On the fatal naïveté of the AI shepherd

Building AI into your work is basically saying 'automate me'.

As well as being told to learn about AI, we’re also encouraged to build the new technology into our workflows. Add value by combining human thinking with AI. Embrace the bright new future of human-machine collaboration.

Confronted with their own obsolescence, the endangered professional aspires to become a ‘AI shepherd’: an intermediary between AI and its end users who adds value by translating business needs and goals into prompts and finessing the results.

‘Shepherd with His Flock’ (c. 1820) by Jean-Victor Bertin

As a copywriter, for instance, you take a standard brief for an ad or a website, translate it into a prompt (or even develop your own custom model) and then use your existing skills to ‘humanise’ the output into something better than the client could achieve for themselves.

It’s worth noting that this approach is more of an uneasy compromise than an advance. Nobody, not even AI evangelists, seriously claims that it gets better results. It’s just a way to reconcile AI-driven demands for ‘fast and free’ with a residual respect for craft. (And I’m not even getting into the ethics of it all.)

However, I also think it’s a short-term solution at best. Because it doesn’t take account of the way AI is likely to develop from here on out.

First, AI will get deeper, in the sense of improving its functionality in writing, coding, or whatever. That will possibly help AI shepherds. However, it will also get broader, in the sense of incrementally expanding its scope to encompass adjacent tasks that require more judgement, analysis, insight or creativity.

AI shepherds are big flashing neon signs showing AI companies which tasks they should automate next. The more value you add and the more money you make as an AI shepherd, the more likely your role is to be assimilated by AI.

Here’s what AI firms will think. “So, you’re adding value by using AI? Converting that value into profit? Great! We’ll take your role and build an AI agent to do it instead.

“Thanks for all those carefully crafted prompts – they’ll be really helpful. We’ll hoover them all up and use them as the basis for the kick-ass algorithm that powers our new ‘content architect’ agent.” (Or ‘marketing manager’ agent, ‘solicitor’ agent, ‘recruiter’ agent – take your pick.)

Then they’ll promote their shiny new agent to all those clients you’ve helpfully found on their behalf. It’ll be like a faster, brighter new rival who works for free. Welcome to the brutal reality of disintermediation, where many copywriters, translators, journalists and customer service folk have been living for a while.

It bears repeating that AI is not your friend. Nor is it some benign force of progress, a rising tide that lifts all boats. It’s a commercial product, and its creators’ primary goal is to increase power and profit for themselves. If they help you along the way, it’s only because it will ultimately help them.

The lesson should be clear by now. If you’re paying little or nothing for a digital service, the chances are that you yourself are the product – or soon will be.

You’re not using AI. It’s using you.