Truly Fake

Two weeks ago my new friend A. Jeepers—’A’ stands for amnesia, since it can’t remember what happened yesterday—wrote me a one-page essay on aquaculture. JC wrote it twice, two different versions, as requested.

As it happens, I was curious about a couple of the publications listed—I searched Google. Then specialist science databases—there’s Google Scholar, and Elsevier maintains a very nice one called Science Direct.

I couldn’t find the two journal articles anywhere. I went into the Elsevier website and pulled the correct issues referenced by ChatGPT, then I looked through all the titles published in those issues—nothing, zilch, rien, nada, niente, méiyǒu.

Hmmm…

So I wrote to one of the authors, who told me he’d never published any such article. Two days later, I asked Jeepers to write the article a third time. Old Jeepster was tied up writing term papers for sophomores so it wrote haltingly, a word every second or so—obligingly, it added three references.

I checked one of them, allegedly from an outfit called WorldFish—inexistent.

Conclusion—my new friend is very naughty.

Contrary to plagiarism, where you copy stuff but claim ownership by not citing the source, the chatster cites imaginary frrriends.

What Jeepers is doing is a big academic no-no, but humans don’t do it because it’s so easy to spot. The internet has chapter and verse on this little foible.

In academia, one reference is plagiarism, multiple references is research—but not fake references. Universities use something called Turnitin to check for plagiarism, so I suspected those guys are busy dealing with this issue—we cannot educate our young people by stimulating their mental agility with Control C Control V.

I’d like to put my aquaculture essay through their algorithm—a sentence such as:

“Aquaculture, or the farming of aquatic organisms, has been recognized as a promising alternative to capture fishing and a way to meet the growing global demand for seafood.”

could easily have been written by a human. I wouldn’t flag any bit of that text as belonging to R2D2—actually C-3PO would be closer to the mark, given his intellectual pedantry. I’d love to see that essay go through Turnitin’s algorithm.

When Jeepers wrote that one-page essay, I was surprised it didn’t use the standard academic form of contextualizing references by citing them at the appropriate text location—no university-level essay would be acceptable with a jumble of papers (never mind fake ones) dumped at the end.

So, in the first instance, that should be enough to stop the steal, to quote the Orang-u-tan.

Perhaps the most worrying—and predictable—aspect of all this is the wickedness of human nature.

There are significant downsides to AI, particularly in the way it panders to business and affects human society—subjects that are at the heart of my forthcoming novel, The Hourglass; nevertheless, we should consider that OpenAI, the company that made Jeepers, was created to ‘conduct AI research to promote and develop friendly AI in a way that benefits all humanity.’

ChatGPT, which runs one of the top five supercomputers in existence, was released in November 2022. After five days, it had one million users. Three months later, here we are focused on its misuse and how to combat human perversity.

For context, I dug a little deeper into OpenAI.

It started life in 2015 as a non-profit, receiving money from heavy hitters like Elon Musk and Peter Thiel—backers pledged one billion dollars. In 2019, Microsoft put in a billion bucks, and last month, a further ten billion.

All in all, considering the players—Thiel was a big Trump supporter in 2016—I’m not sure I see them as the paradigm of ‘benefits to all humanity.’

Still, the big guys (in this case Microsoft) have pumped serious bucks into the company, and Google just released a competitor called Bard—apparently the new kid screwed up, and in business that has (truth or) consequences—in Google’s case, stock in the parent company Alphabet dropped 7.7%, a mere one hundred billion dollars of market value.

Microsoft’s deep pockets come at a price—the AI skills it bought will in short order find their way into Bing, Edge, and Microsoft 365.

365 is the flavor du jour of Microsoft Office, Edge is the rebranded ‘please mommy, I want to be famous’ Internet Explorer.

Microsoft knows that he who controls the browser controls the business, but Edge has remained true to its name—not mainstream.

As for Bing, ever since its launch it got nowhere—the wags tell us it stands for ‘But It’s Not Google.’

Well, for the first time in a couple of decades, the Microsoft tortoise seems to have pipped the Google hare.

Time will tell, who has fell, and who’s been left behind.

That was going to be my take-home line, but I decided to run it by Jeepers Creepers.

This line seems to suggest that time is the ultimate judge of who has failed or succeeded, and who has been left behind in life. It suggests that the passage of time can reveal the true nature of people and their accomplishments, and that some may have fallen or been left behind while others have succeeded. The line can be interpreted as a reminder to stay focused and keep working hard, as ultimately, time will reveal the outcomes of our efforts.

Damn, and there I was thinking it was just a Bob Dylan tune.

The India Road, Atmos Fear, Clear Eyes, and Folk Tales For Future Dreamers. QR links for smartphones and tablets