Skip to content

GPT-3: how artificial intelligence might burn out the social media

July 31, 2020

Thanks to artificial intelligence, this blog and millions of other utterances will soon sink into a swamp of incredulity. As in, nobody will know for certain whether the words I’m writing come from me or from a computer. Thanks to a miraculous piece of technology called GPT-3, all I have to do is come up with a few sentences, and the computer will complete the post. And, based on results I’ve seen, make perfect sense – perhaps better sense than me.

I wish it had been available ten years ago. I would thereby have been saved the effort of writing over a thousand posts and a million words. Though probably I wouldn’t have bothered.

The implications of GPT-3 and its even smarter successors are mind-boggling. Imagine you wrote a letter or email of complaint to the CEO of a utility or mobile phone company. The software would be able to read your letter, make sense of it, and draft a meaningful response rather than the blatherous vanilla reply that usually comes from an executive far too busy to bother themselves with our particular problem.

Now imagine that you run a vlog that posts to YouTube. I’ve often thought about doing this using the content that you see in this blog, but to be honest, it’s a hassle. I’m not particularly photogenic, at least not at six in the morning. Though I like the sound of my voice that’s no guarantee that anyone else might.

So, using another technology called Deep Fake, that creates utterly convincing videos of people who either don’t exist or do exist but have had their heads substituted by someone else’s face, imagine that I could hire, say, the face of Tom Hanks and replace my voice with that of Richard Burton or Anthony Hopkins.

All I have to do is feed in the words, and Tom and Tony do the rest. Fine, so long as people realise that the words come from me. Also fine, so long as the ever-reasonable Hanks unwittingly lends his face to some right-wing shit-stirrer who wants to summon the militia from the hills in order to defend Donald Trump after Biden’s whipped him in November.

In fact, what if the social media is so saturated by AI-generated stuff that’s utterly plausible? If we’re no longer aware that a real human has anything to do with 80% of the content we view on the social media, as opposed to the rabble of fairly obvious bots that currently infest Twitter and the like, then the human content eventually gets crowded out. At that point you’ve arrived at a stage that often occurs in an epidemic. A particularly virulent bug kills everything in its path, and eventually runs out of organisms that it can infect. It therefore dies out, because it’s too successful for its own good.

A similar situation could also occur with the social media, because I doubt whether most of us would want to be fed a constant diet of AI-generated opinion, even if it actually makes more sense that most of the human-created stuff that it’s replaced.

Then, according to a blogger who’s posted an interesting article on the implications of GPT-3, the only spaces in the internet where we can be sure that we’re looking to content that has come from humans will be private areas – probably at a cost – in which the authenticity and humanity of the contributors has been verified.

The likes of Mark Zuckerberg and Jack Dorsey of Twitter are unlikely to be very sanguine about this prospect, because it threatens the entire existence of their current businesses. They either create these private spaces or they die.

I’m making a very big assumption here. It’s that people who hoover up stuff on the social media actually care whether the stuff they’re reading or looking at comes from a human or a computer, so long as it tells them what they want to hear. If they don’t care, then the social media will be reduced to a sump of propaganda that will serve even more than today to amplify people’s prejudices.

An interesting prospect. Fake news burns itself out? The mainstream media regains its credibility?

Whatever the outcome, I suspect that we will be hearing much more about GPT-3 and similar technology for the foreseeable future.

2 Comments
  1. Just as a matter of interest, why do you call yourself “59steps” not 39steps? A la John Buchan.

    • Because I was 59 when I started it, and I’m quite content to remain 59!

Leave a Reply

%d bloggers like this: