[ExI] More thoughts on sentient computers

[BillK]

On Mon, 20 Feb 2023 at 18:15, spike jones via extropy-chat
<extropy-chat at lists.extropy.org> wrote:
>
> Eh, I would agree even without the qualifier of being checked by two different humans.  I am old enough to remember when calculators were new, some geezers would check the calculators by hand.  Now we don't trust human calculations at all.  We already know computers don't make mistakes.
>
> The Vanderbilt groveling apology is really a statement that in principle, a Chatbot can never feel human empathy, even if it writes a beautiful urology.  Well, hmmm, it causes me to wonder if we humans really feel human empathy.  We know actors can fake passion even for someone they don't like (top-billed actors generally don't like each other (because they are in a sense competitors (but they do act out some great on-screen sizzle.)))  So what they are doing is artificial human feelings, ja?  We pay good money to see that, while we turn our eyes away from the real thing when we see it in public.  How many of us here have never faked giving a damn?  Let he who is without a trace of artificial emotion cast the first stone.  Hey, not at me dammit!  I meant cast the first stone at... some other intermittently-phony bastard.
>
> I don't like Vanderbilt's claiming that chatbots can never feel human empathy.  That notion is human-prejudice, analogous to sexism.  If the term hadn't already been used for something else, I would call it humanism.
>
> spike
> _______________________________________________


Many writers are reporting that ChatGPT output needs a tremendous
amount of editing and correction.
Not for spelling or grammar, though. It is fine with that.
The problem is that ChatGPT makes stuff up. Very realistically sounding stuff.
Writers have to check that ChatGPT is saying what they want, not
missing anything and not adding irrelevant items.
It is only an aid at present, not something to have tasks handed over
completely.

BillK