@ouguoc the qualitative evaluations i’ve read have been positive as well (except the one testing a Tiananmen Square prompt).

skepticism is always a good starting posture! i haven’t played with it, and would not in any case be competent to evaluate it.

but i guess i’d perhaps conspiratorially suggest the US incumbents really do not want AI to be “light”, that for example Sam Altman’s ambitions depend on it being capital intensive and difficult to contest. zirk.us/@interfluidity/1138733