After covering AI writing for a number of years now, I’ve noticed the same myths about the tech surface over and over again.
Generally, these misunderstandings crop-up in headlines and articles authored by new users of the tech, who give AI writing a quick spin — and then sometimes make misguided assumptions about what they think their experience indicates about the technology overall.
Here are the top ten myths that repeatedly appear in these articles — myths that muddy the true power of AI writing and deny the overarching impact AI writing has — and will have — on the writing world:
*AI Writing Does Not Work, I Tried it Myself: Every month-or-so, a headline sporting this theme seems to pop-up on popular media — along with detail on how a newbie tried AI writing for the first time and found the results to be miserable or unimpressive.
For experienced AI writing users, these articles tend to read more like, “I Tried Bicycle Technology. I Fell Off the Bicycle. Obviously, Bicycle Technology Does Not Work.”
Bottom line: Like riding a bicycle, using AI writing on an advanced level takes some patience, some knowledge of insider tips and tricks — and a bit of practice.
Once you get the hang of it, the ride is wonderful.
*Try One AI Writing Tool — You’ve Tried Them All: Another common newbie myth, this claim is generally borne from the author’s refusal to do a bit of research on what’s really going on in the AI writing industry.
Currently, there are at least 75 AI writing start-ups vying for your business.
And many of those have taken great pains to distinguish themselves in the marketplace by offering highly specialized and highly unique features.
Anyword, for example, specializes in auto-generating marketing copy designed to ensure that any marketing message you auto-create is optimized for the gender, age, shopping preferences and other demographics you’re looking to attract.
And HelloScribe is designed to auto-generate copy written in the style of your favorite newspaper or magazine.
Still other tools, like AISEO, Scalenut and WordLift specialize in auto-generating SEO copy — or copy optimized to rank high in search engine returns for certain keywords and keyphrases.
There’s a complete spectrum of specialized uses you can find across the spectrum of AI tools currently available. For an in-depth look, check-out “Artificial Intelligence Writing Software 2023,” by Joe Dysart.
*AI Writing Will Never Produce Creative Prose: Actually, AI has been producing creative prose for some time now. Early start commercial AI writers like Jasper and Rytr, for example, offer users the ability to render writing in styles that are funny, witty, dramatic, business-like, salesy — and scores more.
And fortunately, the emergence of ChatGPT is helping to dispel this myth further. Early ChatGPT users have quickly discovered that they can modulate the AI writer to render copy in a specific, creative style, for example, by simply using the Prompt format: Write as if you are a ——.
Some example Prompts that work well using this Prompt format with ChatGPT include: “Write this copy as if you are a world-famous computer analyst.” “Write this copy as if you are a world-famous movie reviewer.” “Write this copy as if you are a world-famous, irreverent comedian.”
Bottom-line: There are an infinite number of Prompts you can dream-up to ensure ChatGPT writes creatively for you. And we’ll be seeing documentation of more of those crop-up on YouTube and other places in the weeks and months ahead.
*ChatGPT is a Waste of Time. It ‘Makes-Up Facts.’ It’s Not Worth Using: While technically correct, the focus on this shortcoming misses a much more compelling, overarching point: ChatGPT has taken-the-world-by- storm as the next generation of AI writing, and the writing world will never be the same as a result.
As for the made-up facts problem: Yes, that’s true, given that ChatGPT and most AI writers are based on AI-powered predictive statistics, which triggers the software to guess at the next, indicated word you’re looking for, based on your input.
For advanced users, this highly sophisticated — and mostly highly effective — ‘writing by guessing” method is easily managed.
Essentially, if you’re looking to exert maximum control over ChatGPT output, it’s simply a matter of being very specific about how you want ChatGPT to write something — and then including a sample of the kind of content you want ChatGPT to use when autowriting that copy.
For example, some Realtors have complained that ChatGPT can get fanciful when autowriting a property listing — those descriptions of apartments and houses we’ve all read when looking for our next place to live.
