The Viewpoint: Learning to speak AI

I’ve spent a fair amount of my working life around foreign languages. I worked for years at a German firm, picked up a bit of Japanese, and can muddle through in French and Spanish. Despite all that, the only time I am truly dangerous abroad is when in a bar ordering a ham and cheese sandwich and a beer. Beyond that, things unravel quickly, although I like to think I am at least polite as I butcher the local language.

When people talk about AI around the office, I am reminded of what it feels like to learn a new language as an adult. At first, everything is awkward. You speak slowly, fumble for words, and probably mispronounce the ones you do remember. You are painfully aware of how limited your progress is, especially in front of others.

When people talk about AI around the office, I am reminded of what it feels like to learn a new language as an adult.

Now most people stop there. They decide languages are not for them and retreat to the safety of listening rather than risk the embarrassment of speaking. A smaller group keeps going, however, not because they expect fluency overnight, but because they sense something useful is happening below that uncomfortable surface.

When AI tools first started appearing in my professional life, I was firmly in the ‘wait and see’ camp. Keep a safe distance, tinker at the edges, but don’t dive in. And yet, here we are, I am writing about the topic and have even been ‘vibe coding’ with the best of them in my spare time!

What I see now, both in my own industry and beyond, is that people are splitting, often without realising it, into two camps. The first treats AI as something to be monitored, maybe even feared, but ultimately left for later. They nod along in meetings, acknowledge the trend, but rarely get their hands dirty. These are the people who understand what’s being said, but rarely say much themselves. In short, the early version of Paul.

The second group treats AI like a language to be learned by doing. They stumble, ask clumsy questions, get clumsy answers, and go again. They are not necessarily smarter or working harder, but they are building fluency. And as with any language, the returns are non-linear. Early progress feels slow and unimpressive, yet it soon starts to compound. This is the ‘new’ Paul.

To my mind, this isn’t about the technology, it is more about mindset. The quality of what you get from AI is almost entirely down to how clearly you communicate what you want. That sounds obvious, but in practice it is anything but. Many people approach AI like a nervous traveller abroad, half-formed sentences, waving of the hands, hoping to be understood well enough to get by. They get something usable, but rarely the best possible answer.

Many people approach AI like a nervous traveller abroad; half-formed sentences, waving of the hands, hoping to be understood well enough to get by.

Prompting, the skill of articulating intent clearly and refining it through iteration, is rapidly becoming one of the most consequential professional skills of this decade. It rewards structured thinking and exposes vagueness quickly. The people getting the most value are not the most technical; they are simply the ones who know what they want and can express it precisely, and who are willing to adjust when the answer isn’t quite right. Over time, the models will undoubtedly get better at helping us do this, probably nudging, probing, and surfacing what we are really asking for, but for now at least, the discipline still sits firmly with the human. To my mind, these are not new skills, they are the same ones that separate an average investor from a good one.

I am not here to argue that AI will solve everything, or that the hype is always justified. I have views on all that and this isn’t the place for them. What matters now is not whether you think AI is overhyped or undercooked. It is whether you are willing to sound bad long enough to become fluent.

The difficulty, of course, is that this phase is visible. Early engagement looks messy. You ask naive questions and you get uneven answers. And in professional environments built around competence and confidence, that can feel uncomfortable. But opting out quietly has its own visibility now too.

In my view, those who engage seriously with these tools, who speak before they are comfortable and who accept a phase of clumsiness are building an advantage that will be hard to close later. Not because the tools are magic, but because fluency compounds and judgment improves. Where teams are willing to trust each other enough to be clumsy in public, that learning accelerates and the average level moves forward faster than it otherwise would. The instinct for when to trust an answer and when to challenge it will sharpen over time.

In my view, those who engage seriously with these tools, who speak before they are comfortable and who accept a phase of clumsiness are building an advantage that will be hard to close later.

Without wanting to be overly dramatic about it, I do worry a little for those who choose not to engage at all. I may yet be proved wrong. This wave may break more gently than it currently appears, or stall altogether, but I am not sure that is a bet worth taking. Sitting on the sidelines feels comfortable, even sensible, but it assumes the world will wait for you to catch up.

For my part, I would rather accept a period of clumsiness and learning than discover too late that fluency was the price of staying relevant.

For professional investors only.  This material is not suitable for a retail audience. Capital at risk. This is a financial promotion and is not investment advice. Past performance is not a guide to future performance. Reference to any security is for information purposes only and should not be considered a recommendation to buy or sell.

The value of investments and any income from them may go down as well as up and is not guaranteed. Investors may not get back the amount invested. Portfolio characteristics and holdings are subject to change without notice. The views expressed are those of the author at the date of publication unless otherwise indicated, which are subject to change. Forward looking statements are subject to certain risks and uncertainties. Actual outcomes may be materially different from those expressed or implied.