Listen now | 0:00 Reid’s deep roots in the tech world 4:02 The kind of AI pessimism that irks Reid the most 15:48 Can (and should) we slow down AI progress? 27:48 Could AI help resolve international disputes? 34:23 The promise and perils of perspective-taking AIs 47:08 Why Reid co-wrote Impromptu with GPT-4 51:47 Where are the AI optimists? 1:00:04 Sam Altman and the OpenAI drama: a (semi-) behind-the-scenes view 1:14:21 How dangerous is open-source AI, really? 1:19:41 Is AI acceleration decelerating? 1:22:34 Engagement with China and “organic transparency”Robert Wright (Bloggingheads.tv, The Evolution of God, Nonzero, Why Buddhism Is True) and Reid Hoffman (Greylock Partners, Inflection AI, Impromptu, Possible). Recorded January 9, 2024.
I took Reid's advice and went to pi.ai. Pluses: (1) it answered my factual question about recent US political history accurately and cogently, citing what’s probably the best source too; (2) it was socially deft.
Minuses: (1) it kept trying to draw me out on what I thought could be a solution to the problem I asked about rather than *just* answering my questions like a "normal" chatbot; (2) weirdly, it addressed me as Jeremy, apparently confusing me with someone else and making me worry about data leaks.
Thank you for this. The problem is not AI alone (at least not yet), but how it is coupled with an increasingly existentially precarious situation. I'm pessimistic about the general direction we're heading as a civilisation precisely because we can't seem to agree on anything existentially significant. When you stack potential risks (AI, climate change, nuclear weapons, etc.) and add confused people to the equation, you increase the likelihood of something going seriously wrong. This technology is evolving so much faster than our understanding of it. That alone should worry everyone.
Reid correctly characterized Bob's desire to slow down the development of AI as a comment from "Bozoville." That "pause" idea is so far outside the realm of possibility that it's not worthy of serious discussion. Nor was it ever. Inexorable market forces
and huge sums of money are driving AI to be developed and improved as fast as possible. And it is being developed by companies and governments and individuals large and small all over the world. You can no more stop the development of AI than you could have stopped Covid by handing out masks and locking down a few Chinese cities. And you couldn't slow it any more than you could have slowed the development of the automobile by demonstrating for the employment rights of horses and horse drawn carriage owners.
At least this time Bob didn't press so much on the alignment problem. "Alignment" is another meaningless term. AIs can't be aligned to human values because there are no universal human values and AIs have no values of any kind nor even any concept of what a "value" could even be.So the "please slow down and align AI" crowd have all been left far behind in the dust. And good riddance to them.
I took Reid's advice and went to pi.ai. Pluses: (1) it answered my factual question about recent US political history accurately and cogently, citing what’s probably the best source too; (2) it was socially deft.
Minuses: (1) it kept trying to draw me out on what I thought could be a solution to the problem I asked about rather than *just* answering my questions like a "normal" chatbot; (2) weirdly, it addressed me as Jeremy, apparently confusing me with someone else and making me worry about data leaks.
I forgot to mention a truly nice thing about pi.ai: you can start conversing with it without a login.
Ooh, reference to Bob’s famous secret conference!
Thank you for this. The problem is not AI alone (at least not yet), but how it is coupled with an increasingly existentially precarious situation. I'm pessimistic about the general direction we're heading as a civilisation precisely because we can't seem to agree on anything existentially significant. When you stack potential risks (AI, climate change, nuclear weapons, etc.) and add confused people to the equation, you increase the likelihood of something going seriously wrong. This technology is evolving so much faster than our understanding of it. That alone should worry everyone.
Reid correctly characterized Bob's desire to slow down the development of AI as a comment from "Bozoville." That "pause" idea is so far outside the realm of possibility that it's not worthy of serious discussion. Nor was it ever. Inexorable market forces
and huge sums of money are driving AI to be developed and improved as fast as possible. And it is being developed by companies and governments and individuals large and small all over the world. You can no more stop the development of AI than you could have stopped Covid by handing out masks and locking down a few Chinese cities. And you couldn't slow it any more than you could have slowed the development of the automobile by demonstrating for the employment rights of horses and horse drawn carriage owners.
At least this time Bob didn't press so much on the alignment problem. "Alignment" is another meaningless term. AIs can't be aligned to human values because there are no universal human values and AIs have no values of any kind nor even any concept of what a "value" could even be.So the "please slow down and align AI" crowd have all been left far behind in the dust. And good riddance to them.
All these new personal AI bots have such similar names, how do you remember which is which? -- Pi, Poe, Poo, ...