Adobe Max, The Techno Optimist Manifesto and Nvidia's Eureka – Live and Learn #27
Welcome to this edition of Live and Learn. This one is absolutely packed because a lot has happened in the last 2 weeks. Adobe had its Max Conference, Nvidia is successfully training Robots within a simulation, and the biggest brain cell-type catalog has been released. And this is not even all… As always, I hope you enjoy this Edition of Live and Learn.
✨ Quote ✨
We trade freedom not for safety but for the illusion of safety.
– Packy McCormick - (source)
Links
Eureka: Using GPTs to train Robots by Nvidia. In this research, Nvidia successfully manages to code reward policies for robots and novel tasks automatically by using GPT-4. This makes policy-based reinforcement learning much easier than it has ever been before. They claim that this approach even outperforms human written Reinforcement Learning Policies on 80% of tasks. Nvidia is doing some crazy stuff with AI these days and if you haven't, you should also check out some of the other stuff they do, like the Isaac Gym or the Omniverse... What they are building with all of this will accelerate robotics and the way that we build smart machines tremendously. This truly is a "Eureka" moment.
Adobe Max Keynote. Another KeyNote by one of the big players in the generative AI field. Adobe has been integrating generative AI capabilities over the last few months into most of its signature apps. And they will continue to do more of the same. Image and video editing will change a lot over the coming months and years.
Largest Brain Cell Type Map Ever Made by Nature. The Human Brain is complex not only because of the millions of neurons interacting with one another but also because there are myriads of different cell types, supporting and helping each other to do their job. Not everything in the brain is neurons, and not even all neurons are the same. And now we have a map of this that can help us gain a better understanding of these different cells and how they relate to one another. The full source data can be found over here.
AGI is already here by Noema. In this thought-provoking piece, the authors argue that the systems that we currently have will be categorized as early versions of AGI in the future. Because the AI that we have right now is already general: It is excelling in many different tasks, many of which the original models have not seen before. Sure they are still weak in doing all of that, but so were the earliest general-purpose computers.
The Techno Optimist Manifesto by a16z. A16z is one of my favorite blogs, and this piece kicks it out of the park. It's about why and how we should be optimistic about the future and the prospects that technology offers. In a few pages, it sums up how the world should be and what is good and bad in it. Growth is good, energy is good, cheap prices are good, and technology enables all of that and more. I enjoyed reading this a lot and it reminds me of the work of David Deutsch, Richard Feynman, Freeman Dyson, and all the other people like that.
How to Eradicate Malaria by DeNovo. We have the means to engineer genetically altered mosquitoes, that can end Malaria once and for all with a technology called a "gene-drive". We had those means for years, but Malaria still exists... And this post tries to answer the question: Why? And then asks if we should simply “do it”, instead of hiding behind politics and regulations. It goes as far as hypothesizing how much it would cost to actually execute this idea and build the facilities necessary to breed gene-edited mosquitoes to eradicate Malaria for good.
Beautiful Places AI. These people build a dataset that only contains data about beautiful places. They make the case that we need more beautiful places in the world, and that AI can help with that. I find this super cool because beautiful things are one of the main reasons why I travel. And having that beauty cataloged so that machines can learn and understand what it means for something to be beautiful sounds awesome. Because this way we can also teach the machines to create more of it. And maybe we can even find new beautiful things this way. Designs and ideas that we haven't thought of before. Things like move 37 but in design. Beautiful things that no human has ever thought about before.
Using Humanoid (and other) Robots by Amazon. Not just Nvidia is doing crazy stuff with robotics. Amazon already uses over 750,000 across its different warehouses and operations. And now they started deploying and testing humanoid robots too. And in typical Amazon fashion, they approach everything from a practical point of view, always looking for ways to improve processes, automate work, and save costs. And turns out: humans are still there, doing jobs that previously didn't exist, amplified in their effectiveness and efficiency by the clever and strategic use of robots.
🌌 Traveling 🌌
🎶 Song 🎶
Apotheosis by Austin Winteroy
That's all for this time. I hope you found this newsletter useful, beautiful, or even both!
Have ideas for improving it? As always please let me know.
Cheers,
– Rico