Love the tread, note the same subject is also discussed elsewhere in talk.tiddlywiki
If you think about it, we have plenty of tools to manipulate the physical world, material science has helped us untap potentials of the physical world. Sometimes we even anthropomorphise our objects.
LLM’s are amongst the thinking tools, TiddlyWiki is one too, and we can make great use of them, if we do as we do today, and remain sceptical and critical. It is wonderful that LLM’s give us a language interface to computers, and we are gaining a new set of tools of incalculable value, but in this case we are even more likely to falsely anthropomorphise thinking tools, when they can talk to us.
- The first trap for identifying A.I. should it ever appear to exist, is our tendency to anthropomorphise.
- As I am sure you realise that it is quite different recognising downsides, and the proverbial “putting your head in the sand”. To me they may reasonably have “banned it from use to produce public communications”, but now they will not learn what it is good or bad at doing.
- Totally agree, calling the product of LLM’s “A.I.” is hype, incorrect and damaging to the investigation of A.I.
- When considering SETI and the possibility of life, intelegent or otherwise elsewhere I do think we need to be at least this “open minded”.
- Yes, it is just as confident making true and false statements
- I expect it to start parroting itself, in time as the training data becomes littered with its own output generated earlier
- But such things we should be sceptical of, always should have, and should be of our fellow humans. For example, a rational universe cannot be constructed from the utterances of Trump, no matter how much you want it to.