Future Digest #1: Space infrastructure; bullet-proof gel; generative 3D models
Good morning ☀️
This is Definite Optimism, the digest from the future.
This weekly highlight brings you the most exciting developments across space, biotech and AI.
🚀 Space
Quantum Space, a start-up founded by former acting NASA administrator Steve Jurczyk, wants to build robotic outposts and spacecraft for use in space near the Moon. They recently raised $15m to build communication infrastructure in the region of space between regular Earth orbits and the Moon - called cislunar space.
Why it matters
It’s been half a century since Apollo 17 astronaut Eugene Cernan left the last footprints on the moon. NASA is aiming to bring the next crewed moon mission in about 2025 or 2026, finally re-achieving what they managed in the 70s with less computing power than an Apple watch. Building space infrastructure will help open up the space economy and make future missions safer.
🧬 Biotech
A new gel-like material made from talin, a protein found in human cells, could be used to create better bullet-proof clothing and protect spacecraft. The gel absorbs the impact of high-speed projectiles and, importantly, maintains its structure. This is unlike current materials which can be destroyed or damaged by the impact, making them less reusable. When struck by particles going even supersonic speeds, the gel can absorb the shock, and even preserve the particles.
Why it matters
Nasa currently tracks more than 27,000 pieces of space debris. There are many more smaller pieces of debris that are too small to track, but can still potentially cause huge damage to spacecraft because they are travelling so fast. This is a growing risk to spacecraft, making space travel more expensive and dangerous. New materials such as this gel could help protect spacecraft from some of these smaller pieces of space junk.
🤖 AI
OpenAI has open sourced Point-E, a generative AI system that creates 3D models from text prompts. The machine learning system generates point clouds, discrete sets of data points representing a 3D shape. Point-E consists of two models: a text-to-image model, which was trained on labelled images to understand the connections between words and visual concepts, and an image-to-3D model, which was trained on images paired with 3D objects.
Why it matters
You will have most likely noticed that “generative AI” is having a bit of a moment. 2D images and text have been key areas of progress, but 3D images are critical across a huge range of industries. 3D images are more complex than 2D images, so take skill and computing to create. This new AI can create models in just a couple of minutes on a standard Nvidia graphics card. It can only create quite basic models at the moment, but we have seen previously how fast these AI models advance.
I hope you enjoyed this digest of the future this week. Let me know what you think by hitting reply or one of the feedback buttons below.
Until next week
Jamie
Did you enjoy this week's newsletter? Help me to improve!
Click on a link to vote:
Thanks for your feedback!