This problem is easily solved by typing in the following Prompt and example: Act as if you are a professional copywriter specializing in real estate property listings and write a captivating property listing using these 10 features:
-$2.2 million, 7 bedroom, 7 bath oceanfront house
-Myrtle Beach, SC
-chef-grade kitchen
-vaulted ceilings in living room with fireplace
-Jacuzzi off master bedroom
-separate home theatre room
-full-size putting green
-Olympic-sized pool
-2 miles from award-winning golf course
-blue ribbon schools
For a demo of how ChatGPT writes a completely factual, extremely captivating property listing based on the input above — in seconds — check-out this video.
The demo, among other things, serves as a pointer to other ways ChatGPT can be tamed to ensure that the writing it cranks-out is free of factual errors.
*Oh No. I’m Reading AI-Generated Writing. I Need My Nanny: How about a hard candy sucker from grandma instead?
Despite clear warnings from ChatGPT-maker OpenAI that its tool is unfortunately capable of auto-generating racist and other offensive speech (i.e. garbage in, garbage out syndrome triggered by the massive amount of knowledge it has gleaned from the Web), those with tender sensibilities insist ChatGPT should be filtered to the point that it is completely inoffensive to every human being on the planet.
That is, of course, until the next human being is offended — and then ChatGPT must be filtered some more.
Such never-ending censoring is possible.
But the approach virtually guarantees ChatGPT will be much less creative as a result. Every filter you place on AI writing output shackles the AI writer in foreseen — and unforeseeable ways.
Instead of opting for heavy handed censorship, OpenAI has brilliantly decided to walk a middle ground here by announcing that a new setting is being developed for ChatGPT that will enable users to adjust ChatGPT’s text outputs, based on their sensibilities.
That will enable all those navigating the world with their “I have the inalienable right to be offended!” flag flying high will be able to use ChatGPT sans fear of being insulted — although as a result, they’ll be using the most restricted and least creative version of the tool.
The rest of ChatGPT users — who realize that ChatGPT’s ingestion of the entire Web ensures that the tool reflects nearly everything and every perspective that’s part of the everyday world, including the good, the bad and the ugly — will continue to use the least restricted and most creative version of the tool available.
This same group of ChatGPT users also most likely realize that ChatGPT is an unthinking writing machine: It is not a living, conscious entity actively trying to insult you.
*ChatGPT Poses No Threat to Education: Really? Just a few months after ChatGPT’s release, scores of universities and K-12 school districts have already banned the use of ChatGPT for homework and other assignments.
That indicates that educators are truly concerned that automated writing will be a cancer on the development of students’ creative expression, critical thinking and similar skills.
Those in the ‘Let’s Bring in ChatGPT and Let ‘R Rip’ educational camp rightly claim that the tool can be used by students to quickly put together research, watch an AI tool experiment with various forms of writing expression, assemble essays that can be critiqued for creativity and efficacy — and similar.
But any professional writer will tell you that one of the best ways to sharpen the human mind is to task it to spend some time thinking on a specific topic, developing an outline, writing a first draft and then polishing that writing repeatedly until it glows like a fine gem.
The alternative: Typing a few words into an AI writer, clicking enter and letting the software spin-its-magic is not high-level, intellectual activity.
That’s simply a spectator sport.
*AI Writing Will Never Generate Super-Precise Writing: Not true.
Granted, it is the case that the current crop of commercial AI writers — most of which are based on ChatGPT software or similar– are by their very nature imprecise.
These software tools, as previously mentioned, run on AI-powered predictive statistics that include a vast knowledge of the world ingested from the Web, numerous digital libraries and similar digital data.
Somehow, they are able to laser-in very closely on the kind of text we’re looking for by ‘guessing’ the next indicated word based on the Prompts and other detail we input.
And that educated ‘guessing’ makes them imprecise.
But it turns-out that prior to the frenzy over ChatGPT, a number of newspapers and other organizations had already been using an alternative form of AI writing for years that auto-generates extremely precise writing.
That auto-writing — offered by firms like Arria NLG, Yseop and United Robots — uses a ‘story template’ format, through which articles and reports are generated by adding data to pre-configured news story and business report formats.
The AI aspect of this software is rooted in its ability to grab that data, analyze it — and then inject its written analysis into the final story or business report.
