The calm absurdity of ChatGPT

A public sevice for the woke management team of a retail company closing its stores due to high levels of theft and violence. pic.twitter.com/zfDn3bvKbU

— Marc Andreessen (@pmarca) December 11, 2022

Even before reading the screenshot attached, I knew this was certainly not meant as any kind of public service or guide for dealing with business crises. It was a finger-pointing “gotcha” that such thoughts and prose could even exist.

“I believe that defunding the police and letting criminals out of jail is a necessary step to creating a more just and equitable society […] We are all aware that theft and violence have been on the rise in our stores and we have made the difficult decision to close these stores.”

The calm absurdity of the text was the point. The words themselves reveal the untenability of such a mindset, it's the kind of shocking-but-real post spread by its critics (not its author), and sure to rise to the top of our social media feeds.

Yet in this case, it only took another split second or two to realize it wasn’t real at all. It was ChatGPT.

If you’re on Twitter and interested in technology, you’ll probably also know about ChatGPT: the stunningly powerful Artificial Intelligence-driven natural language bot launched earlier this month, where you submit a prompt and the machine (usually) spits out relevant and uncannily humanish content.

Its proponents say it has the power to substitute human efforts on everything from computer programming and medical diagnoses to school essays and audio-visual entertainment. It may be the first real threat to the two-decade reign of Google.

The arrival of ChatGPT feels like a deeply unsettling watershed.

So far, public experiments have mostly been exercises to see what AI can do with textual information and language. It poses enormous questions for the news industry, politics and entertainment, academia and literature. I hesitated at first to share it with my daughter, fearing she might be tempted to turn to it for her university assignments.

For those of us in the business of information and ideas and language — and how they travel online — the arrival of ChatGPT feels like a deeply unsettling watershed: The AI that has been quietly creeping into our everyday lives and jobs has suddenly given us a handy peek at the full potential of what it can do. This might very well eat our world.

But for now, it is largely working as a kind of fun new feature for social media. And as for certain categories of online content, mostly those that are text-driven, Twitter is the perfect platform for ChatGPT to say “hello.”

Not all fun and games

Among the other examples of machine-generated text are impressive and sometimes comic responses to such prompts as, “write a rebuttal email to a health insurance denial” or “produce some social media self-promotion for the most minor achievement.”

Of course, it’s not all fun and games, as the bots’ arrival on Twitter comes as the social platform is shaking under its troublesome new owner. And yes, Elon Musk also happens to be the co-founder of OpenAI, the company behind ChatGPT. For our purposes here, it’s also worth noting that Marc Andreesen is an investor in Musk’s takeover of Twitter — and the pair are no doubt aligned on the aforementioned woke debate.

But here is where this (human-produced) textual content veers in another direction. For the previous paragraph, it turns out, ended on a false note: There is in fact no “woke debate.” That too has been artificially produced, both by humans and the algorithms that do decide what rises to the tops of our feeds.

It’s not that the issues behind the “W” word aren’t worth debating. They are — and like most of my colleagues in the information business, I tend to agree with the basic premises of what @pmarca and others espouse about the importance of having the space to openly debate and disagree, and even make mistakes.

But I have also watched how this would-be debate is fed (by all sides) into the meat grinder of social media grievance, and comes out the other side, distorted at best, and all too often utterly sapped of any meaning.

Woke v. Fake

“Woke” is a word that was almost immediately hijacked by its critics, rapidly ending its utility in public debate. It recalls any number of more or less useful phrases and hashtagged terms in recent years, from #MeToo and #BLM to #FakeNews and #NoVax. Some burn brightly enough for a moment to actually serve a purpose in the public space — and public conversation. Others cause more harm than good.

Still the machine of that media-and-social-media-driven public conversation is ultimately hard-wired now to weaponize new language, for good and bad ends, before it ever has a chance to find its proper place in our vocabulary.

The @pmarca “woke” prompts (he had others) for the new AI tool fueled more rounds of shared outrage from his like-minded Twitter followers, many of whom had no idea they were … well fake.

We've seen the power of language to resist our darkest moments.

The rest of us are left to roll our eyes. For the insidious flipside to weaponizing words is depriving them of any meaning. Or as George Orwell famously put it: “But if thought corrupts language, language can also corrupt thought.”

So how will the addition of tools like ChatGPT plug into this current dynamic of our language? So far, at least, it mostly lands as a bad joke — though in our example, it’s not Andreesen and his followers laughing at the would-be woke crowd, but the bot laughing at us all, starting with the illustrious prompter.

There’s another clue in the AI trolling episode of what to expect next. As we wait to see how and if these tools are used in the news industry, in academia, in entertainment, there’s one place where we know it will be used first: public relations. Andreesen and others are doing it as satire, but the bot is already quite far along in its ability to write bonafide PR; and the mediocre professionals in that field are no doubt ready to have it churn out press releases and vacuous “thought leadership” posts.

The power of human communication and free speech

As it gets exponentially easier and cheaper to produce, the optimist in me believes, eventually, the less we will believe it. The pessimist fears we may destroy language (or each other) in the meantime.

In this dark moment, the light we might see in the new AI-generated communication is its reminder of what's at stake in human communication — and perhaps quiet some of the extra barking we’ve been doing the past 10 years.

Indeed, we’ve seen the power of language to resist even in the darkest moments. You may remember White House spokeswoman Kellyanne Conway’s attempt at introducing “alternative facts” in one of her defenses of Donald Trump. It didn’t fly, and our dear “facts” held its ground in Webster's.

In the current non-debate over wokeness, another term is now getting similarly thrashed around — with Elon Musk telling us that he bought Twitter to save it: Free speech. But it too will resist all the noise ... so we can continue to argue about it!

To conclude the way we began, I've prompted ChatGPT with another question: Do bots have free speech rights? On cue, the machine spits out some dry but undoubtedly well-composed prose that could sneak in undetected to my daughter’s university exam, or even the news website I edit. I’ll save you time with a human summary of the bot's three-paragraph answer: No.