Thoughts on leadership

Four Years In, and the Hard Part Still Isn't the AI

One piece of feedback showed me the gap between what I intend and what my team feel. Four years after ChatGPT, the hard part of AI adoption is still people.

4 min readLeadership

Andrew Yardley, engineering leader, hands-on with AI.CareerMore posts

On 30 November, ChatGPT turns four. Looking back, those four years span three jobs and three teams, each at its own stage of AI adoption. In that time we have gone from chatbots to agent orchestration. Some days I feel ahead of it. Other days I feel a long way behind.

This isn't really a post about the technology, though. It's about something a piece of feedback taught me, and it's about people.

The feedback I didn't want to hear

One piece of feedback I had recently was that people feel pushed towards AI, but don't know how to move forward from where they are.

My first reaction was to get my back up. I am encouraging people, I thought. I suggest things to try, I listen, I point people at training. I have been explicit that this isn't about judgement. I want people to be curious, to explore, and to show the team what they find.

But that is exactly the point. What I intend and what people feel are not the same thing. Pressure can be felt even when it isn't meant. You can tell people there's no pressure, but if they can't see their next step, all they feel is the pressure.

Everyone is at a different stage

After a lot of conversations with a lot of different people, one thing I am sure of is that everyone is at a different stage of their AI journey.

Some people are afraid of where it is going, and of losing their jobs to it. That fear leads, understandably, to some resistance. Some feel they have learned as much as they need and are comfortable where they are, and part of that is protecting their autonomy. Many feel judged on whether they have something to show within weeks.

It is worth saying plainly that this technology was thrust on people. The job many of us signed up for isn't quite the job we have now, and that is true well beyond software engineering.

It is also easy to overestimate where you are. I do it myself. The days I feel furthest behind are usually the days I have seen what someone else is doing. Most of us, me included, are less far along than we think.

The uncomfortable part

I should be honest about my own part in this. I am not only encouraging curiosity. I am also leading changes that will naturally make people nervous, such as automating routine operational work so that people can move onto higher value, product focused work. Saying "just be curious" while automating work is a mixed message, and I understand why it can land that way.

The analogy I keep coming back to is the car assembly line. Automation did change those jobs, and the teams got smaller. But people didn't disappear from the line. Someone has to run the machines, maintain them, repair them and improve them. A team of engineers looks after the whole line together rather than one person per robot. They swarm across it.

I think engineering is heading the same way. People will still be needed for direction, for the domain knowledge that doesn't live in the files and the codebase, for creativity, and for the collaboration that makes any of it work.

Where I stand

I feel the anxiety too, especially as a parent. I have been through the doom spiral that social media is very good at feeding, where every story is the worst case. It isn't a healthy place to stay.

So I have made a choice for myself. If I became resistant and ground down by it, I don't think I could support my family. So I lean in, and I try to stay optimistic, partly because it protects my own headspace. That is my reasoning, not advice for anyone else.

What I won't do is write people off because they are at a different stage. Some of them may be among the best engineers I have, and what they need is a bit of coaching rather than a verdict.

What I am changing

My approach so far has been to invest in the people who are naturally curious and let them show everyone else what they have built. It creates pull rather than push: the early adopters bring everyone else along. Alongside that, I have used one to ones with the less keen to give context about why we are doing this.

The feedback showed me that context isn't enough. People told me they didn't know how, not why. So a few things change.

First, I am treating feedback as being about the whole team, not about working out who said it. It is a polite way of saying that something applies to everyone.

Second, in one to ones I will ask about AI deliberately and specifically, not as a weekly formality. "Are you OK?" tends to get a polite answer, so I will try questions like these:

  • What would you try next if you had a free afternoon?
  • What is the one thing stopping you from going further?
  • Do you have enough direction to know what good looks like for you?

Third, I will use the answers to give each person a concrete next step. That might be training, pairing with someone further along, or simply a clearer picture of where we are heading and the part they play in it.

Four years in

Four years after ChatGPT arrived, I spend less time thinking about which tool to use and more time thinking about how the people I lead are feeling. The technology keeps moving. The hard part is still helping each person find their next step.

Read next

Leadership

From Newbie to Rockstar: Mentoring New Engineers

Day one in a new job is a mix of excitement and nerves. How I help new engineers settle in, build confidence and grow, through onboarding and mentoring rather than a rigid playbook.