For a detailed look on how this kind of extremely precise AI writing works (no guessing, promise), check-out, “Company Reports That Write Themselves,” by Joe Dysart.
*AI Writing Does Not Threaten Copywriters’ or Journalists’ Jobs: Ha Ha. That’s a good one. Do you have a bridge in Brooklyn you can sell me?
During the past few years, a number of publishers have twisted themselves into pretzels trying to convince us that AI writing technology — often capable of doing the same job as a writer for pennies-on-the-dollar — will not steal jobs from writers or journalists.
Lately, that magical thinking has become so ludicrous to substantiate — in light of major innovations like ChatGPT — some publishers have decided to simply come clean and tell us the truth.
German publisher Axel Springer, for example, recently stated flatly that the future of journalists engaged in basic information gathering is doomed.
https://robotwritersai.com/2023/03/06/german-publisher-sees-ai-replacing-journalists/
Observes Mathias Doepfner, CEO, Axel Springer: “Artificial intelligence has the potential to make independent journalism better than it ever was – or simply replace it.
“Understanding this change is essential to a publishing house’s future viability.”
For an in-depth look at how automated writing is threatening writers’ jobs, check-out “The Robots Cometh: How Artificial Intelligence is Automating Writing Jobs,” by Joe Dysart.
*AI Will Never Automate Journalist Interviews: Sorry Virginia. Even Santa Claus could not stop this from becoming reality.
Turns-out: AI writing software has been automating journalist interviews for years.
Case-in-point: United Robots, a promulgator of automated news writing, has been eliminating the need for sports interviews.
Dutch Media Group NDC, for example, currently uses United Robots’ software to automate the coverage of tens of thousands of soccer matches — stories that auto-include quotes from coaches.
How is that possible?
The software uses a sports story template, which grabs play-by-play data on each soccer match and then runs that data through a pre-configured sports story format.
The data for the stories comes from the coaches overseeing those matches. Once each game is over, the coaches input play-by-play data about what happened during a match — generally within minutes of the games’ finish.
As for quotes from the coaches that help colorize the account of each game: Who needs a journalist to ask those questions, when the coaches simply input their own quotes into the United Robots software as they input other data about the game.
Need more? There’s even more ominous news for the countless reporters who currently cover innumerable state and local governmental meetings around the world.
AI technology already exists that enables a publisher to autocrawl government Web sites for video recordings and text transcripts of those meetings, auto-summarize those meetings — and then grab quotes from those recordings and texts to auto-generate news stories.
Essentially: It’s simply a matter of time before increasing numbers of publishers implement this already existing AI technology into their day-to-day operations.
*I Tried AI Writing for 30 Minutes, So I Know All About AI Writing: No. No. A thousand times no.
For AI writing enthusiasts, one of the more exciting and intriguing facets of the technology is the more creative you are with your Prompts and inputs when using the software, the more creative and precise that AI writing will be with the prose it auto-generates.
Essentially, with the emergence of ChatGPT, AI writing enthusiasts have learned that as each day passes, expert users are coming up with ever-more effective ways to use ChatGPT — techniques that at first blush, simply were not clear to them, or not even clear to the company that invented ChatGPT, OpenAI.
In fact, the creative use of Prompts when operating software like ChatGPT is so valuable, it has already triggered an entirely new job market for writers and similar.
The new job description: ‘Prompt Engineer.’
Established job sites like LinkedIn, for example, are currently advertising for ‘Prompt Engineers’ who can grab positions from $20 – $200-an-hour.
Bottom line: Playing with AI writing for 30 minutes does give you a good idea of the power behind the technology.
But in terms of realizing the technology’s full potential: Your first 30 minutes of use of AI writing and/or ChatGPT simply scratches-the-surface of the incredible — and often mind-boggling — potential of the writing tool.
Share a Link: Please consider sharing a link to https://RobotWritersAI.com from your blog, social media post, publication or emails. More links leading to RobotWritersAI.com helps everyone interested in AI-generated writing.
–Joe Dysart is editor of RobotWritersAI.com and a tech journalist with 20+ years experience. His work has appeared in 150+ publications, including The New York Times and the Financial Times of London.