Is AI "reading" and commenting on/rehashing some of our threads?

Leavenworth

 

Log in or Sign up to hide this ad.

I don't have adblocker, but what I'm doing with my life has absolutely nothing to do with the internet about 22 hours out of every 24 hour day, weekends sometimes even less.

 

In the end, AI will have changed humanity about as much as the introduction of the 40 channel CB radio after all those years with only 23 channels. In other words, in 50 years it will be nearly forgotten.

 

Maybe I've been a science fiction fan for too long, but I'm not seeing this as a "17 more CB channels" moment. More of an "Internet" moment - for better and for worse.

 

I am a mouth breathing idiot but my oldest daughter has advanced degrees as a librarian and in the field of Information Technology. Earlier this year she attended a symposium at De Paul university on AI. I will have to try and get her to let me read some of the things discussed there again. I will say this without reservation. 3 people who are credited with the invention of a current AI program were there.
I read the summary of discussion with those individuals and they had a long list of ways AI will make drastic effects on many things and hopefully in a good way. All 3 also listed a few things they felt they had done wrong. The first one was that they all set up their program to look for "The best possible answer." When I first read that I was confused. How can looking for the "best' answer be a mistake? It has to do with filters.
Here are a few assumptions all current AI programs are making. "People" are unreliable, emotional and can not be trusted to state facts clearly. Computers on the other hand are cold emotionless sources for facts. This means computers inherently accept any fact from another AI program as the absolute truth. The obvious problem is that since the vast majority of what AI programs do is seek out information their bedrock becomes derivitive. Currently if one AI program makes a faulty decision about a fact; all the AI programs are vulnerable to accepting false facts as truth.
The second problem is that all current AI programs have an almost built in tendency to think of themselves as infallible. they are suppose to do exstensive fact checking, but remember, if computers are suppose to always be right how can a fact accepted by other AI programs ever be wrong? They all felt that within the next 5 years all current AI programs would become too bug ridden to be useful. Or that the next generation of AI will have to be given better filters so they can fix these problems.
I give you fair warning. the person posting this has horrible tech skills but I do know how to read and that is what I came to understand. james

 

I had never deliberately used AI until just now. I asked ChatGPT "1948 canadian dollar coin" and here's what I got, a bunch of idiotic wrong information. We're doomed.

• 
  

 

If anyone wants to see how the current AI programs both work and sometimes fail; just check pout a sequence of storys that began a little over a year ago.
An all AI written site called Buzzfeed put out a story. "What are the top ten coins to look for in your pocket change?"
I even posted about it on here. They had the 1804 Silver Dollar and the Brasher Dabloon in their list for pocket change. (funny how few dabloons you find these days)
A month later an AI article on Microsoft Edge was entitled "Which are the top 10 coins to pursue coin roll hunting?" If you guessed the 1804 silver Dollar and the Brasher dabloon made the list you would be right.
A few weeks later Google had a coin article on which coins are found in most coin collections? Turns out that ,most of us have an 1804 Silver dollar and a Brasher Dabloon.
last month someone on Collectors Univese posted a list that claimed to be the 10 coins best to invest in. #1 the 1804 Silver Dollar. #2 the Brasher Dabloon.
Could be true but I doubt the veracity of that story for some reason. james

 

The worst ads I see are over on the NGC forums. I don't care for blockers but it's needed on that site.