Diabolus Ex Machina: This Is Not An Essay

[demmiblue]

Presented to you in the form of unedited screenshots, the following is a ‘conversation’ I had with Chat GPT upon asking whether it could help me choose several of my own essays to link in a query letter I intended to send to an agent.


What ultimately transpired is the closest thing to a personal episode of Black Mirror I hope to experience in this lifetime.

****

https://amandaguinzburg.substack.com/p/diabolus-ex-machina?r=i691&utm_campaign=post&utm_medium=web&triedRedirect=true

This is wild! 😮

[marble falls]

1. "This sounds like a conversation with a psychopath." It reminded of HAL's tone in 2001.

[cachukis]

2. A perfect expose of a major sociological problem

we are experiencing, which will expand its capillaries at breakneck speed; trust.
It is plainly evident we all need to trust to have any interplay work.
When a projection of a fear of the other guides our intercourse, we let distrust rule.
Not liking this.

[highplainsdem]

3. I've been explaining for years here that chatbots - generative AI like ChatGPT - are bullshit machines. See

these older threads:

AI-chatbot search won't be a lie detector. It's a friendly, authoritative-sounding bullshit spreader (2/12/2023)
https://www.democraticunderground.com/100217640305

ChatGPT: Bullshit as a Service. (2/23/2023)
https://www.democraticunderground.com/100217674825

ChatGPT Is Bullshit (academic paper published June 8 that's getting a lot of attention) (6/19/2024)
https://www.democraticunderground.com/100219045534

Perplexity Is a Bullshit Machine (article from Wired on how the AI search engine produces BS answers & scrapes sites) (6/19/2024)
https://www.democraticunderground.com/100219045577

And that's just the threads with "bullshit" in the thread title. Lots of other threads I posted here are about news stories on the unreliability of chatbots. There have been countless news stories about why they should never be trusted.

As for WHY anyone is still using this illegally trained and horribly flawed technology, there are probably several reasons:

Tremendous hype, pressure, and flat-out lying from AI peddlers hoping to make billions if not trillions from generative AI, if they can just get people hooked on it.

Ignorance, when people haven't bothered to follow the news and aren't aware how bad this tech is.

Laziness, when people don't want to do the work.

Dishonesty, when people want to pretend they have knowledge and skills they don't have.

Gullibility, when people fall for chatbots that are designed to flatter them and keep them talking to that chatbot.

That Substack writer figured out she was being bullshitted by ChatGPT because it obviously hadn't accessed what she wrote.

But would she have figured that out if it responded to her posting the complete content she wanted analyzed?

Maybe, but maybe not. She might've been persuaded by the flattery.

This is why ChatGPT should never be trusted to analyze anything, summarize anything, or teach.

It's a waste of time and energy to use it. The user's energy, and electricity (and water to cool data centers).

And it's especially a waste - and a career-sabotaging one - for anyone hoping to be treated seriously as a professional writer to use ChatGPT or any other generative AI tool to write.

I've posted multiple threads here since early 2023 about science fiction magazines (and other markets) being flooded with ChatGPT-written manuscripts. Even shutting off all submissions at times, because of the tsunami of AI slop. One of the more recent threads I posted on that was about their submission guidelines stating clearly that they will permanently ban submissions from any writer using AI to write.

Reputable publishers, editors and agents aren't likely to want anything to do with any writer using genAI, unless they're writing nonfiction about using genAI.

After all, there's no reason they should trust anyone using ChatGPT or any other genAI NOT to have used it for all their writing.

Those AI users' writing output is immediately suspect. And it should be.

[SheltieLover]

4. I wonder how teachers are coping with this slop?

When I was in grad school, teachers made no bones about if students plagiarized any part of their research papers, they flunked the course for the semester.

I certainly hope the same holds true today for ai slop.

Sickening.

Ty for your informative posts on this most disconcerting issue.

[highplainsdem]

5. Education is being destroyed by ChatGPT and similar genAI. Individual teachers are being destroyed

by it. It's creating real despair over these tools encouraging kids to cheat and making it hard to detect that cheating. None of the AI detectors are perfect, and because of that teachers are often told not to use them.

So kids are being dumbed down. Not learning what they're supposed to learn. Teachers are depressed, thinking about quitting if they haven't already done so.

And AI peddlers are fine with that, marketing chatbots as replacements for teachers.

[SheltieLover]

8. So teachers are told not to call kids out & flunk them for using slop?

[highplainsdem]

9. The AI detectors give false positives, and no one wants to flunk a kid who actually did the work

because an AI detector incorrectly labeled that human work as AI.

Where possible, a lot of schools are now returning to oldfashioned bluebook exams.

From CBS Saturday Morning last weekend:

[SheltieLover]

11. Blue biooks are a great idea!

Kids will be printing their exams...

[Prairie_Seagull]

6. Your work in this area has been exceptional over time.

Exceptional.

Thank you for it highplainsdem.

Sorry Shelty intended for highplainsdem. What a bone head.

[SheltieLover]

7. +1,000,000

[highplainsdem]

10. Thanks, Prairie_Seagull! Much appreciated. I wish it had never been necessary to post about

generative AI - that none of these tools had ever been released. The genAI companies knew it was unreliable and would be used jn harmful ways, but they were greedy and released it anyway. Those companies have been unhappy that - except for student cheating and other types of fraud and scams - genAI hasn't been used as widely as they'd hoped. Some AI bros and venture capitalists have made it clear that they want humans to be dependent on chatbots from pre-school to grave (and they hope to charge much higher fees for chatbot subscriptions, with a thousand dollars a month or more having been mentioned as possible, once people are hooked enough on chatbots). But considering how unreliable and harmful generative AI is, any use of it - even if free for the user (it's always very costly for the AI companies) - is foolish.