Weeknotes 307 - Uncanny listening to your incepted thoughts

By Iskander Smit in weeknotes — Sep 24, 2024

This week a short holiday version with event announcements and a generated podcast on predictive relations via AI. Inception.

Interpretation by Midjourney

Hi, y’all!

This is another short holiday edition. Next week, the usual setup will be back. I wanted to send the newsletter anyway for a couple of reasons. Keeping the rhythm is always a good idea—last year, I prepared some more general editions in advance; there was no time for this time.

Next, I would really like to draw your attention to a couple of activities I will do in October, and repeat the invitation to send your proposals for ThingsCon.

I am invited to host a workshop at Society 5.0 on 10 October in Amsterdam. It will be a Wijkbot workshop, and it is part of a collaboration with Digital Rights House. Later in October, we will use the results to discuss with a group of high school kids. The results will be presented at ThingsCon in December.

Check also the rest of the Society 5.0 program. It has a very nice line-up, and my experience from last year is that it is an inspiring day. The theme is “Transform the Social Fabric.” Here is a leaflet.

Later in October, during Dutch Design Week, we will organize a ThingsCon Salon in Stadslab Eindhoven as part of the special exhibition we are creating for the conference in December with 10 mockups of Generative Things from the next decade. More details follow later.

And, of course, early bird registration is open for ThingsCon. Check the current announced program here.

Triggered thought

There is no extensive triggered thought now. I noted that it looks like there is something about agents happening. See the announcements on Microsoft Co-Pilot Pages. In Dithering, a good case was made for how the agent-like behavior of the new o1 model of OpenAI is a good angle for Apple, for the value of taking time is as important for intelligence as pure compute and data power.

Small and contained AI models that can reason as agents. Is that what the real collaboration between Apple and OpenAI is about?

The hot thing was, NotebookLM introducing audio as a way to interact with your own sources. The podcast of The Verge had a long interview with Steven Johnson on the development of NotebookLM.

The model is not understanding, but the model is doing the thing that human understanding does (link to clip)

Like everyone else, I had to test the tool, and I added two articles I wrote on Predictive Relations some years ago (this one linked to a workshop and this pdf). The result is a great mirror on thinking, how it is interpreted, what makes sense, and what deserves being more precisely written. That is a great value. You can listen here.

Next week more thoughts!