The prose on the write side of critical thinking.

This little piece from the publication, one useful thing, is particularly pertinent considering the recent volley of research around the topic of the impact of AI on cognition. I have also been forming my own view on this so check out 'The Human Side of AI' section below

So Ethan Mollick, one of the more thoughtful AI researchers writing publicly, has a confession: he can spot AI-written prose from a distance, and when his internal detector fires, he stops reading. Doesn't matter how interesting the topic is. The writing is fine, and he's out.

His argument isn't that AI writing is bad really, it's that defaulting to it costs you something real. The specific thing you give up isn't quality, it's the thinking that happens when you sit with an idea long enough to have an actual position on it. Writing-as-process, rather than writing-as-output. The two are not the same thing, and one of them, it turns out, is where a significant portion of human insight actually lives.

What makes this land is that Mollick is not a luddite making a pastoral argument about pen and paper. He's someone who uses AI constantly and has thought carefully about the difference between using it well and using it as a reflex. That distinction is doing more work than it looks like

Read the original ↗

The Human Side of AI

I've been trying to articulate this for a while - hopefully I've got some way there! This piece is about a subtle little weakness AI may be creating. This not stupidity per se but a lower tolerance for uncertainty. The small awkward moments in a task - the half-formed sentence, the unclear judgement, the uncomfortable pause before we know what we mean - are easy to treat as waste. But some of that friction was doing useful work.

The argument is not to keep things hard for the sake of it. It is to preserve the ability to tell the difference between friction that wastes people and friction that forms them. The difficulty is that this may require AI to do something it has no natural incentive to do: notice the moment of dependency, interrupt the rescue, and leave us alone with the problem for a little longer.

Read the full essay →

Have AI Got News For You!

• The best argument I've heard for why AI won't take your job. Amazon hires same intern cohorts despite AI surge, empirical pushback to automation anxiety. Read ↗
• An Interview with Eric Seufert About Models and Ads, and AI's Upside for Humanity. Ad-tech insider examines how AI models reshape targeting and consumer welfare, beyond hype. Read ↗
• Anthropic nears $1 trillion valuation as it overtakes OpenAI. Claude maker's $65bn funding round signals investor shift away from OpenAI's dominance. Read ↗
• Reckoning with the future; and a singularity story - Import AI 458. Sobering case for active engagement with AI's trajectory rather than passive withdrawal. Read ↗
• AI just gave The Boys the ending fans actually wanted ... - Instagram. Fan-generated AI video rewrites unpopular finale, testing copyright limits and creative taste. Read ↗