Wollongong Sci Fight: We should fear AI

This round, we debate: should we fear AI? Just this year, AI chatbots have given people instructions on how to build a bomb, told users that they loved them, and instructed one user to leave their spouse and a child to stick nails into a live power socket. Before now you had to go to 4chan for that quality of abuse. Has Silicon Valley simply built a better troll? Programmers are running behind ChatGPT and other AI trying to fix the problems as they arise and yet for every bit of foolproofing they achieve, humanity builds a better fool. Our one saving grace right now? It’s bad at Wordle. But this will surely not be the case for long. There’s a growing fear about our inability to predict what AI might do - could it build a separate AI that doesn't abide by the programmed controls? Could it create DNA for entirely new…