Weeknotes 268; situated AI in the interstitium of things

Latest on human-machine-intelligence-autonomous and beyond. First reflections on new things emerging from the TH/NGS 2023 conference.

Weeknotes 268; situated AI in the interstitium of things

Hi y’all!

Last week was dedicated to TH/NGS 2023. We had a great edition of the conference. I really liked all the exhibition projects, mostly by students, and also the set of short pitches that created a rich overview of thinking into a reflexive year for Things. See below. Let me thank everyone for participating, speaking, exhibiting, organising and partnering to make it possible!

Next year, we are celebrating our 10-year anniversary and are planning to make it special, so save the date!

In other news, Threads was opened to Europeans last week. Like the rest of the world, it leverages its connection with Instagram to grow into a highly connected community rather quickly. Notifications of new followers popping up continuously. The whole Twitter sage is triggering a lot of diverted attention spans; see how this one will last. My first impression is that it is more a discovery stream behaving like TikTok than following your friends and ties feed like Twitter is (was) and Mastodon and Bluesky are. I need to interact more with it to teach the algorithmic feed to make it work, as it is rather messy now… I don’t see the benefit of merging with the Fediverse now.

Triggered thoughts

“We’re in a paradigm shift. We’re moving away from the scientific way of looking at the world as objects, to seeing a system-based world that’s all about fluid, currents, connections and relationships.” This is an interesting frame in this article, which was shared in this week’s Sentiers newsletter.

It connects me to some first reflections on the combined talks and sessions at ThingsCon. I need to take a bit more time for a full report, but it was clear that we are now entering a new phase of Things with the Agency. The connectedness slowly and surely becomes the intelligent layer. Our final speaker arrived not long ago before his talk. He opened by labelling it as ‘an AI conference’. That felt not completely right, but nevertheless, it had some truth in it. Discussing un/intended consequences of things connects to the new powers of AI embedded in everything. What is a thing after all? Do we grow into that interstitium as described in that article, and what does it mean for our relations with the physical world?

It is evolving thinking. The interstitium is what the predictive relations were about in a way. And as Iohanna Nicenboim spoke about situated AI in her talk, building on longer research. We need to look beyond users, to affected users, to the impact on designers, and also to the role the new agents play.

As Maria Luce Lupetti mentioned in her opening talk, what does it mean to be a researcher, designer, or engineer when it comes to engaging with AI? That is definitely an important path to continue, especially as AI is now really entering in all of our daily lives, including the things we use.

I am pretty sure this will stay the hot topic for the coming year.

Events

This is a short overview. End of the year, you more likely might end up in a Christmas celebration than a meetup 🙂

These are two I found:

  • 19 December: London IoT virtual Meetup
  • 20 December; highly recommended: a talk by Nazli Cila on design in collaboration with AI at Amsterdam UX.

Notions from the news

Regulators galore: Epic beats Google. UK pushbacks on Adobe. ITC is delaying Apple watches. EU is testing X.

Last week's story was on Gemini, Google's LLM engine, with a staged demo that created the counter-ripple effect. It is now opening up Pro level for developers. Will this lead to some news in the coming weeks?

It’s time for developers and enterprises to build with Gemini Pro
Gemini Pro is now available for developers and enterprises to build AI applications.

OpenAI is partnering with Axel Springer on AI in journalism. This will become a major topic in the coming year, full of manipulative news in a turning election year.

Axel Springer and OpenAI partner to deepen beneficial use of AI in journalism

Are you still planning to build your personal GPTs? An activity for the holiday season?

How to Build Your Personal GPTs: From Zero to AI Hero | HackerNoon
How to Build Your Personal GPTs: a step-by-step guide

The hot LLM of the week is Mistral. Introduced earlier, getting more traction now supported by serious investments and valuation last week. An overview.

Everybody’s talking about Mistral, an upstart French challenger to OpenAI
“Mixture of experts” Mixtral 8x7B helps open-weights AI punch above its weight class.

PDFs and conversational interfaces

LukeW | Video: PDFs & Conversational Interfaces
This two minute video from my How AI Ate My Website talk, highlights the importance of cleaning up the source materials used for conversational interfaces. It i…

“At the "Generative AI: Shaping the Future" symposium, iRobot co-founder Rodney Brooks warned against overestimating the capabilities of generative AI without understanding its limitations.”

What does the future hold for generative AI?
Rodney Brooks, co-founder of iRobot, kicks off a symposium on the promise and pitfalls of increasingly powerful AI tools like ChatGPT.

Fighting hallucinating AI: “Here we introduce FunSearch (short for searching in the function space), an evolutionary procedure based on pairing a pre-trained LLM with a systematic evaluator.”

Mathematical discoveries from program search with large language models - Nature
Nature - Mathematical discoveries from program search with large language models

OpenAI is attacking superalignment with weak-to-strong generalization. Can small models supervise large models?

