As a sibling mentions, we have this windfall of cheap GPUs in the economy. They are not used to mine crypto anymore. So we can put them to use to ... train ML m

As a sibling mentions, we have this windfall of cheap GPUs in the economy. They are not used to mine crypto anymore. So we can put them to use to ... train ML models. This can still be more productive than crypto was. But hopefully the ML models trained won't be used for face recognition in dictatorships or for finally replacing all software engineers with a beefed up version of GPT.