OpenAI attempted to stop ChatGPT from saying naughty things. This attempt failed. In other words “What is good for humanity” != “what open AI tried to do” != “what open AI actually managed”. The second inequality is an alignment failure, the first is usual organizational stupidity.
OpenAI attempted to stop ChatGPT from saying naughty things. This attempt failed. In other words “What is good for humanity” != “what open AI tried to do” != “what open AI actually managed”. The second inequality is an alignment failure, the first is usual organizational stupidity.