This is an eye-opening long read for anyone who works with watches and AI, and suggests the very real possibility that we are seeing the destruction of informative and accurate research, quicker than you think.

LLMs that argue with you…

LLMs that argue with themselves…

LLMs that hallucinate details that don’t exist…

If you think that AI can be trusted, or can compare images accurately, this article may make you think again.

‘AI often weighs evidence by volume rather than quality or relevance. It surfaces the narrative with the most coverage and presents it as a finding.’ - a real danger if you work with rare, obscure, or nuanced watches.

‘AI learns from published text, and in our field/hobby the most authoritative published text is often the auction catalogue. But the catalogue record was never clean and will likely be polluted much more in the future with AI slop research and automated tools.’

If AI scrapes information available online, then, before long, most of the information available to scrape will be AI-generated, and we end up in a death spiral of unverified nonsense.

And to those that believe AI is the quick and easy option, ‘Stanford researchers tested paid, purpose-built legal research AI tools sold to professionals for their accuracy, and found they hallucinate on 17 to 34 per cent of queries.’

Tell me, how does your paid-for, purpose-built watch authentication tool fare? Do you know?

But… as @Kingflum says, ‘AI is genuinely useful for gathering catalogue data, finding links, keeping references apart, spotting inconsistencies and organising findings. There is real value. But the conditions are simple: every claim must be verified, and verification is still a human job.’