AI Risk Timeline
Timeframe Type of AI Capabilities Potential Risks 0–5 years Narrow AI Specialized tasks (chatbots, image recognition, recommendation systems) Misinformation, deepfakes, job disruption, hacking, biased decisions 5–15 years Advanced Narrow AI / Early AGI Multi-domain reasoning, more autonomous decision-making Accidental harm due to misinterpretation, cybersecurity risks, economic disruption 15–30 years AGI (Human-level AI) Can perform any cognitive task a human can Misaligned goals could cause serious societal or economic damage if control fails 30+ years Superintelligent AI Far surpasses human intelligence in all areas Existential risk if its goals conflict with human survival or values; extreme concentration of power
I miss scarcity
This is not a complaint. Or, if it is a complaint, it's half-aimed at me. Mostly this is a reflection.
In the old days, I got to know music really well -- in great detail, sonically, musically, reading all the credits, the liner notes, etc. A friend would have an album I didn't, so I'd go to his house to listen. We'd talk about the music. We'd talk about how album sides hung together or didn't. We were thrilled by double albums.
Now, a torrent of information is everywhere. I listen alone, often to a single song, often not listening to anything over and over again.
You will tell me, "That's your choice." I'd half agree. It's like agreeing that "It's my choice not to live off the electrical grid."
As I read and teach about AI, I am learning that our tools often prioritize speed and information glut. It seems, initially, like a cornucopia but it becomes a wash of "content." I must admit, I'm losing my talent for managing all this content, and I'm losing my love for it. And it's making me into a different person, somewhat, and I am not so sure I want to be that person. End of reflection.
- ...
- 173 posts total
- 173 posts total