All of this is just baseless fearmongering, the output of the AI is limited, it will not even write a 2 page long instructions for the things you mentioned

It's okay. There's no need to worry. Technological progress is good and needed to move forward.

"Don't look at where we are now; look at where we will be two more papers down the line."

You have no control over where things will go, so why torture yourself over what may happen? It’s a waste of energy. You might trip, fall and hit your head and die making dinner tonight. Are you going to stay in bed because of that?

Everything is going to be okay. Technological progress, as well as progress related to human well being, happens in spurts.

Honestly if you want to build a bomb you don't have to look too far on the internet to find that information. ChatGPT is like when Siri first came out and everyone would giggle at it telling you where to bury a dead body. I've asked ChatGPT so many questions about my local area which has a fair amount of documented history online and it gave me such rubbish answers. It could very well be the basis of a tool that bad actors can manipulate but if someone wants to achieve any of the things you've stated then they aren't going to be asking ChatGPT about it

Pretty common human pattern to be scared of the unknown

Ok, but couldn’t it be able to come up with new ways to do bad things?

> Runaway self-improvement?

It doesn't do that at all.

>Strategies to acquire and deploy chemical weapons?
>
>how to make several types of bombs

That is all readily available on the regular internet, without AI.

>as it can code, probably can be used to ask for implementations of malware

No, it's really not that good. It can only make little snippets of code and they are often wrong or poorly optimized. It's very cool, but anyone with an undergrad degree in CS can do better than ChatGPT. It is not going to make malware any more approachable than it was before.

Not beyond the bad things we humans have already conjured into existence no, it's a pattern spotting algorithm not an evil genius biding its time

Here's what you don't realize: Runaway self improvement already happened and our reality is, at present, being absorbed into a superior organism.

This is the idea I keep coming back to and I want to write a fictional story about it.

Ask it to cure aging (no euthanasia).

Calm down. We will be fine king.

>This thing apparently knows everything from its vast training data

This AI tool is just predicting the next word based on your prompt and training data, it doesn't actually know anything in the way you mean it. The AI architectures you need to worry about does not currently exist. Existing architectures improved to 99.99% accuracy will not turn them into the AIs you need to worry about.

This, OP needs to realize everyone was afraid of the railroad industry as well.

You better get on it. In like 6 months (give or take) AI like this will be able to write entire coherent stories.

You’re probably right, all these Twitter posts with conversations just look so… powerful. It’s like it knows what it is referring about. And errors seem to occur largely in the details only.

But you can find everything on the internet, You don't need chatgpt...

Chatgpt is not scary at all, But Gpt6 or gpt7 on the other hand, Probably will be...

:D

Damn, I can't wait xD

you can't change it, why worry?

you can change it, why worry?

chatgpt isn't even connected to the internet, and can't even "do" anything. you are overexaggerating.

It also doesn't give proper instructions on how to do anything you mentioned.

It can be 'tricked' into giving instructions for things it shouldn't, like hotwiring a car, but if you read that poem you would see it's absolute trash and you can't conceivably hotwire a car with those instructions.

you should probably talk to a mental health professional before you go full conspiracy theorist

Unlike a conspiracy theorist, nothing would make me happier than be proven wrong