Re:OpenAI chief goes before US Congress to propose licenses ...
by namelosw on Wednesday May 17, 2023 @12:56AM
(#35971413)
Sam Altman and OpenAI are playing a smarter game than Facebook. Facebook's reckless actions have raised alarms among the public and regulators, and have backfired on them, completely screwed their PR making it harder for them to do more damage.
OpenAI's strategy is to work with the regulators and shape the public opinion in their favor. He has earned enough trust to lower people's guard and give him more freedom in his next moves.
AI licensing is certainly going reduce their chance getting disrupted, given how surprisingly small those teams of innovation could be nowadays. (e.g. OpenAI, Midjourney had very few people when they made their breakthroughs)
And no, it's not going to stop AI being abused. Corporations will train their in-house AI, and illegal groups will train theirs, too (remember how small those teams are? Plus, specializing models for bad purpose with technologies like LoRA is not costly). Unless the authority is going to the dystopian point that also regulates who is buying a bunch of RTX 4090 and then inspecting what programs are running on the GPU in the basement.
Eventually, people would have to pay a fortune to those monopolies to protect them from new AI-powered abuses, instead of letting people themselves and startups who are willing to compete with reasonable price to arm-race with those abuses.
OpenAI's strategy is to work with the regulators and shape the public opinion in their favor. He has earned enough trust to lower people's guard and give him more freedom in his next moves.
AI licensing is certainly going reduce their chance getting disrupted, given how surprisingly small those teams of innovation could be nowadays. (e.g. OpenAI, Midjourney had very few people when they made their breakthroughs)
And no, it's not going to stop AI being abused. Corporations will train their in-house AI, and illegal groups will train theirs, too (remember how small those teams are? Plus, specializing models for bad purpose with technologies like LoRA is not costly). Unless the authority is going to the dystopian point that also regulates who is buying a bunch of RTX 4090 and then inspecting what programs are running on the GPU in the basement.
Eventually, people would have to pay a fortune to those monopolies to protect them from new AI-powered abuses, instead of letting people themselves and startups who are willing to compete with reasonable price to arm-race with those abuses.