He controls ChatGPT. But what controls him? Here's the video.
In Spring of 2023, I did a series about AI for Vox (touching on bad hands, special effects, training data, language, and AI art). I developed a lot of opinions while researching it and talking to computer scientists. My strongest opinion was that AI coverage was pretty bad, but my second strongest one was that OpenAI had been horrifyingly mishandled by the media.
At the same time, I felt like my history-focused look at Dr. Anthony Fauci had aged pretty well.
So though I'm a dilettante when it comes to technical matters in AI, I do think that history-focused profiles add value.
Here's a link to the reaction video (for some paid tiers).
Since AI has taken over so much mindshare, both in the media and in corporate-growth-wishcasting, it's become pretty polarized and, in my opinion, misunderstood. I think a lot of the AI hatred is due, in part, to terrible media coverage that focuses on catastrophizing. AI companies haven't helped the matter — they're notoriously bad at branding and marketing (who else names their products with so many decimal points?), and that's made people focus on AI slop as the most visible sign of AI.
I understand how "AI will replace art and human expression" is not an appealing thing. But that's really far short of the potential and uses of AI.
Check out the bad coverage by Kevin Roose. I thought his book Future Proof was interesting, so I've been really surprised by how superficial and incurious his AI coverage is. He focuses on cartoonish, jokey stuff and doesn't seem curious out the real stories in AI. His reporting is infrequent and his questions are...bad. It's a huge disservice to readers.
That said, there is good stuff out there. At this point, I'm an AI-fanboy, not an expert. But these resources have helped me shape my perspective. They're occasionally more skeptical and negative about AI than I am, but they all add value.
Understanding AI is written by Timothy B. Lee and consists of thoroughly researched and reported articles about AI. He's probably best known for his explainer of large language models, which is accessible but still deep.
AI Snake Oil is a book and Substack (I haven't gotten to read the book). It features clear-headed writing about AI while still remaining optimistic about the technology's transformative potential.
I'm sure there are many great productivity-focused, tools-obsessed AI Newsletters out there, but I've been subscribed to Ben's Bites to stay up on the news.
AI For Humans is available on all the major platforms. It's a funny, highly accessible look at AI news and AI tools, and it's honestly how I keep up on a lot of the developments in AI on a week by week basis. But that kinda shortchanges the show — it's very funny and creative and includes a lot of running gags about hot dogs and Guy Fieri.
I'm not sure about the best place to link here, but I'm a longtime listener to Robert Wright's podcast, and he's become interested in AI. His occasional interviews are probably more on the AI doomer spectrum than me (he's worried about societal disruption), but his curiosity comes first.
Cognitive Revolution is a deeper look at AI issues — it's a little in the weeds, but that's what you need.
He's caught up in the political issues of the day, to some degree, but you'd miss a lot of key figures if you didn't include Lex Fridman in this list. I think I'm in hour 3 of his 5.5 hour episode on Anthropic. Dwarkesh Patel is in a similar zone.
I'm not watching a ton of video about AI (though with the blurring of podcasts and videos lately, it's a bit of an arbitrary distinction). But I find AI Warehouse fascinating — you get to watch AIs think through a bunch of interesting experiments.
This would have been a massive link dump, so my sources are here. I don't want to imply I watched all these videos in full, though I did watch a lot. Transcript editing makes it much easier to comb through for big themes and find key quotes.
Robin M
2024-12-01 15:21:01 +0000 UTC