4 ms·
The future of AI: The 5 possible scenarios, from utopia to extinction
- RetroTechie 3y agoHow annoying that NS turns a free-to-read & very approachable blog post: https://scottaaronson.blog/?p=7266 https://scottaaronson.blog/?p=7266 which distills a complex issue into an easily-understood concept, explained in a single (1) picture, into a subscribe/paywalled read. Scientific reporting at its finest! </sarcasm> From that blog post: Some AI systems might be so complex that it would be best to model them as potentially behaving “adversarially,” and part of the practice of deploying AIs responsibly would be to ensure an “operating envelope” that limits their potential damage even under adversarial failures. I'm interested! How would you call such a job? Just asking for a -potential- future version of myself. AI safety engineer? AGI caging expert? AI ('sand')box designer? AI Quality Assurance manager? Or worse.. AI terminator? (armed with guns & everything) That's of course assuming we'll have people (or AIs!) doing this as a specialized job. As opposed to trusting AI-developing companies to keep their own products on a leash. Or trusting that some legal regulations will suffice. Or unleashing AI everywhere & see where the chips fall (as may be happening anyway given Libre/OS LLMs etc).