GTPT

Outright wrong !?

Tiny differences matter – you want to look at Chat GPT and a typo brings you to chat about GTP, of course no problem as you know everything about Guanosine-5′-triphosphate (GTP), this tiny building block which plays an important role in transcribing the synthesis of RNA.

Well, and this is the problem, where we should not really worry primarily about Chat GPT (or google’s bard or …) which may get things reasonably correct as long as we (a) apply the correct spelling, (b) do not expect really clear replies and are happy with answers that are as concise as the horoscope or the statements of any other oracle and (c) come up with simple issues . It is so advanced now that occasionally you may end up with an honest answer:

Sorry, I cannot answer this question. I am an AI-tool.

Yes, there are things where HI, i.e. human intelligence is needed: contradictory, emphatic, combative …

More worrying is the AI-imposed formulation of questions. Embedded in our word processor, or available as AI-writing support like Outright, QuillBot, Grammarly …

Such applications may be helpful at times …, and at first glance. But then the tiny differences may make a big difference:

This is really only the first draft of the initial script and we appreciate very much feedback by the readers. 

This is really only the first draft of the initial script and we appreciate very much feedback by the readers.

This is really only the first draft of the initial script and we appreciate very much feedback by the readers.

Or in Italian

Sono

Io sono

Sono anch’io

Io sono anche

AI comes up and autocorrects a sentence in different ways, sometimes possibly more elegant, and possibly “correcting” something that is actually wrong … because it is new: a new term, concept, … a searching for an “unknown unknown” — yes, borrowing from a militarist like Rumsfeld (MI?) may occasionally be more intelligent than relying on AI.

Intelligent or not … enjoy Christmas … or look forward to Spring Festival while others are celebrating …