Indeed, LLMs have that distintive and slightly revolting quality of friendly, helpful and inoffensive tone that reminds me of American customer service ideals. Also the user's profile is equally toeing this line of uncanny valley.
I guess the only way to identify ourselves as human is to adopt a measure of rudeness, some grammatical mistakes and snark that those stupid AIs still try to avoid at all costs.
I think it has to do with who trained them, and how, and not the technology itself. Just wait until the capability of training on large corpuses is more widely available…
I guess the only way to identify ourselves as human is to adopt a measure of rudeness, some grammatical mistakes and snark that those stupid AIs still try to avoid at all costs.