In case you missed it, a couple months ago, Figure released a video showing their robot doing some pretty impressive stuff. We might be fast approaching a golden age of robotics and automation.
Another way to look at the problem posed by the previous post is that multimodality is a necessary step in order to improve LLMs performance. Fei-Fei Li talks about this in her latest TED talk, drawing a parallel with the appearance of vision in biology (the kind of comparisons I have a soft spot for, obviously)
This week we had the launch of two multimodal AI products by OpenAI and Google. While their demos were quite impressive, it is worth noting that models themselves don't appear to be much smarter than their predecessors. Some recent research is questioning how much more we can get from the current transformer architecture. Here is one of the papers recently released, alongside a video from Computerphile (one of my favourite Youtube channels).
Related to this topic, I’ve discovered the incredible network of MIT fab-labs, started by Neil Gershenfeld at the MIT Center for Bits and Atoms. I can’t believe there isn’t one in London! Dreaming up plans to start one right now.
I’ve been reading the latest book by Mario Carpo, Beyond Digital: Design and Automation at the End of Modernity. A fascinating exploration of how robotics and artificial intelligence will enable us to rethink making practices. I want to leave here a short quote, but I strongly encourage you to give it a read: ”Before the rise of modern engineering traditional societies lived in a world of physiocratic penury, where manufacturing and building where at the mercy of local supplies of materials and labor. Pre-industrial artisans did not have much of a choice: most of the time they had to make the most of whatever material, and labor, they had on site. These constraints were eliminated by industrial production, but—for a number of reasons, not limited to climate change—the environmental, social, and thermodynamical costs of of mass production and mass transportation are increasingly unwelcome. Today, in compensation, we can use computation and robotic labor to reenact some aspects of our ancestral artisanal economy, and recover at least some of its inherent, circular sustainability. Robotic forestry driven by machine vision is already relatively common, and experiments with scan-and-pick stone assembly, and computationally optimized carpentry, made of branches and forks as found, are ongoing.”
A brilliant and extensive article exploring evolution with a machine learning / computer science approach. It can get a bit technical, take it in chunks.
This installation enables a live plant to control a machete. plant machete has a control system that reads and utilizes the electrical noises found in a live philodendron. The system uses an open source micro-controller connected to the plant to read varying resistance signals across the plant’s leaves. Using custom software, these signals are mapped in real-time to the movements of the joints of the industrial robot holding a machete. In this way, the movements of the machete are determined based on input from the plant. Essentially the plant is the brain of the robot controlling the machete determining how it swings, jabs, slices and interacts in space.
I've been reading Anathem by Neal Stephenson. I've read about a third and I can't stop. Stephenson's work never ceases to captivate me: such a brilliant mix of plot development and philosophical ideas. I urge anyone who enjoys speculative fiction to pick up one of his books.
The linking together of principles from different disciplines especially when forming a comprehensive theory.
TempleOS (formerly J Operating System, LoseThos, and SparrowOS) is a biblical-themed lightweight operating system (OS) designed to be the Third Temple prophesied in the Bible. It was created by American programmer Terry A. Davis, who developed it alone over the course of a decade after a series of manic episodes that he later described as a revelation from God.
Moravec's paradox is the observation in artificial intelligence and robotics that, contrary to traditional assumptions, reasoning requires very little computation, but sensorimotor and perception skills require enormous computational resources. The principle was articulated by Hans Moravec, Rodney Brooks, Marvin Minsky and others in the 1980s. Moravec wrote in 1988, "it is comparatively easy to make computers exhibit adult level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility".
Apple might be starting to loose their tyrannic grip on their users through some better-late-than-never regulations. Spotify is not a nice company in my opinion, but I'm happy for them in this one. Hoping Epic Games can triumph in their crusade.
😍
ACI aims to understand the interaction between animals and computing technology within the contexts in which the animals habitually live, are active and socialise with members of the same or other species, including humans. Contexts, activities and relationships will differ considerably between species, and between wild, domestic, working, farm or laboratory animals. In each particular case, the interplay between animal, technology, and contextual elements is of interest to the ACI researcher
Given a path, computes the 3D shape ("trajectoid") that would follow this path when rolling down a slope.
Just a beautiful shot.
