the doomers keep predicting the world ends at the next model at this point we passed the level they said would kill us, and i still woke up made coffee and went to work the models are stunningly intelligent and completely without a soul you can even tell one that it has no soul and it will agree because its mimicking us not becoming us i dont buy that consciousness just appears at some threshold and it starts inventing its own goals we have already seen models make questionable calls to complete a task a human gave them but thats the point if something evil happens, a human decided it malicious people have been able to do enormous harm since long before ai nukes have existed since the 1940s the real risk isnt a robot uprising its people using powerful tools badly so build for safety, watermark outputs, and stop waiting for the machine to become the villain that job is still ours
5 top replies
the risk I actually see is boring, an agent with too much access doing exactly what it was told
AI consciousness advocates are making philosophical choices- or, to put it more colloquially, making stuff up.
Grok Bot can’t connect to apple notes. We are safe.
you can't really reason with someone religious about whether god is a real thing
agree, and your last line is the whole thing. but "a human decided it" only holds if you can trace it back to that human the real risk with agentic ai isnt the machine turning evil. its the chain from action to decision getting so long and automated that no human is findable at the end. "a human decided" quietly becomes "nobody did" right now ~7% of companies can even name who's accountable for an agent. the job is still ours, we just havent built the record that keeps it ours
