Designing Minds

Not Just Tools

Last week, I attended a demo of an AI feature for defining employee objectives in an HR platform. In short: managers are asked to enter a few keywords or rough ideas within the SMART framework (Specific, Measurable, Achievable, Relevant and Time-bound), after which the AI generates perfectly fitted objectives.

To be honest, this demo raised far more questions than it provided answers.

First of all, starting from a methodology makes perfect sense. It provides a common language and helps structure thinking. But should it really be up to a SaaS platform to decide what that framework should be? A company may very well prefer an approach such as CLEAR, or even adopt an entirely different framework that is more consistent with its own managerial culture.

The second question led me to reflect on the kind of cognitive experience we actually want to design.

Here, AI steps in at a very early stage. Is this really the way we want to proceed? I have a motto I come back to often: just because something can be automated doesn’t mean it should be. The person giving the demo replied that this was by no means a mandatory feature, that managers could simply ignore it or edit the output whenever they wished. That is true. But whenever we design a tool, we also shape the way it is used. If AI generation is offered from the outset, it short circuits the very purpose of the exercise. The manager may provide an initial idea, but the tool does the elaborating, the clarifying, the structuring, the thinking. Instead of developing our own reasoning, we end up merely validating a generated one.

I’ll leave you to imagine the possible consequences over time: people gradually losing the ability to think things through on their own, skills quietly fading, ideas starting to look the same everywhere, not to mention a growing sense of not really owning your own work, with, at best, a vague feeling of being a fraud, and, at worst, checking out entirely. Quite a vision of progress.

Now imagine a different design. The company chooses the methodology it wants to apply and the manager writes their objectives themselves, as they normally would, without any writing assistance. Only once this work has been completed does the AI step in to analyse what has been produced. It might, for example, point out that an objective lacks specificity. Do you see the nuance? On the one hand, we optimise a task. On the other, we invest in the capabilities of the person carrying it out.

What if, instead of talking only about performance, use cases or minutes saved, we asked ourselves a different question: after six months of using the tool, has the user developed new skills? Has their reasoning become stronger? If so, how do we measure that? What if we stopped thinking of learning as a succession of annual training modules and instead embedded it directly into the tools we use, so that every interaction became an opportunity to improve? Lifelong learning would become embedded learning, and learning itself would become a property of the tools.

For instance: The HR platform teaches managers how to set better objectives. The presentation tool teaches people how to tell a better story. The CRM teaches salespeople how to better qualify customer needs. The coding assistant teaches developers how to write more robust code.

Do we want tools that gradually replace human judgement, or tools that strengthen it? I have always found it curious to hear that AI should be used "with good judgement". Good judgement is not something we bring to AI fully formed. It is something we develop by repeatedly having to think, decide, make mistakes and improve. And yet this is precisely what many of today’s AI experiences short-circuit, by providing a ready-made answer before users have had the opportunity to think for themselves.

In fact this is a quiet return to a timeless truth: it is better to teach someone how to fish than to hand them a catch. The greatest mentors, managers, parents and teachers have always shared the same ambition. They do not seek to make themselves indispensable. They work to make themselves obsolete by preparing us to stand entirely on our own. Our AI products should aspire to do the same.

In the near future, I suspect this will become the ultimate litmus test for the workplace. We will judge the greatness of a company by whether its technology seeks to capture our attention or truly set our potential free.

MD