Weeknotes 266 - precepts of our entangled artificial life

Hi y’all, Happy Sinterklaas if you are celebrating this…

The week of a climate conference “in het hol van de leeuw” while the urgency is still not clear to all. (Perplexity: “in the lion’s den” describes a situation where someone is in a place or situation where they are in great danger or facing a formidable opponent).

Triggered thought

Listening to the Poki podcast last week, Wietse referred to the concept of precepts and how these might be a different form of mastering AI and GPTs. Or better, how the real co-performance or deep partnership with AI is played out when you can communicate on the level of precepts and not on the outcomes. I had to think about a presentation at This Happened back in the day by an architect working at a Dutch architectural agency ONL of Kas Oosterhuis famous for integrating technology as part of the design material. The project presented was a design for a climbing hall, and the interesting part was that they managed to create a bespoke architecture for the walls by not designing the components as objects but by designing the CNC machine that would produce the components. Which was already on a deeper level even than the so-called “transformative algorithmic design”.

This is the returning theme I think, that might pop-up almost every week. We are increasingly conscious of how nature is based on entanglement (read this book, or this); the next entanglement is the potential of the new forms of intelligence that emerge from our merge with the precepts…

Events to check

• Just opened: AICON Lab in Rotterdam
• 6 December, Amsterdam - The Hmm; Ask Us Anything: Hotline on AI Advice
• 6 December, Amsterdam - DIY Sensemakers Labs
• 7 December, online - v2; Shifting Cosmologies: more than human AI
• 7 December, Delft, online - Data-Centric Design Symposium
• 8 December, Rotterdam - Creative Mornings
• 11 December, online - the future of global AI governance https://www.verses.ai/ai-governance
• 15 December, Rotterdam - TH/NGS 2023 (register for the early bird price with code FRIENDofTHINGS-IS )

Notions from the news

Next to updates on AI, you can do a weekly report on the latest on OpenAI (1, 2, 3,) and also on Elon’s strange moves of the week. The latter is becoming more and more disassociated from reality. After moving to Mastodon and Blue Sky, Threads becomes another option for Europeans too.

Google is pushing back it’s next-level AI, quietly, but ainvent new materials from history. Amazon in introducing a hallucinating Q.

AI assist

Dan Shipper describes how to leverage ChatGPT to assist in decision-making by finding out what others have figured out before.

AI-assisted Decision-making

How to use ChatGPT to master the best of what other people have figured out

Chain of ThoughtDan Shipper

Edge computing was a thing, the thing even, before LLMs became dominant. Could make a comeback with personal AI-assistants.

Bespoke GPTs are taking off but with consequences.

I made a GPT, a custom version of OpenAI’s ChatGPT, and it only took me 15 minutes. Here’s how.

GPTs can specialize in tasks like coding, creative writing, and tech support.

InsiderAaron Mok

OpenAI’s Custom Chatbots Are Leaking Their Secrets

Released earlier this month, OpenAI’s GPTs let anyone create custom chatbots. But some of the data they’re built on is easily exposed.

WIREDCondé Nast

GenAI in the enterprise, longread and data-intense, scenario planning might be a useful tool

Generative AI in the Enterprise

O’Reilly MediaMike Loukides

Use GenAI to Improve Scenario Planning

Generative AI can help organizations overcome inherent shortcomings in conventional processes for performing contingency scenario planning. This article explores how the technology can do that and provides an example: a simulation that the authors conducted.

Harvard Business ReviewDaniel J. Finkenstadt,

New assistants of the week: iA Writer helps to explore what is real, Personal Voice for disabled phone users, smart router, seamless translations.

And Stability AI is introducing realtime text-to-image generation. You still need some computing power, but real-time prompting is next level in human-machine interaction.

AI harms

The warning against harms is not always motivated by the right intentions, researchers found. The solutions are not helping either… On a different level, the carbon footprint of AI is explored in more detail. It is a catch-22 if you are projecting it as a solution for climate change at COP28.

Model alignment protects against accidental harms, not intentional ones

The hand wringing about failures of model alignment is misguided

AI Snake OilArvind Narayanan

“Super-advanced artificial intelligence (AI), left unchecked, has a “serious chance” of surpassing humans to become the next “apex species” of the planet, according to Ethereum co-founder Vitalik Buterin.” And militarised AI is a higher risk than killer robots (is that not the same?)

