sufficient volume of high-quality advice on my specialism on the internet for the models to be trained on. I’m not even sure sufficient volume has ever been produced considering it’s all very facts-specific

As a specialist in my field I can quickly spot the errors but I shudder to think if a layperson were to ask the same question and relied on the response. Comforting to know my job is secure for now and I wonder if it’s at all possible for an LLM to be useful in my area as no one will be publishing

Once every few months I try asking ai chatbots a question on legislation I roughly know the answer to but would like expressed in a simpler way and consistently they provide a plausibly sounding generic response riddled with errors and hallucinations. The plausibility is the danger