First, we failed to deliver our vision of AI-assisted programming because we were 10+ years too early to market, i.e. the tech is not ready yet.
We built the most-advanced AI for helping developers, but it fell short of the 10× improvement required to break through because today’s state of the art for ML on code is not good enough. You can see this in Github Copilot, which is built by Github in collaboration with Open AI. As of late 2022, Copilot has a number of issues preventing it from being widely adopted.
The largest issue is that state-of-the-art models don’t understand the structure of code, such as non-local context. We made some progress towards better models for code, but the problem is very engineering intensive. It may cost over $100 million to build a production-quality tool capable of synthesizing code reliably, and nobody has tried that quite yet.
Nonetheless, we could have built a successful business without 10×’ing developer productivity using AI, and we did not do that.
We failed to build a business because our product did not monetize, and it took too long to figure that out."
Reimagined NASA’s Space Math curriculum (https://spacemath.gsfc.nasa.gov/) as an immersive game, instead of static PDFs. Students solve the same problems real scientists face daily -- calculating orbits and trajectories, dealing with space weather, etc. -- in an interactive way that goes beyond worksheets (or just sending PDFs to an LLM).
https://space-math.academyhttps://www.github.com/dynamicwebpaige/space-math
Powered by Gemini for storytelling and text-to-speech (TTS). Links to the GitHub repo and live link that you can play above, thanks to Google Cloud Run. Please file feature requests, if there's enough interest will add more missions and a leaderboard.