Dear Writers: The Bots Are Coming for Your Job. Think About a Career Change (Just Kidding!) (Sort of)

People who write for a living have enough problems just figuring out how to write for a living, so they sure don’t need some machine elbowing in on the action. Of the 43 gazillion humans who like to refer to themselves as “writers,” maybe 278 make an actual living at it. Okay, that’s an exaggeration. It’s way less than 278 (ha ha!).

Seriously, though. Earning a full-time living as a writer is a cute little pipe dream for the vast majority of people who aspire to it. A study conducted last decade found that nearly 80% of self-published authors earn less than $1,000 a year at it. Even more depressing: Over half (54%) of authors published by traditional publishers earn less than $1,000 a year.

If you write books, you’ll be lucky to sell a few hundred copies – which should fetch you just enough income to pay off the credit cards you’ll max out trying to promote your books.

Most people who write for an actual living – meaning they do it full-time and earn enough to pay the bills and save for the future – work for media companies, technical publications, content sites, etc. But these jobs are shrinking as traditional media outlets continue to slash their payrolls.

I earn a living writing content for a few different websites, and the work has been steady. But I know from personal experience that it can all stop on a dime. You never know when an email will hit your inbox informing free-lance contributors that due to budget cuts, all writing and editing will henceforth be done in-house. Thanks for your contributions and best of luck in future endeavors!

Now the writerly waters are set to get even murkier thanks to the rise of ChatGPT, the artificial-intelligence chatbot developed by OpenAI. The bot was released in November 2022 to great fanfare and hype, mainly because it can do things that only humans could do the previous several thousand years – like write.

According to a recent article on the Mint website, ChatGBT is capable of having “conversations on topics from history to philosophy, generate lyrics in the style of Taylor Swift or Billy Joel, and suggest edits to computer programming code.”

In layman’s terms, you can use ChatGBT and programs like it to write a novel, write a short story, write a poem, song, essay, homework assignment, master’s thesis, business memo, email, social media post, sales pitch, advertisement, news article, opinion column, letter to your aunt – pretty much everything.

More importantly, publishers can use ChatGBT to do all these things and eliminate human writers altogether, if they are so inclined. If you are a writer, ChatGBT is programmed to do what you do – quickly, efficiently, and cheaply. And it will.

Because it already is.

I recently learned through professional contacts that the CNET media site has started using ChatGBT to crank out some of its content. This hits close to home because CNET is exactly the kind of site I write for.

On the bright side, things have not exactly gone smoothly since CNET started experimenting with AI for its content. As Gizmodo reported last month, CNET has been “forced to issue multiple, major corrections” to one of the early AI-generated posts published on the site.

“In one single AI-written explainer on compounding interest, there were at least five significant inaccuracies, which have now been amended,” Gizmodo reported.

CNET had to issue a lengthy explanation, most of which involved ChatGBT’s lack of innate financial expertise, how it got the interest wrong, how it misunderstood compounding, that kind of thing. As of mid-January, the site had published nearly 80 AI-generated articles under the generic byline of “CNET Money Staff.” It wasn’t until other sites called out some of the errors that CNET bothered identifying and correcting its mistakes.

So that’s one problem with ChatGBT writer content – it might not have a deep understanding of its source material, or maybe just gets things wrong because the coding was inaccurate. But these problems are easily fixable. When it comes to news or non-fiction content, 90% of the game is just getting the facts and formulas right.

Fiction is a whole other thing, because it relies less on factual accuracy than style, imagination, plot, creativity, well-written lines, and believable characters. These can be taught (or even programmed) to a point, but they can’t really be mastered unless you are really, really good at them.

How will ChatGPT do when it comes to fiction? I have no idea. Results so far have been mixed, according to an article last week on the Washington Post:

“The bot was dreadful at reproducing the voices of a great novelists of earlier eras and today’s big sellers. For instance, its version of Stephen King began like a bad book jacket: ‘One day, strange things began to happen in Millfield. People started to disappear, and strange whispers echoed through the streets at night.’”

The article went on to say that the bot performed better when it wasn’t trying to replicate the work of the literary masters. You could say the same about the millions of human writers who try (and fail) to replicate the work of the literary masters.

I also recently came across a post on LinkedIn in which the author asked ChatGBT to write the post in the style of legendary gonzo journalist Hunter S. Thompson. Here’s the result:

“Ah yes, the future of tech organizations and their own personal #ChatGPTs. It’s a wild and trippy concept, ain’t it? Picture it, man – all the knowledge of the legendary Dr. Deming at your fingertips, accessible through the mystical force known as “The ChatGPT.” This future version of the #ChatGPT is the Oracle of the tech world, man, the go-to source for all your technical queries.”

As I noted in a comment on the post, that sounds like a bad imitation of a bad imitation of someone trying to write like Hunter S. Thompson and failing. I’ve read a lot of his work and don’t remember him using the term “man” at all unless it was in a direct quote, let alone twice in one graf. Thousands of humans have tried and failed to replicate the unique writing style of HST, so a bot has no chance.

But one day…

———

I am not what you would call tech-savvy. I have average tech skills for someone of my age and social background, which means I can use a computer and smartphone for most of the practical stuff. I am usually among the last groups of people to embrace new technology, mainly out of laziness. As I write this, my laptop is making a weird clicking/grinding noise that I know is bad news but have no idea how to fix.

Even so, one thing I have learned is that we underestimate the power of technological advances at our peril. Technology almost always wins the day. When the first wheel was invented I’m sure a lot of folks said it was just a passing fad, with no real practical use. It ended up changing the world in a million different ways. Same thing with the plow, the printing press, the firearm, the railroad, the internal combustion engine, the transistor, the microchip, on and on.

There are already predictions that ChatGPT and generative AI could be the next tech bubble instead of boom. Many in the literary and media communities downplay its potential impact on published material for all the same reasons they downplayed AI’s potential impact on everything else – it can’t feel, intuit, create, or imagine in the same manner as the human brain.

For what it’s worth, they used to say the same thing about computers playing chess. The theory was that computers could never beat human chess masters because computers could not think intuitively, could not feint opponents into making bad moves, could not predict where an opponent might go, or suss out how an opponent might react to specific moves at specific points in a game. You could program a computer to analyze millions of possible moves and countermoves, but you couldn’t teach it to how to win on sheer talent and instinct.

This theory looked pretty good in 1996, when world chess champion Garry Kasparov beat the Deep Blue computer 4 games to 2 in a 6-game match. But a year later the tables were turned – Deep Blue defeated Kasparov 3.5-2.5, with one match ending in a draw.

Now, games and writing are not the same thing (obviously). But they’re not as far apart as you might think. The best chess players have deep wells of creativity and imagination, and are able to envision dozens of different scenarios for winning games. They can see moves unfolding far into the future and develop strategies to deal with them. This is not just a function of math and science. It’s also a matter of art, intuition, creativity, inspiration, instinct – just like writing.

I don’t think bots will ever be an adequate substitute for human writers. I do think that bots will one day crank out much of the online content you read. I also think there will come a day when you won’t be able to tell the difference between a book written by a bot and one written by a person.

And since publishers will know this too, you can bet they’ll jump on board the second they can make an extra nickel from it.

Am I happy about this? No. Am I afraid of this? No, not really.

But I don’t doubt it will happen. Technology marches on, and we are mostly powerless to stop it. ‘Twas ever thus.

Note: The photo was published on CNBC. The bots don’t mind my borrowing it.