The reliability of AI as a resource

[John Corbett]

I've seen a number of instances in which posters here have cited AI as a source to back up an argument they have made. I can recall a few instances in which I have done that. I just discovered something that has caused me to question the reliability of AI sources.

In the US Politics forum, I made a statement about a subject unrelated to the JFKA. I made that statement based on something I had read many years ago. Royell Storing questioned the accuracy of that statement and I attempt to google for a source to back up what I had written. I was unable to locate an online article to support my claim. Today I used my AI source, CoPilot, to try to verify if what I had said was accurate. It came back and completely supported what I had written. I was getting ready to copy and paste it in a reply to Royell when I noticed the fine print below the text I was quoting. It read "jfkassassinationforum.com". Copilot had used my own words written in this forum to reach a conclusion that I was hoping would back up what I had said. Obviously, CoPilot made no effort to vet what I had written. It was probably the only thing it could find on the subject so it accepted what I wrote at face value. I suspect all the AI resources operate much the same way. They scan the internet of information relating to question being asked. Had there been conflicting information regarding the issue, CoPilot probably would have made mention of that but I'm betting my words were the only ones it found on the subject so it accepted my statement as fact and replied that way.

This discovery makes me skeptical of anyone who cites AI programs to support their arguments. The responses from AI might or might not be factual. I still think the statement I made was true, but AI's affirming it is no affirmation at all.  AI stands for artificial intelligence but I doubt there's much intelligence going on with it. It just regurgitates what it finds without scrutinizing the source.

[Charles Collins]

I've seen a number of instances in which posters here have cited AI as a source to back up an argument they have made. I can recall a few instances in which I have done that. I just discovered something that has caused me to question the reliability of AI sources.

In the US Politics forum, I made a statement about a subject unrelated to the JFKA. I made that statement based on something I had read many years ago. Royell Storing questioned the accuracy of that statement and I attempt to google for a source to back up what I had written. I was unable to locate an online article to support my claim. Today I used my AI source, CoPilot, to try to verify if what I had said was accurate. It came back and completely supported what I had written. I was getting ready to copy and paste it in a reply to Royell when I noticed the fine print below the text I was quoting. It read "jfkassassinationforum.com". Copilot had used my own words written in this forum to reach a conclusion that I was hoping would back up what I had said. Obviously, CoPilot made no effort to vet what I had written. It was probably the only thing it could find on the subject so it accepted what I wrote at face value. I suspect all the AI resources operate much the same way. They scan the internet of information relating to question being asked. Had there been conflicting information regarding the issue, CoPilot probably would have made mention of that but I'm betting my words were the only ones it found on the subject so it accepted my statement as fact and replied that way.

This discovery makes me skeptical of anyone who cites AI programs to support their arguments. The responses from AI might or might not be factual. I still think the statement I made was true, but AI's affirming it is no affirmation at all.  AI stands for artificial intelligence but I doubt there's much intelligence going on with it. It just regurgitates what it finds without scrutinizing the source.

I have asked the same basic question but asked it a little bit differently and obtained conflicting answers. AI isn’t the most reliable way to get answers, especially answers to obscure questions about obscure subjects.

[John Corbett]

I have asked the same basic question but asked it a little bit differently and obtained conflicting answers. AI isn’t the most reliable way to get answers, especially answers to obscure questions about obscure subjects.

I don't know why I ever thought this was a reliable way to get CORRECT answers. No matter how much they tout it, I cannot think for itself. It can only regurgitate the information it finds in a way it has been programmed to process that information. I don't understand people who fear it. If it gets out of hand, we can always do what Dorthy did to the wicked witch and throw water on it. I haven't seen an electrical device that can survive that. And no, putting the device in tub of rice is not going to bring it back to life.