it’s not the “making one” that’s a problem. it’s the making, optimizing and rabid marketing of one in the service of capital instead of humans.

if only a bunch of open source, true non-profits released language models, the landscape might still suck but would be distinctly less toxic.

and if the government (or even a decently sized ngo standards entity) had worked proactively with computer scientists to find solutions like watermarking, labor replacement protections, and copyright protections, things might be arguably perfect. not one of those things happened and so further into the hellscape we descend.