Confession: My Life as a Chatbot

For some years now, I have made much of my income as a human chatbot. That is to say, I write texts that nobody will read. Ideally, nobody would be writing them either. I am the second-best option after nobody. 

I consider myself a good writer, and I have worked hard for a long time to write good stories and essays. But I get paid much, much better when I am a chatbot, churning out what I think of as “my bullshit writing.” I am one of many writers who make money by moonlighting as a chatbot, freelance writing for corporate clients, in order to spend time on the low-paying or non-paying work of writing poetry, fiction, criticism, etc.

As I’ve mentioned already, there are two chief similarities between my own bullshit writing, on the one hand, and the output of a chatbot on the other. The first similarity is that each type of writing lacks an author. The second similarity is that each type of writing lacks a reader.

What do I mean by “lacks an author”? Well, when I freelance for a corporate client, I am not writing in my own authorial voice. At least, I am trying very hard not to. The ideal final draft would contain no trace of my style, nor of anyone else’s. To be clear, it’s not that it requires a different voice—a more professional or constrained or perky one. Rather, I’m trying to write in the voice of the brand. But a brand does not exist. It’s just the shorthand word we use to describe a group of ideas and symbols. At the risk of stating the obvious, a brand has no voice, because it does not speak. Thus, the “voice of a brand” really consists of many individuals’ work, pureed into smooth indistinguishability like pears and bananas and chicken in a jar of baby food, superimposed into inoffensiveness like those photos of composite human faces. The obscuring of the individual writer’s voice is paramount in corporate copywriting. In fact, it’s the driving principle, the style that defines the genre. One becomes part of a chorus of copywriters, all striving to speak in the same disembodied echo of a voice that never existed.

This is, of course, exactly how chatbots work. ChatGPT (to state the obvious again) has no voice. Its verbal output is made up of the mashing-together of many individual voices. This pureeing process is what grants the AI product (or the company behind it) prestige, because the writing seems to emanate from a special new voice, conjured by the mad scientists at OpenAI or whatever. But in both cases, the “voice”—the “brand’s voice,” the “robot’s voice,” is an illusion, albeit a hard-won and impressive illusion. It’s made up of many other voices, speaking in unison, from within a branded interface.

And what do I mean when I say that these two types of writing have no readers? I mean what I say: nobody is reading my bullshit writing. A lot of people probably scroll past it while looking for something else, dimly aware of the words. But they do not read it like literature. They also don’t read it like recipes, or car manuals, or Instagram captions. Certainly, nobody has a favorite piece of the bullshit writing, or recommends any of the bullshit writing to their friends. Nobody ever bookmarks it. I can say with total confidence that nobody is asking for or seeking out the bullshit writing. 

I am sure the writing has some kind of demonstrable value to the companies or they wouldn’t pay for it. Most writers do not get paid very much these days, but they get paid way better when they are writing stuff that nobody will ever read. There are various reasons I can think of that the writing I produce might be lucrative for the company: the bullshit writing provides a background for ads. It offers space to link to products. Above all, it proves to customers that the company can afford to hire a writer. Its very superfluity conveys a status to the people who commission it. It demonstrates that the company is successful enough to waste money on something nobody asked for, and this helps to convey the company’s legitimacy and power to potential customers, investors, etc.

Similarly, as far as I can tell, nobody really wants to read something written by a chatbot. The novelty hasn’t totally faded yet, and sheer curiosity might drive some people to read a ChatGPT story or essay. But I don’t really think anyone cares to, like, curl up with a ChatGPT novel. And again, this is true of non-literary writing too, maybe to an even greater extent. Absolutely nobody in the world is excited to speak with a chatbot customer service representative. With a few exceptions, things like a grocery list, chatbots address problems that don’t exist by producing writing nobody wants to read.

But (you say) the chatbot can solve one problem! The chatbot can write your bullshit writing! This is sort of true. It’s fine that nobody wants to read Chatbot writing, because after all, nobody wants to read my own bullshit. Similarly, nobody really wants to read an eighth-grade essay on The Great Gatsby. The chatbot can, in other words, help you cheat. They are not in actuality good enough to do this effectively—many a student has been failed for an obviously AI-generated essay. But in theory, if the technology got very good, which maybe it will and maybe it won’t or can’t, it could solve the problem of writing things nobody wants to write or read anyway.

But this isn’t a real problem. The moment a robot can write your five-paragraph essay or your bullshit corporate copywriting, the product loses its value. That’ss because the value lies, not in the text itself, but in the proof of the text having been written and labored-over. These texts exist to be submitted, marked up, invoiced for, etc. (I do of course think a school essay has value, in the sense that it helps the student learn—but it does so in the process of being produced and graded, not because it is enjoyable or edifying to read). If chatbots take my freelance jobs away, it won’t be because they do the work more cheaply and quickly than me, even though they do. It will be because their existence nullifies the value of the work in the first place. The companies will no longer be able to use the superfluity of the text as a demonstration of their own superfluous budget, because the text will be cheaply available for anyone with an internet connection.

Right now, an entire PR industry revolves around the creation of writing by nobody and for nobody. The people creating it, by and large, are people who want to be creating something else (screenplays novels, poems). Nobody enjoys reading these PR articles. They, too, would prefer to be reading novels or watching TV or even reading poetry. 

I suspect that chatbots are not coming to take my jobs—because they might never be good enough, and because industries will find ways to shift the presentation of the bullshit writing so that it maintains its prestige. But they might. I’m holding out hope.

It will obviously be bad for me, in the short term, if this work disappears. But then, would it really be a replacement at all?  I’ve spent years, unknowingly, working to resemble ChatGPT. I only realized what I was aspiring toward when the chatbots came along, turning out the absolute platonic ideal of readerless, writerless, valueless writing. Is the chatbot replacing me, or was I just standing in for it all along?