Vitalik Buterin: AI may surpass humans as the ‘apex species’

Superintelligent AI can become the “apex species,” which could result in human extinction, says Ethereum co-founder Vitalik Buterin.

CointelegraphBrayden Lindrea

The militarized AI risk that’s bigger than “killer robots”

The nuclear stakes of putting too much trust in AI.

VoxJeffrey Lewis

What is real? Are memes human? Don’t trust your camera. Or pizza.

AI of A stupid? And how do we learn AI our values?

Researchers Made an IQ Test for AI, Found They’re All Pretty Stupid

Artificial intelligence programs still struggle with basic problem-solving skills that people excel at, new research claims.

GizmodoThe A.V. Club

To help autonomous vehicles make moral decisions, researchers ditch the ‘trolley problem’

Researchers have developed a new experiment to better understand what people view as moral and immoral decisions related to driving vehicles, with the goal of collecting data to train autonomous vehicles how to make “good” decisions. The work is designed to capture a more realistic array of moral ch…

Tech XploreMatt Shipman

AI is eating the internet. And slowly chewing your career.

Slow-burn AI: When augmentation, not automation, is the real threat

Automation eats your job, but augmentation eats your career

Platforms, AI, and the Economics of BigTechSangeet Paul Choudary

Robot intel

Teaching robots smart, co-learning

Princeton Engineering - How do you make a robot smarter? Program it to know what it doesn’t know.

Engineers at Princeton University and Google have come up with a new way to teach robots to know when they don’t know. The technique involves quantifying the fuzziness of human language and using that measurement to tell robots when to ask for further directions.

Princeton EngineeringMolly Sharlach

Sensing

A new product line of IKEA.

Ikea sensors for under $10 each; Matter support promised for hub

Three new Ikea sensors are being added to the company’s smart home product line-up, each set to cost less than…

9to5MacBen Lovejoy

Sensing the weather, and vibes

Prediction Markets as Conversations

Also: DeepMind’s Weather Modelling, BasePaint, and How Running Feels

Scenes with SimonSimon de la Rouviere

Vibes vs. data

It’s asymmetric warfare out there.

NoahpinionNoah Smith

Sceptical futures? Or realistic?

Dispatch from the Future: The Must-Have Gadgets and Gear of 2053

To celebrate WIRED’s 30th birthday, we asked the experts—and our imaginations—to dream up the cars, phones, televisions, and other tools of tomorrow.

WIREDCondé Nast

Paper for the week

To keep in the theme of entanglement: Terms of entanglement: a posthumanist reading of Terms of Service

In this paper, we use ToS as an entrance point to explore design practices for democratic data governance. Drawing on posthuman perspectives, we make three posthuman design moves exploring entanglements, decentering, and co-performance in relation to Terms of Service. Through these explorations we begin to sketch a space for design to engage with democratic data governance through a practice of what we call revealing design that is aimed at meaningfully making visible these complex networked relations in actionable ways.

Seda Özçetin & Heather Wiltse (2023) Terms of entanglement: a posthumanist reading of Terms of Service, Human–Computer Interaction, DOI: 10.1080/07370024.2023.2281928

See y’all next week!

When I started the weeknotes some time ago, I also had the habit of updating on my personal activities, as the weeknotes-format often does. The newsletter grew (deliberately) more into a newsletter than weeknotes, but I thought a monthly overview might be nice to add here. So…

As you might know, I started in October as an independent futures researcher and beyond as a sort of extension from this newsletter. Check the about page for more details.

I have been asked to participate in writing a new agenda for the Dutch creative industries as a part of the KIA MV. And also exploring the forming of a coalition program on the impact of digitalising and precarious groups. I will share more if it is more public.

In the last months, I was very busy organising ThingsCon, and rounding up the Cities of Things LAB010 project. In the new year, I will start looking into follow-up initiatives for Wijkbot and actively detail my future research services (find a rough outline here).

So please reach out if you like to have a coffee or call in early January!

Have a great week!