What Can a Silversmith Teach Us About Using AI Well at Work?

In Season 4 Episode 8 of AI for the Average Joe, Erica and Hayley are joined by L&D expert and author Andy Lancaster. Andy shares how his background in design, furniture making and silversmithing offers a useful way to think about Artificial Intelligence (AI). One example, the planishing hammer, helps make a simple but important point about how organisations are framing AI.

WHAT CAN A SILVERSMITH TEACH US ABOUT USING AI WELL AT WORK?

A planishing hammer is a specialist tool used to create texture and form in silver.

As Andy explains in the episode, you wouldn’t make a beautiful piece of silverware, hold it up to somebody and say, “Look at my wonderful hammer.”

Yet this is exactly how much of the AI conversation currently sounds:

“We’ve got an AI agent.”

“We’ve rolled out Copilot.”

“We’re trialling the latest platform.”

Fine. But so what?

The tool is not the story. What people create with it, improve because of it, or are enabled to do differently with it; that is the story.

ARE YOU ADMIRING THE HAMMER?

There is a lot of technology theatre happening right now.

Organisations are understandably anxious about not being left behind. Senior leaders can see that AI is moving quickly – they know competitors are investing and they have been told that productivity gains are there for the taking.

So they buy the tool, announce the pilot, give people access and wait for the magic.

Then they are surprised when uptake is patchy, outputs are mediocre and people mostly use it to make an email sound a bit more polished.

That is not an AI failure. It is a capability failure.

Andy’s reflection was that he did not learn silversmithing by being handed a planishing hammer. He learned from a master craftsperson. Someone showed him the grip, the rhythm and the judgement required. Over time, he learned what good looked like.

That is the bit too many organisations are missing in their approach to AI training.

MOST PEOPLE DO NOT YET KNOW WHAT “GOOD” LOOKS LIKE

Ask a group of employees to rate their AI capability and you will probably get a fairly confident response. Ask them to show you how they use it, how they spot weak outputs, where they challenge it or what they do when it gives them something plausible but wrong — and the picture often changes.

It is the AI equivalent of asking a room of people whether they are above-average drivers. Most will think they are. Statistically, they cannot all be right.

The issue is not that people are incapable. It is that they have not been taught the craft.

They may know how to type a request into a chatbot. But can they:

Start with the real problem, rather than the first task in front of them?
Give AI enough context to produce something useful?
Ask it to challenge their thinking, not just confirm it?
Use it to generate perspectives they would not naturally see?
Recognise when the output is average, biased, incomplete or simply not good enough?

These are the future skills that will determine whether AI becomes genuinely useful or simply produces a faster flow of bland work.

PRODUCTIVITY IS THE SHALLOWEST USE CASE

Of course AI can save time. That matters. Nobody should romanticise tedious admin, endless formatting or wading through information that could be summarised in seconds.

But if our only ambition is to make people produce more, we need to be honest about where that leads.

More reports. More presentations. More ideas. More content. More work landing in already overloaded teams.

AI can turn on a fire hydrant of output. If we do not change the systems, priorities and decision-making around it, all we have done is increase the pressure.

The more interesting opportunity is to use AI to lift the heavy processing load so people can do the work only people can do well.

Thinking. Connecting. Creating. Exercising judgement. Having difficult conversations. Seeing the pattern behind the data. Asking the question that nobody else has asked.

That is where the AI dividend sits.

AI CAN BE A VIRTUAL GUILD IF WE TEACH PEOPLE HOW TO USE IT

Andy also spoke about the guilds that craftspeople once belonged to: communities where people shared techniques, challenged each other and got better through practice.

There is something in that for L&D.

AI can give people access to alternative perspectives, research support, creative prompts and a way to pressure-test their first draft before they share it with a colleague. It can help somebody who has always said, “I’m not creative” to explore ideas differently. It can help someone who gets stuck in the detail step back and see the wider picture.

But it will only do that if we move people beyond treating AI like a vending machine:

“Write this.”

“Summarise that.”

“Plan my holiday.”

Useful, yes. Transformative, no.

We need to teach people how to collaborate with AI: to say, “What am I missing?”, “Argue the other side”, “What assumptions am I making?”, “Give me the worst possible version of this idea — and now reverse it.”

That is not about clever prompting for the sake of it. It is about developing better thinkers.

THE LEADERSHIP QUESTION IS BIGGER THAN EFFICIENCY

There is also a human issue we cannot ignore. If leaders talk endlessly about efficiency but are vague about what AI means for people’s roles, people will fill the silence themselves.

Usually with fear.

AI adoption needs clarity, honesty and psychological safety. It needs leaders who can say: “We are not introducing this to make you work at an impossible speed. We are introducing it to help us rethink where human energy is best spent.”

Because the organisations that win with AI will not be the ones with the shiniest hammer.

They will be the ones that help their people become master craftspeople with it.

If you want to build AI confidence, critical thinking and practical capability across your organisation, explore our AI Academy and get in touch to discuss what your people need.