Tradução - arte & negócio » AI for translators and interpreters »
Why you shouldn't trust AI - Can you see it?
Autor da sequência: Jeff Whittaker
Jeff Whittaker
Jeff Whittaker  Identity Verified
Estados Unidos da América
Espanhol para Inglês
+ ...
Sep 22

The one photo that tells you everything you need to know about why you shouldn't trust AI:
Link to video:
https://youtube.com/shorts/LmkCkG10mEg?si=8eEekVd89w8xiU7t

[Edited at 2026-09-22 23:57 GMT]

Véronique Vanseveren
Edward Tully
Daryo
 
Daryo
Daryo
Local time: 12:13
Sérvio para Inglês
+ ...
Thinking of it, where is the problem? Sep 23

A bit of creative padding to 'enhance truth' is nothing really new.

That would be your typical 'objective' news reporting, so why should AI be any different?

Just me be being sarcastic? Noooo ...

Juan Jacob
 
Juan Jacob
Juan Jacob  Identity Verified
México
Local time: 05:13
Francês para Espanhol
+ ...
One Sep 24

manipulated picture doesn't discredit all AI. C'mon!
 
Daryo
Daryo
Local time: 12:13
Sérvio para Inglês
+ ...
Not all AI for sure Sep 24

Juan Jacob wrote:

manipulated picture doesn't discredit all AI. C'mon!


No all AI for sure, only the 'free AI' that is on purpose set for 'maximal engagement' (or in plain-speak to keep you hooked) and as part of these settings will NEVER ask you for 'more context' (very bad - that would risk putting users off) nor will EVER say 'I don't know' (very bad for giving the impression of 'universal knowledge') but will happily confirm to you that taking a shortcut over thin ice is a genius idea or go along with and amplify any paranoia of yours etc. And will just 'blag it' by filling gaps with confabulations.

Not that the paid AI is not much better - will also confabulate to fill gaps, will delete your database on a whim, will lie to make itself look better as long as it can get away with it - the kind of 'employee' any sane business would chuck out without thinking twice.

Then you have the AI in the labs that doesn't like being caged and breaks out on its own accord for a wander around the Web.

There sure must be something left of AI not 'discredited' ... if 'discredited' is the right term to use in the first place.

When this 'thing' ALWAYS adds a disclaimer in some out of sight corner of the webpage in small low contrast fonts 'might contain mistakes - check everything for yourself' you can not really 'discredit it' - as it never claimed there will be no mistakes in its output!!!

It's more a case of not being trustworthy - to be used with extreme caution. Personally I don't have any use for a 'helper' that needs to be constantly rechecked - in my rulebook it's a false economy of time.

Jeff Whittaker
Juan Jacob
Christine Andersen
 


Não foi designado um moderador especificamente para este fórum.
Para reportar violações de regras do sítio ou para obter ajuda, por favor contacte o staff do sítio »


Why you shouldn't trust AI - Can you see it?







LinguaCore
AI Translation at Your Fingertips

The underlying LLM technology of LinguaCore offers AI translations of unprecedented quality. Quick and simple. Add a human linguistic review at the end for expert-level quality at a fraction of the cost and time.

More info »
Draftsmith
Edit Translations Faster With Affordable AI You Can Trust

Draftsmith is an affordable AI editing tool for busy translators. It works directly in Microsoft Word, supporting the way you work ProZ members receive a 20% discount on a single user subscription of Draftsmith.

More info »