail to mention the much scarier aspect of producing unnervingly natural content especially after, not that hard to pull off if the money is there, task-specific fine-tuning.

It's not just OpenAI that has the potential of misusing this technology. I could not keep the goosebumps away through the whole "onboarding" flow to just say hi to the thing, not to mention the "safety and production best practices" for any organization interested to build on top of it.

ChatGPT is a big and scary deal. I don't think we as a species can responsibly handle the implications. To say that we are living through interesting times is an underestimation.

A human develops real intelligence over say 20 years. It doesn't spend all that time learning how to string words together. It spends the time actually understanding what things are, how they relate to each other - stuff that doesn't exist as words.

It doesn't understand what it is saying.

Ask it about something you know a lot about. The combine that with another thing you know a lot about. If you're still worried, then I guess I'm in the control group. (https://xkcd.com/2576/)

is it?