Written by AI

A couple of months ago, I wrote an article about A.I. (Artificial Intelligence) generated art.

It’s possible to be glib about it and its repercussions because – well, because I’m not an artist.

But as one who does voice-over work, the idea that a voice, once sampled, could be used over and over without the initial “artist’s” involvement, ever again, has been lurking around the voice talent’s bad dreams for quite a while, but before that, technology created a challenge for the entire industry since it’s now possible for anyone, anywhere, with a computer, a reasonably cheap mic and some software to sit in a closet and record voice-overs. Of course, a good voice or a convincing pattern could still mean the difference between a job and none, but the initial investment and time in a very expensive studio were no longer requirements for most work. What was once a day-long commitment was whittled down to an hour or two.

Now add to it and possibility of using the tone of any famous voice or voice type and simply generating it using A.I.

Actors have rightly looked at what can be done with CGI – computer generated imagery – with some concern. We all know the fame of Andy Serkis, the actor who created the Gollum, the character in The Lord of the Rings cycle. He famously did all his acting in a green suit with tags at particular locations on his body so that his movements could be “skinned” with the body of the Gollum as created by artists. His performance was nevertheless riveting as he crept and leapt and snuck and slunk his way behind a computer-generated mask. But what if you work things the other way around – as some conspiracy theorists speculate – and take a famous face and simply “make” it do or say things he or she never has? There’s a nice little app called “Reface,” and using it plus a clip from a movie of your choosing, you simply tie your face (or a friend’s – or perhaps enemy’s?) and merge it with the movie clip’s main figure for very amusing results. But aside from the fun factor, it works so well and so quickly you have to wonder whether all that you see is “the real thing?”

By now you’ve no doubt seen the clips of actors or political figures merged with someone else speaking for them, again mostly used for entertainment purposes – but with such highly successful results you could easily be tricked.

Not long ago, I went to a writers’ group session, and one of the offerings was an A.I. generated playlet. It was short, and to be honest, not knowing what it was I found it a little boring, though there was an amusing payoff at the end. But to be honest, it never crossed my mind that writing might become an endangered art. The person who brought it to us didn’t tell us that the project had been written by A.I., and though, as I noted, I found it a little ho-hum except for a clever ending, I would never have suspected it to be computer-generated.

At the time of its writing (2022), McKinsey Digital predicted that in the near future, “70 percent of companies will employ hybrid or multicloud management technologies, tools, and processes…5G will deliver network speeds that are about ten times faster than current speeds on 4G LTE networks, with expectations of speeds that are up to 100 times faster with 40 times faster latency.” (What that means is, no noticeable lag in your input and the A.I.’s output.) By 2024, more than 50 percent of user touches will be augmented by AI-driven speech, written word, or computer-vision algorithms, while global data creation is projected to grow to more than 180 zettabytes by 2025.” To put that into some sort of context, my first computer “really fast” computer had 4 megabytes of RAM. One zettabyte is 1 trillion gigabytes. Backing way up, a “byte” is typically a unit of data, the equivalent of 8 binary digits; a megabyte is one million bytes. So you get the picture.

The prediction is simply that computers will get more and more powerful and fast, and more and more of what is said, seen, and written will be generated by Artificial Intelligence.

Take predictive algorithms, like, that spell-checker we all hated so much. Over time, it has become more and more sophisticated and better at correcting, or better put, predicting, the word you’re about to type into your computer or onto your smart phone. The phone lags your computer’s email app by quite a bit, and I find I can often go into just hitting “enter” as the application guesses how to finish my sentences with suggested words or phrases – embarrassingly sometimes doing a better job of finishing my routine email communications more professionally than I might have. That is to say, since its inception, the sophistication and skill of A.I. writing has improved markedly, and will no doubt one day soon be able to emulate the style of say, Dickens, or Poe. Much text has been fed to computers so that they “learn” a writer’s average sentence length, word choices, and other hallmarks of writing that let us quickly determine if a poem is Byron or Yeats as we read it. With that onboard, A.I. will undoubtedly be able to at the very least take a stab at Shakespeare someday.

To play with AI-generated writing, you can check out such apps as Ai Article Writer, Wordtune, or Ai Autowriter, among many others. The apps will either take your copy and “fix it,” or using a few prompts from you, create an entire article or paper or story, and you simply go in and fine tune it, if you wish.

And finally, no, this article is not written by A.I.

Just me.

Nancy Roberts