It will exponentially increase spam
I’ve recently wrote about AI, but perhaps a few more observations are in order. Recently a popular science fiction magazine had to stop accepting submissions due to the huge number of stories written via ChatGPT and similar AI services. I can only imagine what school teachers have to deal with today. When I was teaching literature at college, years ago, it was common for students to get “inspiration” from Wikipedia and similar sources, just changing a bit the wording. Although, once, a less savvy student delivered a paper which was completely copied from Wikipedia, as in Ctrl C + Ctrl V. He didn’t even took the trouble of removing the links or footnotes.
Tools like ChatGPT are harder to detect, and in fact, it is not clear if they can be considered plagiarism or not. At least the prompts are written by the students, and they can edit the final result into a more coherent form.
“I used to spend hours writing creative copy, but now all I do is tell the bot what I need and it writes everything for me. It’s the ultimate AI content writer, and a must-have tool for bloggers, marketers, & entrepreneurs”, says a satisfied customer of a new AI writing service, in a comment which may itself have been written by a bot.
It seems to me that the greatest beneficiaries of the new AI writing programs are going to be spammers. “I used to spend hours writing e-mail scams about Nigerian princes, but now I only need to give the bot a few prompts and it writes it for me, and without spelling errors!”
Now, a bot may not be capable of writing insightful prose or inspiring poetry, but it is, today, perfectly capable of writing a high school paper or journalistic articles. In fact, I assume that a lot of journalism these days, including opinion columns, is written by some form of AI. At least, they read like that. So much of modern journalism is fake, I don’t see why a robot couldn’t do it.
For instance, an article at the left-leaning site Vox assures us that there is something called “robot racism”, and that “people treat white robots better than black robots”. I kid you not. Lots of “woke” news articles read like robot prose these days.
A lot of corporate-speak also feels as if it was created by robots, even when it isn’t.
Could this article I’m writing be written by AI? Probably. But what would be the point?
Another problem of AI is that it doesn’t really save work, money or even energy, but simply transfers it somewhere else. A recent case is a pilot project for self-driving cars in San Francisco. The car can can move automatically without a driver, but it cannot automatically open or close its own doors. If a client leaves the door ajar, some worker will have to come and manually close it. Sure, I suppose new models of cars can fix this issue, but there will always be something that a human will need to do, such as fixing a flat tire or even cleaning the backseats. Besides, while the company may save on drivers, it has to hire more people to monitor the cars route, to assist customers, to do technical service and programming etc. None of those companies so far are making any money.
Perhaps AI systems for self-driving may be better used for the transport of goods than of passengers? Recently a couple of freight train derailments in Ohio, as well as a train crash in Greece, appear to have been caused by human error or bad maintenance. However, it is not clear if a totally automated system would have avoided such disasters. Machines can have bugs too, and they require constant supervision. The maintenance is done by humans too.
Then there is the question of how the growing automation of society and the reliance on AI systems will further increase social atomization in an already very socially awkward age. Most people today prefer to send messages instead of talking on the phone, or talking in person; also many people now work remotely without the need to interact with others. As AI and automation grow to occupy more and more areas of life, our contact with other humans will be further reduced. Wether that’s good or bad, I let you decide, but it is clear that human social contact will probably not increase the coming years.
One day, perhaps, we will only meet other people safely behind a screen or VR headset. And love letters will be written by ChatGPT.
4 comments on “The real problem of AI writing”
Seems ironic that mistakes in grammar and spelling are one of the few tells that a human has written something. Imo the most common mistake is using “it’s” instead of “its”. AI will not make this mistake, yet bloggers I like routinely do. Should we be grateful at their lack of conformity to traditional rules?
I’d always blamed social media for allowing grammar to degenerate (and emoticons to proliferate) but now I see that kind of dumbing-down was a human decision, albeit a techie overlord’s, not a computer’s.
It seems funny to me that lots of native English speakers make this “it’s”/”its” mistake, as well as “they’re” and “their”, which would seem (to me) pretty easy to avoid. As a non-native speaker, for me the major difficulty in English are its (not “it’s”) prepositions. It’s (not “its”) not clear in many cases if one should use in, on, or at, and many times it’s not even clear why we use the same preposition in some cases and not in others. The rules seem pretty arbitrary. In Latin languages I think there are less prepositions. In Spanish “in, on, or at” can be translated to one same word, “en”. German has a lot of prepositions too, but it seems more logical than English, at least it has rules. Lots of rules. Germans love rules.
They’re/their is second on my list of common mistakes. I recently tried to give a short definition for “its” to Rolo. I said “its” means belonging to “it”. It shows possession. I’m not sure that makes much sense to a Slav. Easier to just focus on it’s means it is. If that doesn’t make sense in your sentence, then by default it’s “its”. (lol) getting absurd, no?
You seem to confirm the preposition problem. English prepositions and conjunctions give my German friend the most difficulty. She’s fine w/ nouns and adjectives. Less fine w/ our pronouns.
So your college Lit teaching was in Spanish. That’s what I thought. I have another friend from Argentina whose spoken English is good, but written is limited. I like her spoken Spanish. The tempo is slow, distinct. Her parents were Swedish.
No, the college teaching was in the US, in LA. (With quite a few Latino students who spoke/wrote better Spanish than English, however). It’s been many years since I left Argentina and haven’t been there for years.
Comments are closed.