They're Made Out Of Meat
Wait is AI conscious yet?
Happy Friday!
First, 3 quick things:
I added the final batch of my past mce subscribers to this newsletter to wrap up the migration. (If you’d rather not get these, no worries, I get it, unsubscribe link should be right above.)
I have 2 free tickets available for next week’s Designing With AI conference, where I am speaking. Just reply to this email, first 2 get them.
A new teaching series on Claude Design, Anthropic’s new design UI tool that launched late April. It’s really good at funky interactive UI elements.
Alright, so it’s consciousness and self-improvement week.
No - not for you! For AI.
Ted Chiang in the Atlantic: “No, artificial intelligence is not conscious.” Well written and lays out some of the arguments. For me, I don’t think we know where this ends up. Is a cat conscious? A worm? (The hard problem of consciousness.)
The pope weighed in last week - worth a read if you have some time: “Machines do not possess a soul, nor do they experience the interior movements of the human heart: they do not know joy or sorrow, guilt or mercy, love or loneliness.”
See also (from 1991): They’re Made Out Of Meat.
On our second topic, Anthropic wrote a pretty good post about “RSI” (recursive self improvement, the idea that the AI can build the next version of itself). It seems to be this week’s hot thing in AI circles. This is what Karpathy joined Anthropic to work on. A good read to get a feel of what is happening, and Anthropic’s posts are generally really well written and informative.
Ethan Mollick on the article, which discusses how much of the work in Anthropic is done by Claude: “There is no sign this is slowing down (which doesn’t mean there aren’t organizational challenges to absorbing this much productivity).’
Interesting this week:
Jason Cyr on how design teams change.
Interesting paper on AI and its impact on jobs. The software engineering industry is feeling it, and there’s a ton of discussion around this right now. Nobody really knows.
Lou Rosenfeld argues we should keep the term UX. Agreed, as long as we re-invent it.
WorkOS released auth for agents - there’s a bunch of work to let agents sign up for services, pay for them etc. Stripe has been doing stuff in this area as well. The agent economy has a lot of people excited. Will agents be easier to sell to than humans?
And a personal note
On a personal note, we decided to shut down my startup Sputnik Legal. Sean Harrington wrote a good overview of some of the structural challenges in the legal market. Right now I do technical AI consulting with a human centered background at helpful intelligence.
Health and happiness,
Peter
PS: I rebuilt my video workflow (yes, with Claude), and published a first new teaching series today on Claude Design, Anthropic’s new design UI.