3 ms·

> This seems like a terrible idea, there is no reason for public involvement here, also there is no stated or allocated budget. While I dunno where I come down

> This seems like a terrible idea, there is no reason for public involvement here, also there is no stated or allocated budget. While I dunno where I come down yet on this legislation, it seems truly bizarre to object to the CalCompute cluster. IMO, the biggest AI threat we face is from big tech being the only ones capable of creating new models. Right now, companies like Meta are aligned with open source AI, but that could change at the drop of the hat. It makes sense to preserve state capacity to level the playing field against attempts at monopoly or other forms of overreach. Also, a lot of great work is coming out of UC system labs, so why shouldn’t we invest more in making them successful?

sroussey 3y ago
I wonder if we would have smaller llama models if compute was cheap.
rsync 3y ago
"While I dunno where I come down yet on this legislation, it seems truly bizarre to object to the CalCompute cluster." The calcompute cluster is what jumped out at me. The danger is that this will be a huge subsidy to entrenched UC/CA contractors that are in place solely to navigate - and exclude others from - state procurement mechanisms. In fact, if you were really cynical you might think that the entire purpose of this bill is the calcompute cluster and everything else is just scaffolding to enable a boondoggle. I am not asserting this - I am just pointing to a possible interpretation.
- bugglebeetle 3y ago
  Can you elaborate? I’m not sure I follow.
  - sgslo 3y ago
    Who's going to build the cluster? Who's going to operate it? Who's going to decide how time/access gets allocated? Are results of compute going to be in the public domain? Given the glacial speed of government products, will the cluster be using obsolete tech by the time its finally up and running?
kelnos 3y ago
I absolutely agree, but I'm not sure something like CalCompute is going to do a good job counteracting that. Is the major limitation to others creating new models the availability and cost of compute, or is it the fact that big tech companies can afford to compensate AI researchers well enough that they won't work on open projects? To me it feels like the latter, but admittedly I don't know for sure.

sroussey 3y ago

I wonder if we would have smaller llama models if compute was cheap.

rsync 3y ago

"While I dunno where I come down yet on this legislation, it seems truly bizarre to object to the CalCompute cluster." The calcompute cluster is what jumped out at me. The danger is that this will be a huge subsidy to entrenched UC/CA contractors that are in place solely to navigate - and exclude others from - state procurement mechanisms. In fact, if you were really cynical you might think that the entire purpose of this bill is the calcompute cluster and everything else is just scaffolding to enable a boondoggle. I am not asserting this - I am just pointing to a possible interpretation.

bugglebeetle 3y ago
Can you elaborate? I’m not sure I follow.
- sgslo 3y ago
  Who's going to build the cluster? Who's going to operate it? Who's going to decide how time/access gets allocated? Are results of compute going to be in the public domain? Given the glacial speed of government products, will the cluster be using obsolete tech by the time its finally up and running?

kelnos 3y ago

I absolutely agree, but I'm not sure something like CalCompute is going to do a good job counteracting that. Is the major limitation to others creating new models the availability and cost of compute, or is it the fact that big tech companies can afford to compensate AI researchers well enough that they won't work on open projects? To me it feels like the latter, but admittedly I don't know for sure.