Weeknotes 312 - prevent curling agentic AI parenting

By Iskander Smit in weeknotes — Oct 29, 2024

Agentic AI and new ways of designing the right balance of magic and agency. And other news on immersive human-AI partnerships in tech societies.

Hi, y’all!

Last week was almost fully dedicated to Dutch Design Week (six trends as seen by Dezeen). On Tuesday, I had a session on the the projects of the Expertisenetwork Systemic Co-Design, where I shared and reflected on the progress of the WijkbotKit as a kit for robotic perspectives. Last Sunday, an interview Tomasz and I had with Rijnmond came online, which is both a good representation of our goals (apart maybe from the heading that is more representing the question than our narrative :)

On Wednesday, we had a very good ThingsCon Salon to kick off the project for an “Generative Things” exhibition at the December TH/NGS conference. Sen did a lovely futuring workshop, and we had a nice turnout. Thanks to Stadslab Eindhoven for the kind hosting.

The Design & AI Symposium inspired me (see Triggered Thought), and I visited more of the exhibitions on Friday. As always, I had too little time to see everything.

The coming week will be exciting. Not in a good way, as we will be the last week before the US elections, and there is still a significant chance that Trump will win. We will have all analyses next week on how much AI was part of it or not. Less than expected, etc. And maybe we will only learn that retrospectively. It is part of society for sure and will be in the coming period even more. How it will influence society will depend on who is in office.

Who would have thought that, in the end, it is Elon Musk who dominates the elections and maybe the outcome? It feels like a superhero movie like Spiderman, where the fight is against that one supervillain ruling everyone from a capitalistic agenda. The agenda is power and money, not any usual political or societal ideology (check also the Russian connection). Just like in superhero movies, entertainment value defines success. So the big question is if there is a superhero in this movie…? And I just see this nice post is stated kind of the same feeling.

Triggered thought

Two keynote speakers opened the Design & AI Symposium on Wednesday morning, organized last week during Dutch Design Week by TU/e and TU Delft. Tobias Revell did a lovely talk about AI and magic entanglement. To summarize, one of the messages is that AI is not magic, and claims of scientific underpinnings are more marketing than based on real research. Companies are suggesting research to sell AI as magic. In his view, money and power are dominating AI. There are not really artistic goals.

After him, Bas van de Poel presented his work at Modem. He is a great designer who has found ways to explore futures by inviting collaborating designers to publish research reports on certain topics. The output is often also a design to underpin visions.

From personal encounters, I can say I respect both of their work a lot. it might seem to represent two visions, but it is also one story with two branches. One of Modem's works made me think about how AI also drives a new aesthetic in product design. The Terra project is a speculative device that helps you navigate a city without following a specific route. It lets you wander a city, maybe even get lost, with the security that you will reach your end goal. This sentence already has an interesting concept: Are you free to wander if getting lost is not possible? But that is something to explore another time, maybe.

Terra is not a new idea. It is probably for Modem and their partner Panter&Tourron, as the earlier example was made by a startup in 2015 that failed to execute successfully. The company created a smart Bluetooth device called Wander that was doing the same: a device that looked like a watch, only helping you with a pointer to the end goal, not offering any routes. It was up to you to decide how much you would like to wander in that direction. The device was too complex to make, and the app that was the pivot was not exciting enough, so the startup adventure ended in 2019.

Terra is not just a new iteration; it is building a new product based on the current state of our relationship with technology. It does not leave it all up to you to decide; it offers opportunities to use the capabilities of generative AI to form a partnership in getting lost and wandering without panicking. It is, in that sense, representing the often-mentioned co-performance with AI.

But it is also clear that this relationship is very layered. Is it a form of curling parenting, taking away all frictions? It is tempting to connect it to the focus on Agentic AI. What can we learn from these types of first explorations to design these new agents?

In 2019, when I was diving deep into predictive relations with things that predict (and the potential alienating consequences), I made a first draft of a design approach. See also this short paper.

Deconstruct what a smart product makes smart (or intelligent)
Think about the impact of adding predictive intelligence to our interactions with that thing. In current times you would say, what would be the impact of a thing getting generative capabilities
Design engaging interactions between human and AI based on respectful co-performance.

