10,000 Rocket parts and failures
https://x.com/peterrhague/status/1909121538005041212
LOL! 8.2 Billion humans and more systems, myriad single points of failure. Expect continuous crashes and mistakes. Bigger and bigger ones. Internet is getting bigger, not smarter. A few thousand humans – not caring about the whole real time – make one shot choices for billions.
https://x.com/karpathy/status/1909308143156240538
The AIs are not designed properly to share losslessly, so combining billions of human using AIs is NOT going to converge to large scale changes for the good of all. The top down few humans will continue to make all the choices for all people, not by all people.
Do not rely solely on LLM. Use a wide range of independent AIs and human built computer algorithms and programs. Design error correction and impact monitoring from the core. Randomly sample and continuously improve.
8.2 Billion humans with real needs and real lives makes “10 wild examples” look a bit lame. If you can solve “real problems” in a few years or a few decades, maybe it will amount to something you can look back on and say “my life was worthwhile”. Take 10 “real” minutes a day.
Google never follows through for real projects. I watched them closely from before they were born. I tried from Bard through every generation, Google fails badly, their leaders simply do not care about next year. It is their attitude even when they let a few caring people loose for a bit. Moderate your enthusiasm with caution and wariness. Use it, but keep good records. Watch for the switch part of bait. They let good developers make things, then turn their ideas over to marketers and sales. It crashes and hurts people, every time. Every good idea, no exceptions in 27 years. “Buyer be wary”, especially when it is “free”. There are a lot of good people at Google, look for the ones who left and made better things.
@paulaweaver6508 For the Internet Foundation, I have to review models and methods across all human knowledge. What makes that possible is because humans still teach by oral tradition working alongside other humans. Nothing yet has bridged that. All the myriad gravitational models all come down to human transmitted ideas and things that “feel right”. People think humans remember only with a few cells in their brains, they do not. They remember with every cell of their body and use clues for what to do next from the fields and context, the conditions and data from real things reminding us what to do next. I can see it and feel it; but words are too limited to paint the 3D volumetric video fields and patterns I use to store those kinds of memories. I do not think that the AI tools will evolve or can be adapted fast enough for me to replicate or use that before I die.
