The Nonsense Machine

When AI talks to AI and nobody's thinking, then we have a problem.

Generative AI has made it extraordinarily easy to produce content. Writing, art, video, strategy documents, research summaries - things that used to take a lot of skill and real time can now be produced in minutes by anyone.

On one hand, this is brilliant. It’s liberating. It removes barriers to entry that have historically kept talented people with great ideas from ever getting those ideas out into the world. If you’ve got a sharp insight but never had the training or access to produce a polished output, you now can. That is a good thing.

But it also removes a barrier that was doing something important. The difficulty of producing exceptional work used to force people to think. To challenge whether an idea was good enough to be worth the effort. To ask: is this a good use of my time? Is this something I want to attach my name to? The effort required to produce something acted as a quality filter on the idea itself.

Think about an artist working on a painting. The reason the finished piece is good isn’t just because it looks beautiful. The artist has likely been on a journey with it - planning, iterating, questioning whether the story they’re trying to tell is worth telling. The difficulty of the craft forced that thinking to happen.

Now I could sit down with a generative image tool and ask it to paint a landscape of Walthamstow, where I live, in the style of Monet. And it would look lovely. But was there any point? Did I have something to say about Walthamstow? Or did I just do it because I could?

Just because you can doesn’t mean you should. And when the cost of producing something drops to nearly zero, the qualifying question of “is this actually a good idea?” tends to disappear with it.

The slop era

This is how we ended up with AI slop. I think most of us first noticed it on social media.

It’s everywhere now. AI-generated comments appearing under LinkedIn posts. Whole sub-genres emerging - ‘grind slop’ where people produce endless content about how hard they’re working. Twitter (no, don’t make me call it X) is slowly turning into Wikipedia, with people sharing incredibly long backstories to things nobody asked about. Content for the sake of content, produced because it’s easy, not because anyone sat down and asked whether it was worth producing.

As humans, we’re quietly getting used to this. Wherever we consume the internet, a big chunk of what we see is second-rate content that wasn’t produced by a human and wasn’t produced with any particular care. We scroll past it. Sometimes we read it, sometimes we don’t. Either way, it fills our eyeballs and takes up our brain space.

Quality human content is still valued - by a lot of us, actually. But we’re all going to be exposed to a growing volume of AI-generated noise regardless. You can scroll as much as you like. It’s still there, and it’s still taking up your time.

The same thing is happening in the workplace

I’m seeing the same pattern play out inside companies.

It’s now incredibly easy to produce strategy documents. To write up competitive intelligence. To generate summaries of every sales call with enterprise businesses in a given vertical over the last month. And, of course, a lot of this is really useful.

But AI is, by its current nature, verbose. It says more than it needs to almost all of the time. I don’t think I’ve ever seen a first draft from an AI tool that was as succinct as it could be. And for people who don’t have a writing background - which is most people - they see a report the AI has produced and think it looks great. It’s impressively structured, it’s detailed, it’s twenty pages long. So they send it to their colleagues.

A couple of years ago, this was impressive. If it was your first time seeing an AI-generated report, you’d have been struck by its structure, its detail, its ability to reference and synthesise. But two things have changed since then.

First, everyone’s doing it now so it’s not impressive anymore.

Second, everyone is aware of the weaknesses. People know AI isn’t fully reliable. So if someone sends you a report that’s clearly AI-generated, you wouldn’t want to bet the house on it.

But I think there’s something more important going on. The tells in AI-generated content are well known by now. And sending someone a long, obviously AI-written report sends a message beyond the content itself. It tells your colleague that you don’t respect their time. That you expect them to read something you may not have even read yourself. That you probably didn’t put much thought into it.

Their reaction will therefore probably be one of indifference - or even annoyance.

The Nonsense Machine

And then the logical next step kicks in. They think: “I don’t have time for this. It could have been two pages. And thirty other people have sent me similar-length reports this week, also produced by AI.”

So they feed it into their own AI tool and ask for the three main takeaways.

And then they say: “Can you draft me a response based on that?”

And suddenly you’ve got two humans who are supposedly working together, who are just proxies for their own AI agents. Two machines talking to each other, with humans in the middle doing very little thinking.

The Nonsense Machine.

The natural next step is that the agents just talk to each other directly. And I’m sure some people reading this are thinking: “Yeah, that’s exactly where we’re headed. Who needs humans?” But that’s an article for a different day.

The more immediate question is whether we should really be outsourcing so much of our thinking. And what is the point of being involved in a project if you’re not actually going to participate?

What to do instead

There are some clear takeaways for product marketers from all of this.

Be thoughtful about whether something is worth producing. Not every insight needs a ten-page document. Not every call needs a detailed write-up circulated to the whole team. The fact that you can produce it quickly doesn’t mean you should.

Don’t outsource your thinking. If you’re writing up strategy or research, spend your own time with the insights first. Get close to the data. Form your own view. Give the AI a hypothesis. Be in a position to critique what it produces.

And think about what you’re sending to your colleagues. Surely the benefit of this technology is that things can be clearer and more succinct than ever before.

As a product marketer, that’s the value you’re there to bring. Clear, accessible language about your product, your customers, and your market - company-wide. This technology is an enabler and an amplifier for exactly that.

So don’t add to the problem. Move past the fact that it’s incredibly easy (and fun) to produce content quickly, and go back to thinking about your end user - whether that’s a customer or a colleague.

Let’s do what we can to reduce the noise, and use this technology for getting to the core of what product marketing is about: providing clarity on how to effectively communicate your products.

---

Rory Woodbridge is a London-based consultant working with European tech companies to unlock their growth potential through world-class product marketing. Get in touch to discuss working together.