What I like to focus on and -as note to myself- would be good to revisit, is the first rough design approach. For me, the last part was key to investigate, and IMHO I think this is still key and not a standard part of the routine. Designers often stop at the Magic of step two, not diving into the consequences of step 3. That should also trigger deeper thinking about the consequences and dilemmas of agency.

To end, some wishful thinking. Would the rise of generative AI and its mundane use train us not to overtrust technology? We might have a bumpy learning curve, but let’s hope that we will end with the continuous realization that the answers our AI agents provide might be based on hallucination, and we need to always add some grain of salt before blindly following the opinions of the tools we are offering. And could that even teach us a habit of being more critical in the polarised synthetics society? Let’s hope.

For the subscribers or first-time readers (welcome!), thanks for joining! A short general intro: I am Iskander Smit, educated as an industrial design engineer, and have worked in digital technology all my life, with a particular interest in digital-physical interactions and a focus on human-tech intelligence co-performance. I like to (critically) explore the near future in the context of cities of things. And organising ThingsCon. I call Target_is_New my practice for making sense of unpredictable futures in human-AI partnerships. That is the lens I use to capture interesting news and share a paper every week.

Notions from the news

So it was another hot week of AI agents. It seems the Agentic Era has started, or at least the hype. It has been a promise for some time, and as someone said in one of the tech podcasts, the Rabbit device was early to the party, jumping on the bandwagon that was not really on the move. Overpromising, though, that was soon clear. Now, Anthropic is getting it for real.

Some announcements: OpenAI is announcing its next big update for December. Or not. Google is introducing new AI music features. Notion is introducing a self organizing mailbox. Meta is introducing slim models for mobile use.

Human-AI partnerships

The computer does not need a human in the loop anymore. The new Claude functions trigger new steps in agentic AI.

Apple is rolling out the first AI features now but is still very careful. And focusing on enhancing existing functions. Confusing tech writers.

I am curious to see if this will work out; the cynical take could be that the real product of TikTok is the AI, not the stream of short-form movies.

AI watermarks in generated texts.

AI enhanced reading.

Robotic performances

A framework for safer autonomous vehicles.

AiDa is painting a portrait of Alan Turing, that will be auctioned. It feels like a piece of performance art together.

Waymo is gaining new growth money

Is that smart screen the same as the home robotic device?

A returning promise to support the disabled with robotic support.

Immersive connectedness

Humane was the promise of AI immersiveness but failed to deliver on the needed intuitiveness of immersive experiences. It is dropping prices and pivoting to software platforming.

Changes in the Raspberry Pi OS.

Immersive warfare.

What can we learn from our immersive sensing?

Tech societies

A self-improving AI implosion

The AI Hype Index

On the last Hardfork podcast this story on a teen committing suicide after building a relation with an Character.ai bot. The lawsuit would give insights in the responsibilities of the companies.

The relationship between AI crawlers and their sources is constantly shifting.

The new trade wars are on AI.

Biotech-inspired product inventions. Emotional soup.

Paper for the week

The fall and rise of sovereignty

“This article attempts to make sense of this fall and rise of sovereignty. Motivated by genealogical concerns, it historicises political and scholarly conversations on the concept. First, it examines the perceived decline of sovereignty following the Cold War. Second, it traces the proliferation of sovereignty discourses in the European 2000s and 2010s back to debates surrounding the 1992 Maastricht Treaty. Third, it explores the concept of sovereigntism, charting its move from political discourse into academic debates. Finally, the article concludes with a reflection on the trajectory of sovereignty in a post-Liberal International Order world.”

El Taki, K. (2024). The fall and rise of sovereignty. The British Journal of Politics and International Relations, 0(0). https://doi.org/10.1177/13691481241284220

Looking forward

After a week full of events last week, this week will be dedicated to making progress with the new research project at Civic Interaction Design group, and synthesizing the results of the hoodbot workshop iteration. And I will check the work of students at TUD with robotic explorations using the kit on Friday. Let’s see what Apple will introduce this week, how much of the Apple Intelligence will land in the new iOS and MacOS. Full functioning is not to be expected here before April.

A quick look at the calendars of events, I see there is a unconference on Public Tech this Friday afternoon in Rotterdam, and Speculative Futures Amsterdam is exploring food servings in 2100 this Thursday evening. For after the elections; this discussion is worth watching I think. Breaking Futures.

Enjoy your week!