Another week is full of AI news. Steered, of course, by that big happening in Austin as mentioned last week, it was a moment to announce new versions (GPT-4, Baidu, Claude), or integrations of AI in tools (Google, Office, Linkedin), language models (PaLM from Google). While Microsoft is getting extra attention, laying off the Responsible AI Team. Pretty sure GPT-4 was the most anticipated and discussed. The CEO did an impressive demo where the more conceptual thinking capabilities and the visual capturing stole the show. And now it is passing the bar exams with high numbers. Creating a -still quite ugly- functioning website from a sketch on paper was impressive.
Having a conversation with logic-aware language models that are creative and balanced (MidJourney)
Hi all,
Last week’s newsletter had a couple of wrong links; I was pointed to me (thanks, Harm). Luckily it turned out that only some of the links were in the introduction and update, so I chose not to send an updated newsletter. Of course, I corrected it in the online version. This week I will do an extra test for sure!
Some updates; we are very happy to have two speakers confirmed for the ThingsCon Salon on Listening Things on 14 April in Eindhoven. Bob Corporaal of CleverFranke will discuss an inspiring project visualising the feel of smart homes. Elif Ozcan Vieira is an associate professor at TU Delft, specialising in design for and with sound. Find more information on the website. We expect one or two additions to the program this week.
“news as artifacts that represents the truth in inconvenience”, by MidJourney
Hi all!
Welcome (again) to this weekly update newsletter. For those who are relatively new; (almost), every week, I capture interesting news on human-tech relations, from robotics to intelligent systems and protocol economy to … what catches my eyes. I also check interesting events that I might visit (if I have the time), so they are mainly in the Netherlands or online. If I manage to visit, I will try to share some impressions. And I share every week a paper that got my attention or I read before.
As my weeks are rather busy working at the startup Structural and running some research projects around Cities of Things, and organising ThingsCon events in the leftover time, the reading list is for a large part ‘hope to read’, but I always try to capture the gist of it at least. :-)
Enough for this extended introduction. I did not attend an event (only one movie). Happy to have a new student team and graduation student starting at the Lab010 Wijkbot project. I will keep you posted on this for sure.
A bit further into the future: Mozfest will be held end of March. Last year I liked the broad mix of topics and engagement. Trustworthy AI is an on-topic theme, of course. Navigating the programme is always a challenge.
The impact of Augmented Reality… It is a topic that can be interesting from a conceptual level or too focused on finding new applications for existing media… I hope this one will be more than the first.
Still, there is a problem with the behaviour and manners in the conversations with a couple of the users. It brings back memories of an earlier attempt by Microsoft to create a companion bot, now one of the most used examples in AI presentations. Tay became, within 24 hours, an ultra rude bot trolling, and was switched off right after.
It will be a pity if this influences the development of AI and the like. There is so much potential in professional services based on these tools.
One of the strategies can be to build opportunities to oppose the behaviour. Contestable AI, Kars, is doing a whole PhD on this topic and he presented shortly during the 2 year anniversary of Responsible Sensing Lab on Thursday. I could not make it in person in the end but watched the recordings. These are some of my impressions.
Peter-Paul Verbeek had a keynote on democratizing the ethics of smart cities. We are now entering Society 5.0, living a digital life. Robots become citizens. He wondered if AI is learning the same way as humans do, if AI has its own agency or if it is always derived from the relation with humans.
He showed how sensing is also shaping the way we see our world. Technology is not only a tool; it is also shaping how we are in touch with our world. This is a topic to address as designers of these sensing environments. Smart cities are part of politics. In AI systems, it is always the question of who has agency: the operator, the control unit, or the drone itself. Citizen ethics can be an important concept; ethics developed by the citizens. There are three stages: (1) technology in context, (2) the dialogue, and (3) options for actions.
After the keynote, Thijs Turel and Sam Smits looked back on two years of RSL and to the future. An important theme is to design data collectors in the city to design for just enough. An interesting aspect is how, via “designing” tender requirements, we can set goals for climate, among others. Next to a lot of examples, the presentation included the development of the scanned car; what does that do for the people living in the city? If we have a 100% chance of being fined, it should be taken into account in the democratic decisions that were the basis for the level of fines.
Kars Alfrink was sharing his concept of Contestable AI. Leveraging disagreements to improve the systems, validating the concepts of enabling civic participation, ensuring democratic embedding, and building capacity for responsibility. For example, we should not discuss if scan cars or cameras are part of political programs and promises; we should connect them to the values behind them and acknowledge – like Peter-Paul Verbeek was presenting – how technologies mediate not only the things and services we use but also the political decisions behind these.
It relates nicely to an essay by Maxim Februari that was published in NRC on Saturday. Democracy is not a product with a certain outcome; it is the process that counts, and that should be stimulated. There is much more to say, but it is best to read it yourself.
