Post #4232233

I think for me AI research is so super interesting because it is one of the very few fields that has genuine science left in it. Like materials science, for example, has degraded to a phase of almost pure trial and error through simulations. It is a field in which all major breakthroughs have been achieved, at least the low-hanging fruit ones. The high-hanging major breakthroughs are either not there or they are not funded because no one wants to give money to people that just goof around half the time. Materials science is too mature for major breakthroughs that occur from individuals who know the field deeply and just sit there and think and dream for a while, instead they try to build simulation data centers and autonomous laboratories to increase their throughput of brute-force experiments. The AI field is a little bit like physics in the 1920s... it is primarily people that sit there and think how to bend the laws of information and latent spaces and then they go and try something out no one has ever done before. There is for now no simulation available that can predict what this idea might yield, it is just human-driven science that has a lot of amazing low-hanging fruit to be picked. I am quite happy to be born in a time that has basically created a new field of science that is rapidly progressing with reality-shattering breakthroughs every few weeks. If the world had UBI, I would not even be anxious about all of that. #llm #ai #ubi #science