Weeknotes 264 - unintended consequences of AGI projections

By Iskander Smit in weeknotes — Nov 21, 2023

The whole OpenAI saga is still developing, so these reflections might be outdated… There is more news that contributes to longevity insights, though.

Midjourney's interpretation of unintended consequences of AGI projections

Hi, y’all!

This week is elections in the Netherlands, with even some extreme right-wing parties as potential winners. The devastating situation in Gaza is at another critical moment; in Ukraine, chances seem to change for the worse, and the UN warns for 3C heating. I felt the need to acknowledge this first before diving into another big news event in the context of this newsletter that unrolled this weekend: the OpenAI saga…

At this very moment, Microsoft seems to play a master move in the end, having all the IP secured, controlling the computing power and now also having the senior leadership on board… Ben Thompson wrote a great analyses:

What is clear is that Altman and Microsoft are in the driver seat of AI. Microsoft has the IP and will soon have the team to combine with its cash and infrastructure, while shedding coordination problems inherent in their partnership with OpenAI previously (and, of course, they are still partners with OpenAI!).

I am not even trying to share all the news articles and tech analysts’ opinions here (like this, these, this, this, or this); it is clear that it is not so easy to build a world-dominating technology with society guardrails from within a company space. I hope that initiatives such as the EU AI act can offer some resistance to growing into villain-types we know from the superhero movies…

Also, just dropped on Youtube, a TED talk taped on 17 October by one of the remaining OpenAI tribe, Ilya Sutskever. He is optimistic both at AGI surpassing humans and a future where humans and AI live in a healthy collaboration mode.

Triggered thoughts

To elaborate a bit more on this, the predicted future of an AI-driven apathetic life building in the coming decade as we created for the earlier mentioned Design & AI pre-conference workshop, a future ACM Interactions magazine. The Prompt Fatigue was the metaphor here, in an interview this weekend (Dutch) with French philosopher Eric Sadin, the same phenomenon was indicated as promptism. He rightly points out that it is not about the total extinction of humans that we need to fear but that we might lose our human capacities, being numb without any agency.

More than fighting AI as a tool, I think that we need to create conscious designers who are aware and design human friction into the systems and be brave enough not to create the most easy-to-use seamless apps to support our life into a numb experience.

And, back to the opening, hopefully, we will not be forced into a state of escapism due to all that happens on a global scale (think Ready Player One)…

Events to check

Good to shift perspective; where can you meet interesting speakers and people?

Of course, check out TH/NGS 2023 (15 December); we updated the website with the program details.
23 November, Rotterdam, Neural, critical digital culture and media arts
24 November, London, How is Artificial Intelligence transforming Urban Design?
27 November, Amsterdam, Design for impact and a living, distributed design
29 November - Product Tank AMS on data https://www.meetup.com/producttank-ams/events/297437877/
30 November, Amsterdam, rescheduled, Pinch about Responsible apps

Notions from the news

There is more than OpenAI. Before all developments were unrolled, Microsoft introduced custom AI copilots. As The Verge podcast rightly mentioned, the branding of copilots is messy mixing specific tools and category.

And in another way, brands and AI can clash.

Deepmind introduces an AI pop star clone that can create new music including videos of existing icons. Youtube is preparing…

OpenAI as a peer for AI Models capabilities might change, but still, for now:

New AI-powered tools by Instagram, and more picture tools within Meta to distort reality. Last week, I described the possibility of having a GPT from your personal link list and other saved sources. It seems that Notion is working on this: Q&A your own enhanced brain…

AI tools can be discontinued too, like Clyde from Discord.

An overview of some AI tooling that has entered the world of architectural design. I hope the London event mentioned above will be captured, as it will deliver some more fundamental thoughts on the change in the design of urban space.

Mundane coding is part of developing low-friction living (see above); Make it Real creates code for games from drawings.

The designer's role in AI and AI in the design is that the AirBnB founder has an opinion; you need to team up.

Create new ways of discovery, mixing real and fake products to understand better what you really like; interesting experiment by Google.

More Google: solving the biggest prediction game of all…

An AI that can learn language with human-like efficiency through "meta-learning for compositionality…

Following also the bigger developments sketched above, Scott Belsky dives into the implications: “When we lose the reins of our attention, we compromise our judgment. And, why should hybrid teams miss out on the benefits of proximity and spontaneity? Let's explore the implications of these topics.”

I think we did not have AI images influence our elections here, but some parodies were made. I wonder if it was different in Argentina…

Oh, this summary made within Readwise is rather convoluted… The article argues that the current approaches to artificial intelligence, which focus on formal mathematical expression and leaderboard success, are reminiscent of the discredited epistemological foundations of logical positivism and behaviorism. The fixation on formalism and benchmark performance fails to consider alternative hypotheses and inhibits scientific progress.

Here are some more thoughts on wearing a device like the Humane pin.

Smart home devices with advanced artificial intelligence (AI) models, such as generative AI, could significantly increase energy demands, according to a report in the journal Joule. No surprise, however, there might be a way to optimise the activity in the end?

Robots as sergeants…

I like artists that create experiences and narratives to make you think. Like Tinkebell in the Netherlands and Maria Abramovic too.

Holiday gifts? Or unnecessary gimmicks? Can be the same of course.

Paper for the week

This feels timely… Authoritarian urbanism: on the normativity of a research field (German)

In view of increasing authoritarian, illiberal and populist trends worldwide, not least within established democracies, the examination of authoritarianism in interdisciplinary urban research has gained in importance in recent years. This article takes this renewed popularity as an opportunity to examine the topicality and development of the research field on authoritarian urbanism. The article reconstructs what was understood and researched under authoritarian urbanism and how it developed over time. In doing so, he identifies established assumptions and normative attributions in the study of authoritarian urbanism, shows how these arose and have influenced the debates since then, and discusses how these stand in the way of a better understanding of authoritarian urbanism.

Zupan, D. (2023). Autoritärer Urbanismus: zur Normativität eines Forschungsfeldes. sub\urban. zeitschrift für kritische stadtforschung, 11(3/4), 17-41.

https://doi.org/10.36900/suburban.v11i3/4.864

See you all next week!

Hold tight. A lot of things can change in a week!