This week
Last week I went to the University of Cologne to talk about writing, reading, and thinking with AI. I was the last talk and I was slotted at 6:30pm. For a conference that had started at 9am. God bless those who were still in the audience, and especially those whose eyes were still open. To make matters worse, we were running behind (of course). So I walked up to the front of the room just after 7. The organizers had booked dinner for all the presenters at 8pm. Everyone (myself included) was hungry and antsy.
It sucks to go last, but somebody always has to do it. Even though my talk was a bit rushed, the discussions before and after made it all worth it: human-machine collaboration, the future of education in the age of AI, what skills the next generation will need.
Back in June I wrote about the moment I started thinking more seriously about the consequences of writing with AI: a pre-ChatGPT TV episode where a line of dialogue made me think this sounds like AI wrote it months before AI could have written anything. I promised I'd share the longer argument after I gave it in Cologne. It's both the story of how imperfectly AI imitates human writing and how that imperfection is already starting to spread, even to people who don't directly interact with AI.
Human writing is still twice as unpredictable (read: interesting!) as AI-generated content. No model or prompting technique has been able to close the gap. And no matter how exponentially AI models continue to improve, I don't believe that any model ever will (more about that in today's slow note).
Next week, part 2: why I'm still heavily invested in a bright future with AI. And why I have a certain Star Trek captain from my youth to thank for this.
What I'm reading
Prof G Conversations · Ethan Mollick
Why CEOs Are Getting AI Wrong →
Wharton professor Ethan Mollick on why the people deploying AI often understand it least. The leaders building this technology are often sincerely afraid of it, and Mollick's answer is to reroute that fear towards near-term adoption so people actually thrive, instead of fixating on distant doom.
Book · Maryanne Wolf
Reader, Come Home (Harper, 2018)
One thing I'm trying to do is read more long-form content written before AI. Wolf, a neuroscientist of reading, is an excellent option: she argues that deep reading is a skill cultures build... and can lose as we outsource our reading skills to screens.
National Catholic Reporter · Justin McLellan
Pope Leo calls to "disarm" AI →
A report on the encyclical from Pope Leo XIV, entitled Magnifica Humanitas ("Magnificent Humanity"), in which he lays out the importance of preserving our humanity in the age of AI.
Slow note
Speaking of pre-AI long-form content, I'm currently reading Descartes' Error. While the book itself is completely unrelated to AI, it is opening a door in my mind that I believe will lead to more insights into the gap between human-written and AI-generated content. In the book, Professor Antonio Damasio from USC explores the connection between emotion and reason. His work has shown that the body (including feelings and emotions) and the mind (including our ability to reason) are not two opposing forces: they work together. And when one suffers, the other does as well.
Though common sense tells us that emotions poison reason, Damasio is showing us that the absence of emotion and feeling can equally poison our ability to reason. When it comes to AI, it may be the lack of the corporeal that makes the difference between artificial and human intelligence. Pope Leo points towards this in his encyclical too, writing: "So-called artificial intelligences do not undergo experiences, do not possess a body, do not feel joy or pain, do not mature through relationships and do not know from within what love, work, friendship or responsibility mean...".
Until next week,
Louise