Wars and natural disasters, rough times… Hard to shift to that totally different dispute, another week full of developments in the A.I. Arms Race… Microsoft declaring war… Google lost 100 billion in value due to one presentation; Last week I mentioned the announcement of Bard, but the presentation on Wednesday was disappointing. Everyone was surprised at how little progress the AI-first company had made in years. Microsoft’s CEO Satya Nadella, in the meantime, made an interesting frame: all computing will start with a draft…
While preparing an update and roadmap for the AI strategy for Structural, I got back to a reflection pre ChatGPT on the role of the model by Bratton and Barcas “The Model is the Message“, referencing the famous McLuhan of course; the difference between services will be defined by the models you use; OpenAI, Google, or others. And with ChatGPT the UI is becoming more and more key to defining freshness and quality.
The integration of AI functions has become part of all. Or just ChatGPT integrations, like in Opera browser. Microsoft is going one step further with AI copilot as part of Edge browser. ChainGPT, is an AI model designed for blockchain. The development was nicely framed by The New Yorker as “ChatGPT is a blurry jpeg of the web”. New variants are popping up in different big tech, like in China with Alibaba. Much more can be found in the noticed news of last week.
Before jumping to these, here are some potential interesting events to join online or in person…
A webinar on tech ethics with Cennydd Bowles amongst others: Privacy, Seriously (23rd of February).
This Thursday the Responsible Sensing Lab is celebrating one year of existence with a great line-up.
New Art in v2, opened last week, runs through till March.
“Yielding a road with multiple driving robots playing music as bards” (MidJourney)
Hi all!
Thanks for landing here. Another week went by fast. And as all editions of the last couple of months, the biggest things are about generative AI developments. This week there might be another key moment as Google is believed to introduce a response in a special announcement event this Wednesday. It also invested in a competitor of OpenAI. On Monday, Google announced Bard; “We’ve been working on an experimental conversational AI service, powered by LaMDA, that we’re calling Bard. And today, we’re taking another step forward by opening it up to trusted testers ahead of making it more widely available to the public in the coming weeks.” The company is a bit nervous about the rapid growth of ChatGPT and the stakes of Microsoft in it. This video -which was suggested to me by the YouTube algorithm- gives a nice overview of this “A.I. War”. Or maybe is A.I. Arms race a better framing.
One thing that I think might not be the right angle is the focus on the search fight. I agree that we might evolve into new ways of searching, specifically more conceptual questions, but I don’t think that the real battlefield is in the search domain. The sense that Bing is not a competitor for Google, even if they would integrate natural language search conversations into it. Plain search will remain kind of the same. But we will find integrations in all other things we use, from digital tools and services to even things that become more understanding and start to predict. Having GPT4-empowered Office tooling is much more powerful, and the co-pilot in GitHub will make programming so much more accessible; allowing you to create your own personal apps for everything easily. These were promises for years; we live a computational life that will only become more real, without notice.
I had to think back, of course, to the research explorations I did back in 2018 and 2019 to predictive relations. What would it mean for our perception of using things as the behaviour is based on predictive data? How would design change? Will it alienate or enhance our relations? One of the students I coached brought it into research through design practice for a vacuum robot and wrote this paper about it. I maybe should revisit this topic if I have time… For now, I am focusing on the development of the AI-powered platform of Structural.
“morphing fantasies of Weizenbaum warnings”, via MidJourney
Hi!
Today I do a short update and keep it to the captured news of the week. Last week it was mostly crunch time at Structural. I attended a meetup of IxDA London on AI for creative work, discussing the potential for creativity and having a new type of user research. Like this Figma add-on.
We also had a nice design session with the think tank of Afrikaanderwijk for the Cities of Things Lab 010 project. We are now starting to design together a ‘robot’ for the neighbourhood as part of the co-design.
Lorna and I are shaping a ThingsCon Salon in Eindhoven as part of the STRP program on Listening Things. More information soon, save the date: 14 April 19-21:30 at AI Innovation Lab, High Tech Campus, Eindhoven.
Speaking of events. These are some happening the coming week:
A room that helps mapping knowledge in a hybrid combination of physical and virtual thinking. by MidJourney
Hi everyone. Last week’s top-of-mind news was a bit hard to think of… Scrolling back the layoffs in Silicon Valley remain on top of the news; Alphabet, Spotify, and no exceptions it seems, apart from Apple. The latter is doing some silent product updates and switching on some hidden features in the Homepod mini, which is the way to go for product development nowadays. Nice to see how someone improved Siri by using a ChatGPT integration. Speaking AI; The speed-up is still happening, and Microsoft has now officially made an additional investment in OpenAI, with new tools continuously appearing. Questions about the Intellectual Property of the generated visuals and text have been raised, and legal claims have been announced. In the meantime, Google is believed to be freaking out on ChatGPT. At the same time, Google Research is sharing its research progress. It is passing the Turing test already. It needs cheap human labor still to get it less toxic, Time claims.
I attended the launch of a new book – or rather, a collection of “cahiers” – called Action Design for Urban Futures, written by Ben Schouten and many other contributors. It is intended to form a Civic Empowerment Toolbox to help local movements organize activities to improve their neighborhoods and collective personal environment. It is targeted towards future designers, scholars, and policymakers, but also – and perhaps especially – those who are initiating these kinds of changes, with a planning tool to jump-start their civic initiatives. The planning tool is a matrix covering aspects of empowerment on one axis (mobilization, organization, operation) and societal levels of influence (individual, collective, institution). Previous initiatives are evaluated to gain learnings. The most actionable part is a game-like canvas to explore all these aspects together.
It is definitely an interesting toolkit. In the end, it is probably more useful for the professionals involved in bottom-up initiatives. I will look into it more deeply as part of our initiatives connected to neighborhood-driven design for the Cities of Things.
In the meantime, we have started the development of (software) tooling for the STRCTRL method and language. Working on the prototype and preparing for the next iterations will be the focus for the coming months.
Events in the Coming Week: What to Do and See? For those who are not following me on Instagram, there is a nice hidden gem in the old Foodcenter Amsterdam we explored; Markt Centraal is organizing evenings and lunches in the central market building that is worth a visit if you are curious about the heritage (like an original Keith Haring piece on the old Cooling Building) and nice food.
The more regular tips:
Tonite, Design for Planet is happening again in London and online, with a lovely line-up.
“A man on the couch of an AI therapist”, by MidJourney
Hi all! Another week full of ChatGPT and comparable AI explorations. Below are some new examples. I will keep the update short this week. Working on the startup and keeping track of the news. This week there are some events planned that I might attend so will report on this next week.
This is the second newsletter I sent out via Ghost instead of Revue. This is a symptom of what is happening with Twitter; winding down in a way. Or crashing in a race to the bottom… Next to the newsletter saga, now my favorite Twitter client Tweetbot is broken due to new API rulings of Twitter. Getting everyone on their own apps is an understandable strategy for commercial reasons maybe but it might break trust in the tool even more… The open character of Twitter stimulating the platform to grow as a messaging platform was once a strong strategy. But that was when Twitter was more of a community and communication tool, nowadays it has become more of a publication platform…
It is easy to make a connection with ChatGPT. You can make the case that the tool is in the same phase of opening up to all makers to create their own new applications or at least find their own applications in an open platform. Not exactly the same, of course, 15 years later, the baseline of use of platform tooling has changed. Nevertheless, we might see a more closed strategy as soon as the business model becomes dominant. I wonder if this timeline from an open platform to a closed system is described in the literature… The first pilot of ChatGPT professional just started.
Events this week
I might check out the network Newyear drinks of Amsterdam Trade, who we are in contact with for the MUC AMS Cities of Things field lab
On Wednesday there is an event on Action design for urban futures organized by some scholars I know from Amsterdam University of Applied Sciences and Technical University Eindhoven
Back in the day, I attended Product tank Amsterdam quite regularly. Tonite, there is another meetup (I will miss).
Hi all! As mentioned last week I am moving the newsletter to a new tool as the former one (Revue) is discontinued from next week, and all data will be lost. I am now extra very happy to have been very consistent in copying the content of my newsletter to my blog. So I have been considering a couple of tools as new email engine, like Emailoctupus, Beehiiv, and Buttondown, and even just using my WordPress blog as email-tool too. In the end I choose for Ghost however. Mainly as it has the best way to make nice bookmarks previews in the email, what is the main part of this weekly newsletter of course. It is not as good as Revue was for that goal, it misses part of the workflow. But for now it seems to work nicely. Welcome to Ghost!
So I imported the subscription list and so you have received this newsletter via Ghost now at the usual time. I will send out another short one via Revue in case this one ends up in a spam folder. So in case you did not receive this one in your mailbox, you might check your spam folder (or else subscribe :-) )…
So long for the practical stuff. My first week of 2023 was dedicated to working on Structural. We moved to our first dedicated office in Amsterdam West, as mentioned last week, we are settling and occupying the walls by mapping out our language and plans for tooling. I also wrote some 2023 outlooks (see below) and caught up with Wilbert and Marco (as OG Behavior Design AMS team :) ).
News update: the core of this newsletter: what were the interesting news articles? There was CES still, of course, with lots of new (TV) screens, strange cars (see below), and the introduction of Matter in everything. The Verge has a nice overview. Or listen to the podcast of Stacey Higginbotham to catch up on all that Matter…
target is new 