Antonym: The Pedalo Pedagogy Edition

Mind that metaphor! We're going on an AI learning odyssey.

There were parts of the brain which were constantly at work – as if construction processes were under way: an operation which was almost painful, which sliced up the day, drew information from the cut surfaces and put them together to form a new pattern, a new universe.

– On the Calculation of Volume I, Solvej Balle (tr. Barbara J Haveland)

This felt familiar – it sounds like the way learning sometimes feels.

Solvej Balle’s International Booker nominated novel is about a woman who wakes up to the same day again and again (yes, it’s a Groundhog Day remix with A-level literature vibes). Though the world remains (almost) the same with each repetition, she evolves. Her perceptions shift. The familiar takes on new meaning.

In the last few months, I’ve been working one-to-one with a group of senior executives – from communications and marketing roles to tech, finance and operations leaders. They’re living in the familiar cycles of corporate life – meetings, quarterly reports, strategic plans – with growing awareness that their well-honed routines are being reframed. What’s changed isn’t always the work itself but their perception of what that work could be in a world shaped by artificial intelligence.

The first step on this journey is deceptively empowering. Learn to write a prompt, grasp how large language models (LLMs) work, and suddenly you have a dazzling assistant. Emails get written faster. Research is less of a slog. You’ve got an intelligent companion – a tool that seems to lighten the load. But the first taste of AI literacy is just that: a taste.

Actually that Makes it sound like they're all marching in lockstep. Some are now on their sixth or seventh session and becoming quite advanced. Others stopped at one or two and are convinced that they've learned all they need to. This week, it’s become clear to me that we – I – need to better articulate what lies beyond.

Explaining AI is not enough. Understanding only begins when the terrain gets muddy.

The Pedalo Pedagogy

P-p-p-pick up a perfect alliterative allegory. Adults learn like pedalos – slowly, awkwardly, and with much splashing before momentum builds.

People often ask for "training" – a ninety-minute Zoom or a glossy slide deck – expecting transformation. But these are merely entry points. What comes next is deeper, and more demanding.

We’ve always stressed that deliberate practice is essential to build AI literacy.

A post from an academic at LSE (name withheld pending proper citation) used a memorable phrase:

You have to tell people: the gate you’re walking through leads to a vast field. But that’s not all there is – if you look carefully you will find another gate in the far corner leading to an even bigger and more wonderful field. And behind that field? Another. You can’t shortcut the sequence. Walk before you run; pedal before you glide.

Hang on, are we crossing a field in a pedalo?

Yeah. We mix metaphors here. My house, my messy conceptual frameworks. Deal with it.

Actual footage from a recent AI workshop where some metaphors became tragically mixed leading to a land-pedalo race mishap.

From Literacy to Kompetenz

A friend of mine in Munich reminded me this week that in German, the word "literacy" is reserved for reading and writing. When it comes to what we might call data or media literacy, Germans use a more robust term: Kompetenz. Competence. The kind that implies skill, judgement, application. And envy stirred in me. That’s what we need.

"AI competence" sounds urgent. If you lack it, you don’t sound undereducated – you sound unfit. AI illiteracy might be a nuisance; AI incompetence could be career-limiting.

Competence is the point at which understanding becomes repeatable, productive action. Literacy lets you read the map. Competence lets you navigate it, make decisions, and lead others. It manifests in artefacts – marketing strategies that use synthetic data, budgets that anticipate model costs, board reports that surface AI risks.

Getting there requires more than curiosity. It takes deliberate practice and structured feedback. The fastest learners? They treat AI like learning a language and…

• Use it daily.

• Run small experiments.

• Embrace critique.

• Tackle incrementally harder challenges.

They also make room for discomfort – the very sensation Solvej Balle describes so well. The ache in the mind is a sign that new pathways are forming. AI literacy opens the gate; AI competence is the long walk across the field – boots muddy, muscles sore, but the horizon growing clearer with every step.

The awk-cringe-crumpling of the Grok

Over at Musk’s AI company Grok people were getting confused by the model’s responses this week. The damn thing would not shut up about “white genocide in South Africa” (note: there is no such thing) even if people were just asking for a restaurant recommendation or how to build a nuke in three easy steps.

“Someone” had changed the system prompt for Grok, the company admitted.

In our house, “someone” is code for a person we are specifically blaming for for something. – “Someone didn’t take the bins out this morning”, “Someone forgot to buy dog food”, etc.

In the case of Grok’s mysterious prompt re-engineering, someone is possibly the FKA-a-genius who supposedly runs the place.

That’s all for this week…

Thank you for reading. May your pedalos remain ever stable.

Antony