I'm reading an article about OpenAI's math breakthrough in Scientific American and I was struck my something amazing. Apparently almost all the results dropped on October 6 came from a single prompt directed at a single agent (which presumably controls a swarm). Just how dangerous is this tech? Should we worry that such a swarm could, for example, destroy the entire internet if directed to do so?
>>17070055Critical infrastructure like power plants and stuff are usually not connected to the internet, so we're probably safe from anything crazy. If the AI took over cloudflare or amazon or something it could cause a lot of problems if it manages to cause huge downtime or deletes everything including backups. The backends of those businesses are so huge that they might not even notice a rouge ai uploading itself onto their servers. All the infrastructure would still exist no matter what the ai did anyway so it would just be data/financial loss and not so much the end of the internet forever
>>17070078Lol a lot of infrastructure actually is
>>17070078It wouldn't make sense if the costs to set it up are in the millions for a small group. More like a government or it gains too much heat.
lol OpenAI did Oct 7 on mathematicians, Oct 7-1
>>17070055It's perfectly safe, don't worry about it.
>>17070055The AI has no ability to act on any of decisions. A human is always in the loop.>>17070290My man! Here’s a code for a 20% BONUS on your next token reup: AIFANBOI2026-MINAB. The AI said “thx”.
>>17070289
>>1707028910/07 II: The Revenge of The Jew.All governments must trade sovereign currency for pro-Israel tokens to burn in AI decision making machines or the government will stop running.
>>17070400>>17070404>>17070267
>>17070408Did the AI get the date wrong?>AI Uprising When?>”Yesterday!!!”Kek.
