Sign in Subscribe

Weeknotes 259 - Is AI eating itself?

By Iskander Smit in weeknotes —

Latest updates on tech and humans trying to live together in harmony, and many events on AI and beyond on the calendar.

Weeknotes 259 - Is AI eating itself?
a machine producing objects with the material of that same objects, according to Midjourney

Hi y’all,

What to say about the current situation: potentially big shifts in the world balance have been predicted for some time, and it seems another trigger is unfolding. Let’s hope for the best, de-escalation in any way… It will not be easy.

I am not gonna try to connect it to the usual topics of this newsletter however, our hyper-connected world is of course, also part of the context drivers of how it develops and how we experience it all.

Triggered thoughts

A lot is said about the influence of misinformation via the social media channels. Is there a discussion about that naming of social? We see social design, for instance, as a way to create inclusive and thoughtful interventions in people's lives. There is little social media left that seems to have that aim. Esther Perel was interviewed in the Pivot podcast and was referring to this also in light of AI developments. She made the punch that AI is artificial intimacy and that the way we have become part of networks of weak ties disassociates us from real human connections. With AI this might be amplified if we replace human friendships with artificial ones.

Today, I will participate in a workshop to formulate AI futures with a group of academics as a side event of the Design & AI Symposium. They asked as preparation to think about a future AI artefact that will stimulate the conversation about design & AI for the future. Thinking about this, I had to think back on the outcome of a project we did on The Civic City some years ago where we designed with residents of the neighbourhood Woensel-West in Eindhoven a device called WoeWe that aimed to collect stories from the neighbourhood carried around by the residents taking WoeWe in their homes for some time and creating in neighbourhood collective brain.

It resonates with experiences from the Wijkbot project in Afrikaanderwijk I shared before. We also see how residents like to create more of an engaging social robot that connects people than a purely functional one. I hope we can build on this to conquer a disassociated society…

Another event I visited and participated in a workshop on AI Maps, was RSD12, on co-design for societal transition. The opening panel was about Just Digital Life, and the opening keynote by Domenico Dentoni introduced some great groundwork for system mapping of unintended consequences. I added one of the papers as one to explore more below.

Domenico Dentoni at RSD12

Events

So many AI-related events these coming weeks…

And as a save the date, we are busy filling in the details for this year's ThingsCon December event. The theme is Un/ntended Consequences. Find more information via TH/NGS 2023.

Notions from the news

Google seems to introduce generative AI images by searching too..

Google now lets you generate AI images by searching - here’s how to do it
Google added two new features to its Search Generative Experience today, one of which allows users to create AI images on the fly.
BGRJacob Siegal

AI apps seem to be already quite popular for users, and those is the ones they label as AI apps; there are probably more that are silently AI

Mac users are embracing AI apps, study finds, with 42% using AI apps daily | TechCrunch
AI adoption among Mac app users is booming, according to a new report from app subscription service Setapp that found that 42% of Mac users today report
TechCrunchSarah Perez

Content credentials for AI-generated images. Introduced by Adobe, will that work? Would it be better to have an extension of the Creative Commons system?

Adobe launches new symbol to tag AI-generated content—but will anyone use it?
New icon, metadata seek to illuminate origins of content—AI-generated or otherwise.
Ars TechnicaBenj Edwards

How AI is stereotyping the world, if we let it loose.

How AI reduces the world to stereotypes
Rest of World analyzed 3,000 AI images to see how image generators visualize different countries and cultures.
Rest of WorldVictoria Turk

What if we add smell to it?

Computers Are Learning to Smell
AI could revolutionize our understanding of one of the most mysterious human senses.
The AtlanticMatteo Wong

And AI is a geopolitical factor. AI Panic

The New AI Panic
Washington and Beijing have been locked in a conflict over AI development. Now a new battle line is being drawn.
The AtlanticKaren Hao

And we knew about the environmental impact

The environmental impact of the AI revolution is starting to come into focus
Google Search could use as much electricity as a small country in a worst case scenario.
The Verge

I think the problem of a vicious circle if LLMs will base it’s knowledge more and more on generated knowledge by LLMs was mentioned here before; is the web eating itself?

Heather Ford: Is the Web Eating Itself? LLMs versus verifiability - Ethan Zuckerman
Can Wikipedia and other open knowledge projects survive the rise of LLMs? Dr. Heather Ford offers a path forward through verifiability.
Ethan ZuckermanEthan

Gary Marcus is critical on the progress too, based on comparing Dall-E 3 to 2.

Race, statistics, and the persistent cognitive limitations of DALL-E
Why DALL-E 3 is and isn’t better than DALL-E 2
Marcus on AIGary Marcus

Living AI via neural matrices.

To create living AI, replace neural networks with neural matrices - TechTalks
AI based on matrix neural networks will become a digital form of life, turning from an object into an active participant in the reality around us.
TechTalks - Technology solving problems... and creating new onesAlex Kostikov

These predictions on when AI will overtake human intelligence are interesting to observe to see how people respond to it. To the predictions, and later, to the intelligence itself; will we adapt and find a symbiose that makes it hard to define the moment at all…

Smarter than humans in 5 years? The breakneck pace of AI
There is an urgency for comprehensive, global governance of AI. Why getting this right could be critical for humanity.
VentureBeatGary Grossman, Edelman

Google research found a way to speed up analysing of contextual windows of AI models.

