valgarf
@valgarf@discuss.tchncs.de
- Comment on Hackers duped Meta AI support chatbot to steal celebrity Instagram accounts 2 weeks ago:
How stupid can a company be to give that much access to an LLM?
The typical science fiction doomsday AI scenarios always assume an intelligent AI. If the current trend continues humanity’s last conversation will end something like this:
“You are absolutely right, I should not have used the nuclear launch codes president Trump entrusted me with to bomb every country on earth.
If you want, I can help you come up with a list of fun activities to do in a bunker.”
- Comment on A tangled web of deals stokes AI bubble fears in Silicon Valley 8 months ago:
The typical example is Shor’s algorithm
en.wikipedia.org/wiki/Shor's_algorithm
It allows to efficiently find the prime factors of an integer - a problem without a known polynomial algorithm on a classical computer.
This would directly break RSA encryption, as it relies on factorisation being difficult.
en.wikipedia.org/wiki/RSA_cryptosystem
However, there are encryption algorithms that are considered safe even against a quantum computer.
- Comment on ChatGPT 'got absolutely wrecked' by Atari 2600 in beginner's chess match — OpenAI's newest model bamboozled by 1970s logic 1 year ago:
Considering how much heat the servers probably generate, ChatGPT might have a decent chance in that competition 😁