My first public foray into the breach of AI (Artificial Intelligence) was an attempt at figuring out just how “intelligent” it is right now. You can read the post on the you’re listening to radio revel blog, or if your a podcast fan, you can listen to it anywhere you get your podcasts (I think!).
In this post I ask myself and ask other writers the question in the title: be honest with your feelings about the threat that AI may present to our jobs. Will AI replace online writers? Well, maybe, but let’s cast our minds back 46 years.
It’s the late ’70s and the popular TV series The life and times of Grizzly Adams was, at least for two seasons, all the rave. Now, I can’t find anything to confirm this, but I had heard somewhere that this was the first TV series that was at least partially written with the help of computer software. The writing for both the series, and the 1974 movie that inspired the series, was attributed to a Lawrence Dobkin, a character actor who also directed for TV and wrote screenplays. Older readers (like me!) might be able to drum up an audio memory of what this text sounded like:
There are eight million stories in the naked city. This has been one of them.
That was Lawrence Dobkin.
The closest I can find to computers taking part in that series, though was this tidbit from AFI Catalogue:
According to a 24 Mar 1975 DV article…. Sun Classic Pictures relied heavily on computer analysis to determine which film projects to produce. Unlike major studios, which made movies to appeal to a majority of filmgoers, Sun spent an average of $85,000 per film on a five-stage computer program to find subjects that would appeal to at least three percent of the audience that the majors overlooked.
Considering what computer software must have looked like in 1975, it may be stretching things a bit to say that AI took part in the choosing of appealing subjects for viewers. Still, it seems to be a germination point.
In that same time period, when I was in high school, registering for classes involved being given a computer punch card that had our “identity” punched in. We walked around the gym, visiting department tables (English, Math, Science, etc.) and collecting individual punch cards for each class we were interested in. We gave those cards to a secretary at the end of the ring of tables.
Those cards were sent to be processed somewhere on a big, fat, probably room-sized punch card computer. Class list print-outs would be produced about a week later. So, with that thought in mind, that a computer was grinding data and spitting probable story lines, or project ideas, seems plausible. Writing scripts, though, is a different matter. A screenwriter, a human being, wrote the script.
Bringing it all to my work as an online writer
I believe (and the stats support the belief) that I have few readers because I have consistently refused to write in the same way as all other online writers do. You know the stuff I’m talking about:
- Identify your audience.
- Identify a problem they need solved.
- Offer a solution to the problem.
- Rake in the bucks.
I’m the kind of writer who knows that his audience will be reduced. I am not “looking for an audience”, nor do I need to “identify” a particular audience before writing. So, most of my titles are far from “click bait”, my images more often than not have nothing to do with the theme of anything I write, and those who do read my stuff have found me because they’ve bothered to look for me or something I write about. So, not many “followers”, hardly any “likes”, and I don’t mind at all.
Still, I do have to put food in the pantry and pay the Internet provider. So, I share some of my craftsmanship with a platform that is willing to pay me to write stuff for them, following the above mentioned scheme, taking advantage of my knowledge of the subject. They didn’t know what they were getting themselves into on hiring me, though I knew exactly what I was getting myself into.
Identify your audience
The audience for this platform, from what I could glean, coax out of them in the initial interview, was people who are learning about, using, or interested in persuing, a particular subject; they might have doubts about terms, rules, norms, might even be interested in using the pay-for service the platform offers. That was only partially true.
The true objective of my posts was to draw traffic to a blog, and as the objective of the blog was email harvesting for marketing, the real audience was not people, it was the audience of one: Google Search Engine SEO Algorithm (GSESA). My information was bait, not knowledge or advice. Let’s keep that in mind and move on to the next step in the process.
Identify a problem the audience needs solved
Again, not completely clear, though the prey using Google will certainly have a problem, something like “I don’t understand such-and-such“, but that’s not the problem GSESA needs solved. The only real problem GSESA might have is parsing through gazillions of words and word combinations, categorizing, indexing and prioritizing them for their “greatest hits” pages. And GSESA has already solved that problem by inherently being that algorithm that focuses on certain things while ignoring others.
Look at this sentence: The teacher has breakfast at a diner in the morning.
Now, let’s chop out all the unnecessary words, leaving us with: teacher / breakfast / diner / morning
Now, let’s apply standard English sentence word order:
- subject = teacher
- verb = too common to include
- object = breakfast (implies a related verb, like “cook“, “prepare“, “eat“, and the common “has“)
- where = diner (we know it’s a where because it’s a place, so also implies a preposition “at” or “in“)
- when = morning (again, it’s a time word, and the preposition “in” can be assumed)
Those five words, then, contain enough information for GSESA to “understand” the sentence. They would be called keywords by those of us who write with GSESA in mind. This type of writing is called SEO writing, “search engine optimization”, that is, we write with keywords in mind, reduce ambiguity and straw words as much as possible, to help GSESA resolve the simple problem of “too many words”.
GSESA’s “intelligence” may (and probably does!) go farther than assuming prepositions like in and at. For example, the subject is teacher and the place is a diner and since teachers don’t generally work in diner’s, GSESA can reduce the verb pool to things a teacher may do in a diner: order, pay for, have, eat, buy, while ignoring others: prepare, serve, cook. Good old BASIC coding: if x then y. ESL teachers would call that the First conditional.
Offer a solution
So, while I am writing to GSESA’s interests, I’ve also got to keep in mind that secondary audience, the users looking for answers. I am currently “creating content” which, run through the “problem solving” of GSESA, will make some kind of sense, answer some kind of question a human being searching with Google is asking. Kind of gives the feeling that I am helping the end user get the information he/she needs to get on their lives. But remember, what I am writing is the bait….
