“The most beautiful thing we can experience is the mysterious. It is the source of all true art and science. He to whom the emotion is a stranger, who can no longer pause to wonder and stand wrapped in awe, is as good as dead —his eyes are closed. The insight into the mystery of life, coupled though it be with fear, has also given rise to religion. To know what is impenetrable to us really exists, manifesting itself as the highest wisdom and the most radiant beauty, which our dull faculties can comprehend only in their most primitive forms—this knowledge, this feeling is at the center of true religiousness.” Albert Einstein
Solving problems in one domain produces solutions for others. Leonardo Da Vinci designed weapons for a living. I am not advocating for war tech, just noticing that throughout history it has been behind a lot of technology that may not have been developed otherwise. DARPA wasn't just the internet - also chips, computer interfaces, compute clusters, etc..
What can a neuron compute?
Real biological neurons are complex, but how capable are they?
Using a new method, we found that a single cortical neuron can classify cats vs dogs, recognize spoken words, and solve 10-bit parity, all tasks thought to require entire networks. (1/15)
Everyone building programs with AI needs to now:
1. Be able to swap out models, in short order.
2. Be able to verify, regularly, that your output isn’t silently changing.
(You should have been doing this for any production system, with evals, but now there’s no excuse.)
I can't talk to Claude about anything - my 'memories' are full of biological and quantum research and it keeps pulling them in and demoting the model. This feels like the start of something dark under the figleaf of 'safety'. I have also been discombobulated/slightly depressed all day. This is just going to drive people to other model providers. It is so short sighted.
There is a vast amount of open biological data. It all has value in building models of biology. The paywall/private moat does not exist for transformative models in biology - the only value in the private/paywalled data are the 'labels' for the application of the models to particular diseases etc.