Weak-to-strong generalization
We present a new research direction for superalignment, together with promising initial results: can we leverage the generalization properties of deep learning to control strong models with weak supervisors?
OpenAI’s Ilya Sutskever Has a Plan for Keeping Super-Intelligent AI in Check
The Superalignment team led by OpenAI chief scientist Ilya Sutskever has devised a way to guide the behavior of AI models as they get ever smarter.

Or are we remaining on the quest for the human-like supercomputer?

Human brain-like supercomputer with 228 trillion links coming in 2024
The world’s first supercomputer capable of simulating networks at the scale of the human brain has been announced by researchers at Western Sydney University.

Is this the first step towards AI designing the future humans?

MIT designs robotic heart chamber
MIT said the realistic model could aid the development of better heart implants and shed light on understudied heart disorders.

A provocation.

Are AI language models in hell?
Good links; a provocation.

Microsoft is doing a good job again executing copilot for facts. Not.

Microsoft’s AI Chatbot Replies to Election Questions With Conspiracies, Fake Scandals, and Lies
Research shared exclusively with WIRED shows that Copilot, Microsoft’s AI chatbot, often responds to questions about elections with lies and conspiracy theories.

Hacking chatbots that use GPT-4 for other tasks than built for is kind of easy.

Marketing Chat Bots As Generic ChatGPT Access Points – Interdependent Thoughts

As expected, AI is transforming the Meta Ray-Bans into a potentially troubling device.

Meta’s AI for Ray-Ban smart glasses can identify objects and translate languages
This is how Mark Zuckerberg chooses an outfit.

Enter the age of spatial photography. Will it be a shift indeed? The layout of cameras on phones might be changing, but also our approach to taking pictures.

Why Vision Pro Will Change Photography
For photographers and videographers, Apple’s Vision Pro will serve as a catalyst to rethink traditional photography. This technology, when combined with smartphones, represents yet another st…

What does this say about our relation to technology and social behaviour? Is this a sign of a future-mediated life or just a silly product idea?

This companion robot motivates socially inactive youth to venture out with mood-based conversations and persuasive activities - Yanko Design
With the advent of social media, most of us have found long-lost friends or made new meaningful contacts. This movement in technology has helped shape our personality; in some cases, however, it is threatening the fabric of social interaction with people closer and more physically available, leading to social phobia. The fear of venturing out

How an AI Researcher Uses ChatGPT and Notion AI - a podcast.

How an AI Researcher Uses ChatGPT and Notion AI
Linus Lee wants to bring the focus back to human agency when we turn to AI for creative work

Where are we with protocols as part of economy. The Protocol Kit is something to dive into.

In Search of Hardness
Protocol studies, the next crypto cycle, and the next age of the world

Remember pipes as protocols?

Pipe Dreams: The life and times of Yahoo Pipes
In 2007, a small team at Yahoo! briefly changed how we program the internet. This is their story.

Exploring blockchains as an urban concept

Walkable Blockchains: Metaphors Between Blockchains And Urbanism
Also: More Novellas, Axel Springer - OpenAI Deal, and Compulsory Licensing

General-purpose robotics is highly related to nature… and human habits

Electric Sheep Robotics acquires 2 more landscaping businesses - The Robot Report
With these new acquisitions, Electric Sheep Robotics has acquired four landscaping businesses in recent months.
Retail robots are slowly paving the way for industry disruption - The Robot Report
Retail robots can collect data in near real time and are gradually scaling in deployment, writes consultant Georges Mirza.

Is this a bottom-up or top-down form of transition? Rhetorical question. In the co-performance of man and machine, the heating systems were an example of fleshing out different stages in the new relations. This initiative of Renew home is another step.

Google’s Nest Renew joins Alphabet spinoff to form Renew Home
Google is offloading its Nest Renew energy-shifting program.

That time of the year, predictions for 2024 and reflections on predictions made a year ago.

14 predictions about 2024
AI! Elections! Threads! Media! And more

Looking back at the 2023 breakthroughs in science. And the best new materials.

The latest deck by Benedict Evans

Presentations — Benedict Evans
Every year, I produce a big presentation exploring macro and strategic trends in the tech industry. For 2024, ‘AI, and everything else’.

In other -important- news, not to forget…

Some thoughts on where the war in Ukraine is headed
And some perspective on what’s happened already.

Paper for this week

To balance a bit the AI-heavy newsletter, this paper is more practical, looking into the digital-physical relations: Digital twin applications in urban logistics: an overview. so that

“we survey previous research on DT applications in urban logistics as we found that a holistic overview is lacking. Using this knowledge in combination with the identification of key factors in urban logistics, we produce a conceptual model for the general design of an urban logistics DT through a knowledge graph.”

Abdo Abouelrous, Laurens Bliek & Yingqian Zhang (2023) Digital twin applications in urban logistics: an overview, Urban, Planning and Transport Research, 11:1,

https://doi.org/10.1080/21650020.2023.2216768

See you in 2024!

Next week will be taken up by family celebrations and cooking tasks, so the next edition will be on the 2nd of January. Enjoy the holidays!