AI harms your reputation

People who use AI to outsource their thinking and communication are silently losing respect from a subset of their colleagues, potential business partners, clients, readers - and they don’t know it.

When someone sends me a text that was clearly AI-generated, or even worse, when they send me “here’s what Claude told me about X”, in my mind, I just downgrade future expectations related to that person.

If I want to speak with an LLM, I can speak with an LLM myself. I don’t need a human-shaped proxy. I know at least one case where a business project didn’t go through because one party decided that speaking indirectly with an LLM wouldn’t go well when the project actually got complicated. If you cannot think through things when they are simple, and you cannot signal this ability to others, then it is unlikely that you will get this ability in the future, once things start requiring even more human attention and engagement.

In another business, I know of a case where someone who was paid for their insight and decision-making decided to start mass-producing AI slop, thinking themself “AI-enhanced”. This resulted in shipping broken features, which cost the business trust from a big client. And that person already had quite a damaged reputation within the company among their colleagues, because of all the slop that the colleagues had to fix.

In both these cases, there’s an actual financial cost to the misuse of AI-generated text. And of course, there’s the reputational cost both for the business and for the person, which might outweigh the financial cost. Once you get the reputation for being “the guy who just copy-pastes everything to/from chatGPT”, it’s hard to get rid of that reputation. People won’t tell you that, but they will absolutely take that into account in every interaction they have with you.

You might also live in a bubble where this type of communication is not seen as too bad, and so you might temporarily experience positive effects from fully outsourcing your thinking to OpenAI’s servers. But it’s temporary, and there will come a time when you will come to know the true price.

I myself have probably sometimes done the same mistake, but I don’t think I’ve ever been as egregious as some examples that I’ve seen around me. In any case, it’s not a mistake worth making. The productivity gains in using AI do not come from copy-pasting responses from chatGPT when you’re communicating with another human being. The productivity gains come from quickly writing automations for repetitive work; from taking quick stabs at things that would have taken many months in the pre-LLM era. You have a bunch of unstructured data, no real skills for data cleanup, but you would like to extract at least some insight from the data. Use an agent! You’ll get something, and it’ll probably be more useful than not having anything.

But if you’re trying to strike a deal with a business partner, and you’re just passing on what chatGPT tells you, you’re not getting a productivity gain. You’re potentially (probably, even) just losing their respect.