Simple Known Dangers with AI, and Possible Future Dangers Too.

[fox]

I would have said in the past, that an essay is a good way to evaluate a person's ability to use information creatively to tell a story or test a hypothesis. But in one of the courses I taught called Fisheries Assessment and Management, I learned before the days of AI that it was too easy to plagiarize an essay on a relevant topic - from past ones written, from other students, or buy purchasing one online. After awhile I felt that I no longer had the ability to catch these violations, and which ones I caught was just a matter of luck. (Most of the cheating I caught in this course was from students lifting paragraphs from published papers, not the other kinds of plagiarism mentioned above.) So I came up with a creative way to issue the assignment that would be much, much harder to plagiarize or buy. I called it "advances in ....". You would take a fisheries topic, identify five relatively recent advances in the science around that topic, indicate why they were; advances, and review the five published papers where the advances came from. The paper also had to summarize the state of the art on that topic. It worked very well and probably cut the rate of plagiarism to zero or near zero. But I'm not sure that AI couldn't handle this; I never had to find out.

[Jason]

The paper also had to summarize the state of the art on that topic. It worked very well and probably cut the rate of plagiarism to zero or near zero. But I'm not sure that AI couldn't handle this; I never had to find out.

Unfortunately, that is one of the things that AI is excellent at. What it wouldn't be good at is writing the essay in the student's voice. But unless you have your students writing several essays and you don't have too many students or an amazing memory, that would be hard to pick up on, I'd imagine. But, AI does format its answers in a particular fashion, and if you use it enough, you might pick up on its voice. Handing out take-home assignments for high school teachers must be particularly awful. I would bet that more students are using it to cheat than at the university level. Could be wrong, though.

Have you used any AI bots, Fox? I've used ChatGPT occasionally but not the others. I did use Grok recently to ask if it was still racist. If you don't understand why I asked that, check some recent US news having to do with Elon Musk and Grok. The answers seemed to be too "woke," so Elon Musk made some changes to fix that, and the bot started acting racist with positive comments about Hitler. You'd have to read them yourself. I don't want to offend anyone.

[fox]

....Have you used any AI bots, Fox? ....

I might have used ChatGPT once or twice, but the only AI I think I have been using is the AI you get on MacOS, iOS, and iPadOS.

[buster]

Another interesting unfortunate 'invasion' of the nonliving shows up in a CBC website article. It makes sense, but I didn't realize that bots were used as moderators for social media because of the huge numbers of participants, many of whom are also bots disguised as humans.

There are a number of people apparently who have been removed from sites by non-humans because of some perceived transgression, and phoning to complain gets you nowhere because it's impossible to connect to a human.

   https://www.cbc.ca/news/canada/toronto/teacher-wrongly-accused-child-exploitation-meta-account-apology-1.7599595

[Jason]

I was one of those people who had my account disabled because of unclear reasons from Instagram, which also made me lose my Facebook account. Some of my friends were formerly Facebook friends, and their accounts don't let you send friend requests to them, so there are friends I'll never get back unless they look for me. Facebook and Instagram don't have numbers or emails you use to contact them when it happens. And new subscriber requirements make it extremely difficult to create a new account. They make you do a short video to see if your picture matches a disabled account, as well as other measures. I asked ChatGPT how to bypass them so I could create a new account. Hard to bypass a video unless I got somebody else to pose for me!

I finally got back on with a new account by using an existing account that I created for a family member who ended up not using it and passed. Not the way I wanted to do it, but I was able to change the name.

At least I have friends outside social media. There are a lot of shut-ins who don't, or people with mental illness who go on Facebook with other similar conditions to get peer support. And they're just cut off, just like that, and probably by AI and a faceless corporation that won't help you if you could find a flesh-and-blood person to do it.

Btw, it's not just mods that are AI-controlled now, tech companies are using AI as the front line of support. You click on the chatbot to ask questions, and many require you to use the bot first before you can even try to reach a real person. It can be frustrating. Most of the time, the questions I have can't be answered by a bot. It's because it only searches the help files to answer your question, and I've already gone through the help files. If it was in there, I wouldn't be trying to find a real person.

[buster]

Reading the local edition of our newspaper yielded two more examples of unpleasantness.

a) There is a picture of a massive fire in British Columbia near a highway taken from the air appearing in social media. The authorities warned people to disregard the fire, because it does not exist. The picture was created by AI.

b) AP reported a study done using fake 13 and 14 year old teens in contact with ChatGPT getting information on drugs, best ways to commit suicide, and even how to write notes to leave for the parents. This age group seems to be the most committed to asking the program for advice.

A simple search AP and ChatGPT will produce the article.

[buster]

Interesting article about AI in The Examiner today by Gwynne Dyer about the unforeseen consequences. Found this copy in the Maritimes that might let you in:

https://www.saltwire.com/cape-breton/opinion-cape-breton/gwynne-dyer-investigating-the-looming-artificial-intelligence-crash