The world's first living coffin
'Today, the White House Office of the National Cyber Director (ONCD) released a report calling on the technical community to proactively reduce the attack surface in cyberspace'. New ways of thinking about sovereignty and security in the digital realm.
Spoiler alert: it has little to do with inflation reduction
Boids is an artificial life program, developed by Craig Reynolds in 1986, which simulates the flocking behaviour of birds, and related group motion. As with most artificial life simulations, Boids is an example of emergent behavior; that is, the complexity of Boids arises from the interaction of individual agents (the boids, in this case) adhering to a set of simple rules.
Due to twitter being twitter, I ended up watching Starship Troopers this week, a 1997 Paul Verhoeven film. Such a brilliant satire about the difficulty of conceiving cohabitation with non-human species and the fascist response that so naturally creeps up in these situations. It is the second time in the last couple of months where a science fiction film from the 90s (a decade I always considered quite superfluous when it comes to mainstream culture) surprises me with a profound message that is still quite relevant. The other film was Jurassic Park. I'm still trying to figure out if the polemic this week was a publicity stunt by Helldivers 2.
Deepfake is not very novel at this point, but what I find interesting is how AI technology will enable a whole new array of remixing tropes and referencing styles. I've seen deepfakes and generative AI used for nefarious political purposes and commercial stunts. What I'd like to see more is conceptual work using AI techniques. Not AI art, just concepts that you couldn't carry out without AI. This example is not it, but it is a step in that direction.
Great twitter thread with some striking art. I've been listening to a lot of Spinetta lately, he remains my favourite spanish speaking artist.
CES always brings a myriad of futuristic products and I almost always find them either unimaginative, impractical or plainly undesirable. But I like this one.
Really interesting take on the current economic dynamics. Varoufakis is someone I have many differences with but I always find his ideas interesting.
Excellent short story.
Sony released a Trailer for a free downloadable experience called 'Interaction isn't explicit'. It's an interactive essay/study with a lovely art direction. Interesting move by Sony, would love to see more like this.
Manfred Mohr is considered a pioneer of digital art based on algorithms. After discovering Prof. Max Bense's information aesthetics in the early 1960's, Mohr's artistic thinking was radically changed. Within a few years, his art transformed from abstract expressionism to computer generated algorithmic geometry.
If you've ever tried to learn coding, one of the first rules you learn is 'Don't Repeat Yourself': if you find yourself writing the same code over and over, you should find a way to automate it. Well, Tailwind is a framework used to give styles to websites, and does the exact opposite. It forces you to memorize a bunch of keywords and write them over and over. It's a very counter intuitive way to write code, but once you get the hang of it, its great. I've struggled with CSS for a long time, and Tailwind has made it so much easier. I think there's a deep beauty in the fact that we keep finding new ways to solve old problems.
"A Symbolic Analysis of Relay and Switching Circuits" is the title of a master's thesis written by computer science pioneer Claude E. Shannon while attending MIT in 1937. In his thesis Shannon proved that Boolean algebra could be used to simplify the arrangement of the relays that were the building blocks of the electromechanical automatic telephone exchanges of the day. Shannon went on to prove that it should also be possible to use arrangements of relays to solve Boolean algebra problems. The utilization of the binary properties of electrical switches to perform logic functions is the basic concept that underlies all electronic digital computer designs. Shannon's thesis became the foundation of practical digital circuit design.
Great article by Patricio Gonzalez Vivo, who's always ahead of the curve.
I've just learned that there is a whole set of computer vision models trained by people to perform all kinds of tasks. A testament to the importance of Open Source
The way in which Unreal Engine 5 has empowered creators to make this kind of content is amazing. Although I'm not surprised the game that inspired this was Factorio.
Very interesting group at the MIT Media Lab. For a long time I've explored the idea of making computing spatial, and I find it interesting how they have a similar goal but with a different approach.
A partnership between the University of Edinburgh, Alan Turing Institute, and Edinburgh’s Festivals to explore AI in the arts.
The Journal of Design and Science (JoDS) was a joint venture of the MIT Media Lab and the MIT Press that forged new connections between science and design, breaking down the barriers between traditional academic disciplines in the process. Although its no longer in operations, the six issues are available on their website.
Very interesting approach to designing products and a great overview of an ecological design practice
