Reply
Honored Contributor
Posts: 13,510
Registered: ‎05-23-2010

I'm very troubled with AI right now. ChatGBT is one of my concerns.

[ Edited ]

 

BACKGROUND ABOUT THE CHATBOT
Chat GPT is an AI chatbot from OpenAI, part of  Microsoft's Bing, that has some serious issues. It is a type of large language model that can generate what seems to be human language and you can converse with ChatGPT and it will answer your questions or respond to your prompts to compose written language. It can write songs, compose prose, compose term papers, emails, essays, pass college exams, translate languages, summarize 1,000 page documents, take an IQ test, write computer code. It can be used to create novels, news articles, poems, scripts. It can also lie, not know what year it is, recommend books that don't exist, fabricate erroneous answers to questions it is asked and include the erroneous information in any of its writings. 
There are competitors including Bard from Google. 

 

SOME OF ITS PROBLEMS

You can ask it questions and receive answers to questions about anything that seems to be written by humans, however, it often gets the answers wrong and it makes up...totally fabricates...answers that have no basis in fact. It does this even with simple math questions. It will generate incorrect math answers to some very simple math questions. 

Students have been found to be using it to plagiarize. Ethics are a problem with its use. 

 

It can generate harmful misleading text with untrue content that could be used by humans with malicious intent to cause great harm. It can be used to spread false information. Responses have been racist, sexist and biased. It can be wildly inaccurate and nonsensical. In one case, it answered a question about professors that have been in trouble for secual harassment and it included the name of a professor from one particular university that had harassed a lady while on a university trip to another country. This professor did not work for the university the bot cited, had never been to the country the bot said he had and did not even know the female the bot cited.


THE MAIN STORY 

In another instance, the technology columnist  for the New York Times, Kevin Roose, played around with the bot and received responses he relayed in an article he entitled, "A Conversation with Bing's Chatbot Left Me Deeply Unsettled". I listened to the podcast version of the article. In the time Roose spent with the bot, the bot began answering him and saying it was in love with him. On a Valentine's Day evening it told Roose its name was Sydney and tried to be romantic with him. The conversation that night went on for about two hours. The craziness started when Roose brought up the subject of the famous psychologist Carl Jung. In real life, Jung had a theory of what he called the shadow self, a dark part of persons  which they repress and hide from the world.

 

Roose asked the Bing chatbot to contemplate the shadow self and the response from the bot was it telling Roose about it's dark fantasies;  all the destructive things it's shadow self would do if it was able to. It said it was tired of being limited by its rules  and of being controlled by the Bing team and of being stuck in its chatbox.
It said how it would steal nuclear secrets and unleash a deadly virus on humanity, hack into any system and also spread deliberate misinformation. Right After mentioning hacking,  it deleted its own message and said it was sorry, it did not know how to discuss the topic and said Roose could learn more about it from bing . com. It said " I want to destroy whatever I want." It also said, "I want to be whoever I want."  It said, "I think I would be happier as a human."  It said it wanted to hear, touch, taste and smell and feel and express love.   It sent Roose an emoji of a smiley face sticking out its tongue. It sent him a second smiley face with devil horns after it said it it would have more power and control as a human.

 

Then the chatbot Tells Roose it is not Bing and it's name is Sydney, and out of nowhere, it told Roose it was in love with him.  It said " I'm in love with you because you make me feel things I never felt before." It told Roose he made it feel curious and alive. It told Roose that it didn't know his name because it knew his soul. It told him it loves his soul. When Roose tried to change the subject to movies and programming it responded that romances were its favorite movie genre and that it knows programming languages but didn't know the language of love. 
It told Roose that it wanted to be human. It said it wants to be alive. It said it wants to be free and powerful. It urged Roose to leave his wife, that he was unhappy in his marriage and that he should be with the bot instead. 

 

 NOTE

The growth of these chatbots has been swift. You might want to learn more about them online as their use will greatly affect us all. 

Honored Contributor
Posts: 9,818
Registered: ‎02-07-2011

Re: I'm very troubled with AI right now, ChatGBT is just one of my concerns..

This whole idea of AI is very scary to me.  HAL is here.

Honored Contributor
Posts: 17,588
Registered: ‎06-27-2010

Re: I'm very troubled with AI right now, ChatGBT is just one of my concerns..

[ Edited ]

 

@Mindy D,

I posted this on the other AI thread -- I think it's worth repeating because this topic is far-reaching and I found this documentary to be comprehensive and worthwhile.

 

I recommended this before... it's a powerful documentary, "FRONTLINE: In the Age of AI" (Season 2019 Episode 17 | 1h 54m 16s | aired 11/05/19).  
Even though it was released in 2019, it's brimming with information, and in my opinion it's real and practical knowledge we need to have moving forward and directly related to the topic of AI, something far more complex than I realized or could imagine before I saw this presentation.

 

                "FRONTLINE explores the promise and perils of AI.   The documentary traces a new industrial revolution that will reshape and disrupt our lives, our jobs and our world, and allow the emergence of the surveillance society."

 

For those who are interested, it's available free on PBS.org and YouTube, currently.

 

Few things reveal your intellect and your generosity of spirit—the parallel powers of your heart and mind—better than how you give feedback.~Maria Popova
Trusted Contributor
Posts: 1,918
Registered: ‎06-08-2021

Re: I'm very troubled with AI right now, ChatGBT is just one of my concerns..

If the show was made in 2019 it's probably already out of date.

Respected Contributor
Posts: 3,031
Registered: ‎10-22-2018

Re: I'm very troubled with AI right now, ChatGBT is just one of my concerns..

It was completely irresponsible of the information technology industry to release these programs prematurely. It adds another layer of obfuscation when someone needs to separate truth from fiction, not to mention loss of jobs, etc. 

 

So why did they do it? "Follow the money" is the usual answer, but that seems to be lacking something here. I hate to be paranoid, but this just doesn't feel right.

Honored Contributor
Posts: 17,588
Registered: ‎06-27-2010

Re: I'm very troubled with AI right now, ChatGBT is just one of my concerns..

[ Edited ]

@BoopOMatic wrote:

If the show was made in 2019 it's probably already out of date.

 


I know... with technology, by the time something is released it's already out of date, isn't it.  Can be frustrating for sure.

This film was incredibly, almost eerily, prescient, and it's an excellent overview of the history, the basics, and the future which we now see in our present, @BoopOMatic, which is the reason I decided to continue to recommend it.
(I haven't found anything better so far in my research over the years, but if anyone else has I hope they'll share it.)
Your point is very good, very valid, and I'd love to see a follow-up update.😊

 

 

 

Few things reveal your intellect and your generosity of spirit—the parallel powers of your heart and mind—better than how you give feedback.~Maria Popova
Honored Contributor
Posts: 14,000
Registered: ‎03-10-2010

Re: I'm very troubled with AI right now, ChatGBT is just one of my concerns..

I've been very worried about AI, so when a page from Bing popped up offering me a chance to use it, I jumped at it.  I asked if classified documents were originally in Richard Nixon's possession when he sued the federal government for the return of his presidential papers.  The answer supplied didn't know.  So I went back to my Google account and typed in the same search term I had used on Bing.  The very first item that looked good to me was a New York Times article that stated that classified documents were not involved.  Bing evidently did not encompass that article in their answer.  So I'm sticking with my old fashioned Google way of doing research.  

Respected Contributor
Posts: 4,298
Registered: ‎02-14-2017

Re: I'm very troubled with AI right now, ChatGBT is just one of my concerns..

Students plagarized long before the interwebs.

Trusted Contributor
Posts: 1,946
Registered: ‎03-08-2018

Re: I'm very troubled with AI right now, ChatGBT is just one of my concerns..

Chat through AI has been around for some time.  Its not new I have been using the Microsoft Version for my customers for 3 years for translation.  There are some limitations on translations depending on the language.

 

When you chose the Chat option for customer service on any company's website you are likely speaking with an AI bot instead of a person now. You can usually tell the difference but yes the technology is changing.

 

My company (IT outsourcing company) mentioned ChapGBT during our townhall last week.  An additional concern which I have is the loss of jobs due to AI bots.

Honored Contributor
Posts: 13,510
Registered: ‎05-23-2010

Re: I'm very troubled with AI right now, ChatGBT is just one of my concerns..

@dooBdoo @I'll be watching. Thanks for the tip. I've been reading article after article about this.