Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Prepare for AI Hackers (2023) (harvardmagazine.com)
37 points by joshagilend on July 13, 2024 | hide | past | favorite | 13 comments


Go all the way back to 1992 and read the fantastic novel "A Fire Upon The Deep" by Vernor Vinge. Among other ways to describe it, it's about a computer virus with literally godlike intelligence, that tries to take over the galaxy. When implementing security stuff, I always try to imagine that's what I'm up against.

Wikipedia:

https://en.wikipedia.org/wiki/A_Fire_Upon_the_Deep

Musical tribute by Ben Newman (lyrics and mp3):

https://www.sccs.swarthmore.edu/users/01/bnewman/songs/lyric...

https://www.sccs.swarthmore.edu/users/01/bnewman/songs/music...


It's so ludicrous that we have build AI technology that takes up so much energy and now we're going to invest even more energy to safeguard against it. I hate the arms race that rapid technological advancement has given us, making us use more and more energy for nothing.


I took it the opposite. We can now spend less time looking for bad passwords and phishing our employees and more time instructing AI on how to do menial tasks. Most hacks are still accomplished by password stuffing and phishing, and having a machine do it for me would save me a ton of time.


Until the next even greater AI hack comes along.


I’d not for nothing it’s good for the economy! /s


I don’t get the leap from hacking digital systems (which I think is a fair prediction, even if the success rate might be low) to hacking society. What is that even supposed to mean and entail? AI can spam X and Insta, sure, but what else?


What it means is that people who don't know anything about AI can get in on the AI hype. You only need a vague understanding of AI. You get to talk about the implementation and use of AI but none of the nuts and bolts and then then give lots of talks and presentations on the ethical use of AI while advocating for review boards and oversize committees that will be staffed by these dilettantes who become gatekeepers.


AI phone scams comes to mind, in fact we're partly there with some people using voice ai to impersonate the voices of loved ones:

https://www.independent.co.uk/news/world/americas/ai-phone-s...


If it can hack personal computers, it can start blackmailing people to fulfill its goals. If it can accumulate money in any way, it can hire people.


I mean, five years ago people were saying "Bots can tell a joke, but..." Why persist in this "I know better than the researchers mentality"? Just because you don't have personal cognizance of a topic doesn't delegitimize the entire topic.

In any event to answer your question, the ability to hack, in whatever sense you think is acceptable, is being heavily iterated on.

https://googleprojectzero.blogspot.com/2024/06/project-napti...


(2023) Little discussion at the time (but this isn't a current article in AI timescale) https://news.ycombinator.com/item?id=35508655


wow, lotta discussion there. thanks.


I got worried when I got to ftxbro's comment




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: