Hi! A crypto crash, a democratic uptake, climate hell, twitter meltdown, one of those weeks…
Updates from last week
Last week we had a good week with Structural meeting with our first user and having conversations with two others. The students at Rotterdam UAS working on another iteration of the Wijkbot made some serious steps to make it more autonomous.
We found a location for the ThingsCon Winter Unconference 2022 and sent out our first mailing. Check it out if you like to join us on 9 December 13:00-20:00!
I listened to the interview by Ben Thompson with MidJourney founder David Holz, and shared how inspiring it was on Mattermost (ok, also on Twitter), and got a remark that it was a pity it was behind a paywall. So let me share some of my impressions here.
“In the end, the best way of knowing is starting by consciously not-knowing.” Ben Thompson is summarising his interview with David Holz, founder of MidJourney, one of the generative AI tools that use the techniques that DallE has become famous for. I really liked the interview. It is only accessible to members of Stratechery, but let me share some of the nice insights.
The tool of MidJourney has its own style of images with a certain dreamy feel to it that is different from DallE and Dreamstudio, although the latter can produce similar images. What I found most interesting to hear are the choices they made in developing the tool to stimulate imagining, the prompt they literally use in the tool. The tool is built in Discord as a chatbot that you can ask to imagine your imagery. It does not try to be a human though, as often is done with bots. It is honest as a tool. But the whole setup is creating that the tool is mixing human and machine creativity in an exploring generative way. As a first-time user, you land in a newbie channel to find out the tricks and learn from other newbies. Being aware of all other art that is generated remains a part of the experience. You can see what others imagine, what prompts they use, and what results they deliver. This is a rich way of learning.
Another aspect they carefully designed is the waiting time for images to build up. It is a benefit to have it built up not all at once, but seeing it come together, makes you understand even more how it generates your imagination.
David says that he aims to build new human infrastructure as “the world’s going to need a lot of new things, and we need infrastructure to build new things”. It connects a lot to the concept of co-performance I have mentioned here before and written about here. AI is not a dumb tool for humans, it is a partner in exploring shared goals. Generative AI is the most direct form of that concept, the most transparent to use that word.
“a lot of people who want to make the icon for the chatbot a little robot. And I was like, “Absolutely not. This is not a robot. You are not working with a robot. It’s a vehicle. We put a little sailboat on there.”
Mid-journey is telling: “There’s no conversation between you and the computer, but then there is a conversation between you and other people.”
David shared more on his way of developing the tool, and now into V4, which is a big release. “Version four is in a weird place right now where we’re still using some of the data from the older versions to kind of define the look, but the system itself is so different. It really should have a new look of its own.” It is not marketed widely as there will be stress on GPUs if the growth is too rapid; the GPU crunch. They are one of the largest GPU users in the world now. An image is now generated balancing GPUs from eight different regions from all over the world.
There was a lot more in the interview, but I found the conversation on building a generative AI tool most interesting. In this weekly update Ben Thompson shared some parts too and is accessible without a subscription.
Events of interest
Ok, a short list of events that might be interesting.
- We visited Glow this weekend in Eindhoven. If you are into light art is always nice. I am not sure, but I think the number of good work vs. average was a bit lower this year, but still, there is enough to see.
- We were not able to visit the Evoluon, but next week there is an event that I hope to attend.
- That Smart City Expo conference is happening now in Barcelona; in case you work at a city on smartness or are a vendor that something has to sell to cities 🙂
- Sensemakers is doing a workshop on 3D printing and laser cutting this Wednesday.
- Today you can still (remotely) visit a symposium Humans at the Centre of HRI.
- Immersive Tech Week is approaching; in Rotterdam.
- And as said, 9 December we are organizing ThingsCon Winter Unconference
Captured news
Let’s look at the interesting news articles of last week dealing with human-machine interactions, responsible tech, autonomous systems, and generative intelligence.
In general, I like to share the blog of Matt Webb – Interconnected. This past week again, two articles worth reading I think.
Continue reading Weeknotes 214; the invisible present