Re: OT: The AIs have it...

On 2/26/2025 7:22 PM, Rhino wrote:

 On 2025-02-26 6:50 PM, moviePig wrote:

On 2/26/2025 4:24 PM, Rhino wrote:

On 2025-02-26 3:45 PM, moviePig wrote:

>
Me (to Microsoft Bing):
>
   Which word in this question is misspeled?
>
>
Microsoft Bing (to me):
>
   The word that is misspelled in the question is "wieht". The correct spelling is "weight".
>
>

Congratulations on finally understanding that ChatGPT and its fellow AIs are quite capable of non sequiturs (answers that don't make sense in the context of the question). If you haven't already encountered it, you should also be aware that AIs can - and do - lie shamelessly.

>
To say that today's AIs "lie" is to deceptively anthropomorphize them. It's important to bear in mind that, despite the lights, nobody's home.
>
>

 I know the difference between a human being and an LLM (Large Language Model) you patronizing twat!!

Then perhaps you don't know the difference between a lie and an error...

The fact remains that some people will take the output of an AI as TRUTH and it is *NOT*. Humans created AIs and the AIs clearly have the ability to lie - and do so. I've seen it myself in live demonstrations.

I don't know what you've seen, but for an AI to actually *lie*, it would've had to be programmed to do so. i.e., to deliberately render "facts" it "knows" to be false.