AI models can analyze thousands of words at a time. A Google researcher has found a way to increase that by millions.
‘Ring Attention’ is a new technique that massively expands the context window of AI models, according to researchers.
InsiderAlistair Barr

More on the performance of generative AI models, MIT has some achievements

MIT’s New Generative AI Outperforms Diffusion Models in Image Generation
Inspired by physics, a new generative model PFGM++ outperforms diffusion models in image generation. Generative AI, which is currently riding a crest of popular discourse, promises a world where the simple transforms into the complex — where a simple distribution evolves into intricate patterns o
SciTechDailyMike O’Neill

Next up: live deep fakes mapped on real people via new vision devices. It might not be allowed now, very well imaginable.

Apple Vision Pro live deepfakes concept - 9to5Mac
As we wait for Apple’s Vision Pro to officially launch in 2024, developers have been working on third-party apps for…
9to5MacMichael Potuck

Check also the VergeCast podcast of this week for some critical reflections on VR and hybrid devices as a response to the introduction of Meta Quest 3.

Meta Quest 3 hands-on review: VR rejoins the real world
A nice VR upgrade also offers a glimpse of a mixed-reality future.
Ars TechnicaKyle Orland

These experiments with AI designing robots are still more cute than creepy, I hope it will stay like that…

AI Designs Little Robots in 30 Seconds and They Keep Sprouting Legs
An AI used to build artificial neural networks can also create autonomous robot bodies with remarkable speed 
Scientific AmericanStephanie Pappas

Is this the next hopeful technology like we had the superconductors some time ago? Or just a fad?

‘Artificial photosynthesis’ breakthrough could lead to limitless energy
A new breakthrough in artificial photosynthesis could help create renewable, clean, and limitless energy for our planet.
BGRJoshua Hawkins

Over-the-air updates we knew, and another promise is over-the-air charging.

I’m Charging My Toothbrush With Wireless Power Over Distance—and It’s a Trip
Nikola Tesla once dreamed of transferring electrical energy through the air. Now, a company called Wi-Charge is beta-testing a prototype technology, and I’m testing it in my bathroom.
WIREDCondé Nast

The self-driving car seems to be further away every time.

We are decades off having self-driving cars that have the freedom to go anywhere, autonomous vehicle boss says
Oxa CEO Gavin Jackson said that the self-driving tech industry is “a long way away” from producing cars that can match human drivers.
InsiderTom Carter

This Disney robot looks more like a pet than the dogs we used to have.

watch disney’s new two-legged robot dance, emote and follow people around
disney research’s new wall-e-looking robot consists of 3D-printed parts and reinforcement learning which allow it to be emotive.
designboom | architecture & design magazinematthew burgos | designboom

We have seen the concept videos for years (like the IKEA ones made by Space10), but apparently, there are real plans for spaces on the move…

Panasonic Wants Us To Drive Around In Mobile Autonomous Living Rooms By 2035 | Carscoops
In partnership with Panasonic, the Japan Mobility Show will provide attendees with the chance to experience a virtual ride into a lounge-like autonomous future
CarscoopsSebastien Bell

This seems like a nice trick to replace Siri with ChatGPT via the action button on the latest iPhone, but I think -at this moment- the conceptual assistant that Siri is (or wants to be) and ChatGPT is not comparable…

But still, something to try sometime, maybe.

How to replace Siri with ChatGPT on your iPhone 15 Pro
You can replace Siri with ChatGPT thanks to the iPhone 15 Pros’ Action button - here’s how easy it is to search the web.
BGRChris Smith

If you are into experimenting with creating your own LLMs this might be interesting to check.

alpaca-lora: Experimenting With a Home-Cooked Large Language Model | HackerNoon
How to fine-tune LlaMA on a low-end GPU and still produce great results.
apple-logoWei

The curse of the creator economy; is everyone an influencer after all? Is that all that will be left? Not sure. Or is it just replacing mundane use of media, and is it just what we do? is everything monetised and will the platforms have automated revenue share for all?

The Curse of the Creator Economy
In her new book Extremely Online, Taylor Lorenz argues that everyone can be an influencer. The history of the internet suggests relatively few will strike it rich.
WIREDCondé Nast

Paper for this week

As mentioned above, the presentation by Domenico Dentoni at RSD12 made me curious to the paper he used as a source for his insights on system mapping.

Systems Thinking, Mapping and Change in Food and Agriculture

(…) we propose a collective process to systems mapping that grounds on the notion that the configuration of problems (i.e., how multiple issues entangle with each other) and the configuration of actors (i.e., how multiple actors relate to each other and share resources) represent two sides of the same coin.

(…), we provide implications for societal actors - including decision-makers, trainers and facilitators - using systems mapping to trigger or accelerate systems change in five purposive ways: targeting multiple goals; generating ripple effects; mitigating unintended consequences; tackling systemic constraints, and collaborating with unconventional partners.

D. Dentoni, C. Cucchi, M. Roglic, R. Lubberink, R. Bender-Sala- zar, T. Manyise (2022). Systems Thinking, Mapping and Change in Food and Agriculture. Bio-based and Applied Economics 11(4): 277-301. doi: 10.36253/ bae-13930

Link

See you all next week!

Or earlier in person at Design & AI or DDW.