The thing is, I am not answering any questions posed by real people. In writing the explanations, I am posing the questions myself. And that’s only in part because I think someone may ask the question. More importantly, I have been told that GSESA finds it easier and more productive to index and prioritize questions followed by answers. How does that work, then?
How does it work?
It begins with “headings”. I don’t use them too much in my blogs, but this platform I use, WordPress, has a feature that allows me to place such “headings” throughout my work. Headings have been around for a while, don’t know if WordPerfect offered them, Microsoft Word has for decades.
A heading is simply a headline, perhaps a section title or chapter title, which you “code” to reflect a visual style (maybe larger font, or bold faced). That code is not only for formatting, it flags the text as a type of heading, the common ones being: Title, H1, H2, H3, etc. GSESA loves code, that’s its native language. No superfluous words to ignore, no meanings to understand, no sentence structure, no assumptions to be made, just nice, clear, clean code.
So, if I put a heading in H1, GSESA will assume that “them thar is important words“. And if I pose those words as a question, using one of the question words when, where, why, who, what, how, GSESA knows it can ignore the next word automatically, that will just be an auxiliary verb like do or have or be, functional words for people, straw words for GSESA. Any words that follow will probably be keywords, so GSESA now knows from the outset what will be talked about under that header.
But it doesn’t just stop there. I may ask:
“What is the price of tea in China?”
GSESA will think:
- what: question about a thing
- price: the thing: monetary value
- tea: keyword
- China: a particular geographical location
If I begin the following text with:
“The price of tea in China in 2023 is $45 a kilo.”
GSESA will have both the question and the answer, GSESA will assume that the writer has complied with “Identify the problem” and “offer a solution”., GSESA get all excited and give the post brownie points that will move it up the ranks!
But, if I write, after that question on tea in China:
“Since the days of Marco Polo, the importation and exportation of goods from the Far East has had a significant impact on world markets.”
GSESA will wrinkle its brow. That’s not an answer. GSESA may give me a second chance and read a bit further. If I get that price and China into the next sentence, well, I may only loose half a brownie point. If I don’t get around to the price of tea in China until the next paragraph, well, GSESA may not even bother reading that paragraph, GSESA will ask:
“What does that have to do with the price of tea in China?”
Zing! You’ve lost all your brownie points.
Rake in the bucks
As long as we writers write to make GSESA’s parsing of our writing easier, as long as we are trying to earn those brownie points, then, yes, we should be worried about AI.
It generally takes me between 9 and 15 hours of work to get a post approved. The process is overly-complicated:
- I write an outline (around 1 hour)
- I submit an outline.
- An editor and the project manager look at the outline and suggest improvements.
- I make the improvements. (another 15-30 minutes)
- With luck, the improvements are accepted and the outline is approved. (My new team manager recently tossed out the tired old “the outline is half the work!” meme, I wanted to send back a snappy retort, but my stoic attitude saved me from the uselessness of doing so!)
- I write the post and submit it to both the editor and the project manager. (another 3-4 hours)
- They “edit” the post and send it back with suggested revisions.
- I do those revisions and send it back, hoping for final approval. (another hour)
All three of us on the team are fairly expert in the field and fairly competent writers. Though there are some differences in opinion on factual interpretation of the content itself, we easily debate those points out and manage to come to negotiated agreement.
The majority of the revision my work suffers, though, is because of GSESA. Comments begin with things like “For SEO we need….“; “Make this heading a question….“; “You need to structure the post like this for Google….“. All these lead to unnecessary revision time, no matter how closely I follow the scheme, it is assumed that something is not on tack. Plus, Google changes the rules all the time….
At the beginning of the project, we were encouraged to develop a voice, to be creative, friendly with the end user, use images, gifs, even try to become a “mini-influencer”. Then the company hired a “SEO Guru” and we are being forced into a mold, a structure, a “content creation” straightjacket that has turned our writing into a lesser version of what AI could whip out in mere seconds (remember, it takes me hours…. well, the actual writing only takes 5-6, it’s all the “revisions to meet SEO standards” that takes up most of the time, including special formatting, also for GSESO).
So, yes, the company will probably finally figure out that AI can whip out those email-harvesting posts they need for a fraction of the marginal pay they are currently compensating my work with (I’ve just read that bicycle fast-food delivery people have been given a minimum wage of around $17/hour in NYC, that’s pretty close to more than I get paid for these posts. Having done the bike messenger thing when I lived there for about a month, I’m tempted to do a pro/con on the stress level of both…. messenger work might just be less stressful!).
What can be done?
Well, following that scheme very liberally, you are my audience, you may share the problem I have, that our writing has been funneled into looking like AI’s slightly slower older brother and AI will be able to do it much better. Our jobs are in danger of being taken over by AI! Danger Will Robinson!! Now I’ve got to share some kind of solution, haven’t I?
Stop writing for GSESA
Forget about Google. Google will find you or Google will not find you. Google is not a reputable audience. Google doesn’t care if you have a clever turn of phrase, if you can handle plot or dramatic action. Google won’t leave you a kind or devastating review after having read your work. Google is a machine. Do you yell at your coffeepot when the coffee comes out bitter? (Well, maybe you do, but that’s something different.) Do you make coffee for the pleasure the cup feels as you pour it in?
Write first for yourself….
Then write knowing that you are not alone in this world and that perhaps someone else can connect with you through your writing. Stop trying to solve everyone else’s problems for them. Show your knowledge, share your knowledge, build a reputation, sign your name only to that which is your writing, not that which is SEO writing. SEO is a blurb on a flap. What you are writing is that which the blurb superficially describes, with the objective of piquing the interest of a potential buyer. It is the bait, your writing should be the catch, not the user.
Oh so much more to say about this, right?
Cheers,
revel.
revel arroway / ctohp publications 
One thought on “Writers, are you threatened by AI?”