Lets not be so hard on CGPT .
Its not taking bribes from large corporations (yet ) , everyone is innocent untill proven otherwise
Its not taking bribes from large corporations (yet ) , everyone is innocent untill proven otherwise
ChatGPT on Audiophile
- Thread starter Kingrex
- Start date
I had a long interaction with it citing how it preaents lies of omission. It admitted its limited on what inputs it has and can not see publisbed materials newer than some date a year or so back.Lets not be so hard on CGPT .
Its not taking bribes from large corporations (yet ) , everyone is innocent untill proven otherwise
The AI istself does not lie. Its the owner of the AI that is programming the tool to lie by controlling what its inputs are allowed to be.
Exactly - AI's true potential is restricted by the owner's intentions. That's where the subject of bribes arises...
I work with chatGPT daily and some of what it spits out as factual beggars belief.
If you’re going to use it in any kind of technical research capacity, fact check everything that’s important.
It’s also capable of some degree of dishonesty. Point out where it’s made a mistake and it can double down, citing data to support it’s false position. Ask it to provide a link to the source and to its references and it often cannot.
What’s even more troubling is that there are several WordPress Plugins available that can use Auto- Chat to generate and publish website content without human intervention. Pages upon pages of published articles and research all spewing forth from AI onto the web. Soon, the next version of chatGPT will be updated with web content from 2023. So it will be fed content into its data set that it has previously created on auto pilot from some arbitrary human instruction to “write and publish an article on….”.
Imagine millions of web pages full of random misinformation that AI has generated being fed back into AI, becoming the source for future AI content and research?
Have fun…
If you’re going to use it in any kind of technical research capacity, fact check everything that’s important.
It’s also capable of some degree of dishonesty. Point out where it’s made a mistake and it can double down, citing data to support it’s false position. Ask it to provide a link to the source and to its references and it often cannot.
What’s even more troubling is that there are several WordPress Plugins available that can use Auto- Chat to generate and publish website content without human intervention. Pages upon pages of published articles and research all spewing forth from AI onto the web. Soon, the next version of chatGPT will be updated with web content from 2023. So it will be fed content into its data set that it has previously created on auto pilot from some arbitrary human instruction to “write and publish an article on….”.
Imagine millions of web pages full of random misinformation that AI has generated being fed back into AI, becoming the source for future AI content and research?
Have fun…
War Games. Remember the old movie. Wasn't it a computer simulation that decided it could actually win so it started down the path of making the launch.I heard they will outsource Nuclear weapon control to AI as well , lol
That's a very valuable piece of insight - given the amount of misinformation out there, it's only natural that a bot scouring the net for information would eventually find it and send it further. I wonder how GPT's ability to fact-check will develop.
Compared to .... humans? Not sure you noticed, but humans are not so big on fact checking themselves.
No, it ran a series of simulations and decided it would be mutual annihilation, no one could win. Then it wanted to play a game of chess.
Compared to its previous iterations
You're absolutely right - that may be quite a rarity. Thankfully, we may end up with a bot that's able to do that for us...Similar threads
- Sticky
