The Age of Suspicion · Ben NorthmoreBen Northmore<br>Ben Northmore<br>Developer journal.

Engineering manager building useful things, automating the boring stuff, and sharing what I learn.<br>Wales, UK

AI didn’t just make content cheap. It made authenticity expensive.

Ten years ago, if you read a well-written article, you probably assumed the author knew what they were talking about. Today, one of the first comments is just as likely to be “This reads like ChatGPT.” A beautiful illustration? “Midjourney.” An impressive voice recording? “Probably cloned.”

Even when something is completely genuine, we now instinctively wonder whether it is. Not because AI can create convincing content. We’ve had Photoshop, CGI and editing tools for decades. The change is that the default assumption has shifted. We’ve moved from trust to suspicion.

The debate we rarely unpack

There’s a tension running through the whole AI argument that people rarely state plainly.

On one side: using AI is cheating.

On the other: if the information is correct, who cares where it came from?

Spend enough time on Hacker News or Reddit and you’ll see both positions treated as obvious. I’ve watched threads where the debate isn’t whether something is useful. It’s whether AI was involved. People ask “Did you actually write this?” before they ask “Was it worth reading?” Some circles treat AI use less like using a calculator or an IDE and more like a confession.

I think the second question is the more interesting one. Not because tool choice is irrelevant. Because it forces you to separate what you’re judging.

The uncomfortable question

Imagine you read a blog post that taught you something genuinely useful. A week later you discover it was 95% AI-generated.

Did the knowledge become less true?

No. The bits in your brain didn’t disappear. The code still works. The idea is still valuable. So what actually changed?

Maybe we don’t value information. Maybe we value human effort, or originality, or expertise, or authenticity. Those are different things, and we often collapse them into one vague complaint about AI.

When someone asks “Did AI write this?” they might actually be asking several different questions at once. Should I trust the author? Does the author understand this? Is this original? Is this worth rewarding? Those aren’t questions about the information itself. They’re questions about the person behind it.

Knowledge has no provenance. People do.

The facts don’t carry a history. Authors, expertise, accountability and trust do. That distinction matters more than picking a side in the cheating debate.

A thought experiment

Imagine two articles. Article A was written entirely by a human. Article B was generated with AI, then carefully verified and edited by an expert. You don’t know which is which. Both teach exactly the same concepts. Both contain zero factual errors. Both are equally enjoyable to read.

Which one is better?

If you can’t answer without knowing who wrote it, that’s worth sitting with. It suggests that at least some of what we call quality isn’t in the text at all. It’s in what we believe about how the text got there.

When provenance is evidence

This isn’t an argument that provenance never matters. There are domains where knowing who produced the information tells you how much confidence to place in it. Medical advice. Scientific papers. Legal opinions. Historical research. News. In those contexts, the source isn’t decoration. It’s part of the evidence chain.

So the more interesting claim isn’t provenance doesn’t matter. It’s that people increasingly treat provenance as more important than the content itself, even when the content can be independently evaluated. We reach for the tool story before we’ve finished reading the argument.

Engineers already judge outcomes

Software engineers have been living with a version of this for years.

Nobody asks whether code was written in Vim, IntelliJ, Copilot or Cursor. They ask whether it compiles, whether the tests are green, whether it’s maintainable. Code has always been judged primarily by outcomes. The diff is evidence. The tool is incidental.

Writing is one of the few domains where people still obsess over how something was produced. A polished demo no longer proves engineering ability. A beautifully written post no longer proves someone is a good writer. Every impressive piece of work now comes with an invisible asterisk: maybe they made it, maybe AI did.

For the people creating genuine work, that uncertainty is frustrating. Artists post time-lapses. Photographers upload RAW files. Developers livestream coding sessions. Students are asked to explain assignments they genuinely wrote. Proof is becoming part of the creative process.

The proof economy

For years, technology has steadily reduced the cost of creating things. The printing press made copying information cheap. The internet made publishing cheap. Social media made distribution cheap. Generative AI has made creation...