Judgement Is a Combination of Science and Art

You’ve had a long career across IT and consulting. What would you say is irreducibly human — something no case study or data analysis can quite capture?

Firstly, what’s unique for all of us is critical thinking: how we think and how we approach a problem. It’s all about solving a business problem and getting paid for it, right? So all of us have our own unique thinking capability. Critical thinking, identifying a problem and approaching it for resolution — that’s something very human and stays with all of us.

What you can only identify once you’ve worked long enough in an industry, is this: you know the vibe, the customer behaviour, whether something is going right or wrong despite what the data is telling you. That comes with experience. Your judgement, basically.

Judgement is a combination of science and art. Science is the data you’re able to see, from AI these days and from earlier analysis, even when it was manual. But there’s an art to it too, which is entirely your own: it could be your risk appetite, it could be your empathy, it could be your value system. All of that combined together is judgement, and it’s unique to you. It stays with you throughout the process.

You’ve also said that speed and quality need to move hand in hand when it comes to judgement. On a time-critical, accelerated project, how do you know when you have enough information to actually make the call?

We have to stop somewhere. It’s all about being efficient at every step of the process. I can’t spend my entire day looking for information and then spend another day taking the judgement call. I need to know: this is what I’m looking for from this analysis, and I need to stop here.

That’s instinct, or gut, or experience, and that’s also judgement. How much information I need to look for, that’s judgement. How I take a decision based on the information I have, that’s judgement. Going after the right problem in the first place, that’s judgement. And how I align the right resources to execute the problem-solving, that’s judgement too. Judgement runs through the entire cycle.

So, coming back to your question: how do you know when to stop looking for information? That comes with experience. I should just know: this is exactly what I was looking for, this was the business problem I was sitting with, and I have the right amount of information now. However many times I repeat the question to AI, or to anybody else, the outcome is going to be the same, with different flavours. I need to know where to stop.

I’ve always believed judgement is something innate, that no matter how much you train it, it doesn’t quite translate into the corporate system unless it comes from within. Would you agree?

Exactly. Training alone cannot give us judgement. We have to fall, fail, learn from experience, and then we know what the right judgement is versus the wrong judgement in a given context. When the context changes, my training effectively starts all over again.

So it all depends on how I’ve put my judgement to use, whether I succeeded or failed and learned from it. That becomes my knowledge base of judgement. I know that, at this point, in this particular context, I can use this judgement call and I’m going to be right.

Across the different organisations you’ve worked in, what’s one unique thing each of them taught you as you climbed that ladder?

I’ve had multiple organisational changes through my career, and each of them has taught me something unique. Right now I’m with Accenture, and what I’ve learned there is that each business problem is unique, each customer’s need is unique. You start from that business problem, but when you provide a solution, don’t expect it to be static; it has to be dynamic, changing, evolving. That’s the learning from Accenture.

When I was with IBM, they were engineering- and sustainability-solution-oriented, so I learned about reliability in the product and solution you’re building: it should meet standards, industry best practices, and so on.

And I’ve also worked in startups, which taught me that doing a giant is different from being a giant. I learned to be a giant: adapt, adopt and continuously evolve. Those are the learnings across my career.

A lot of the systems in large organisations are still legacy, not really built with AI in mind. When you’ve worked on projects where that AI readiness simply wasn’t there, what challenges did that throw up?

Even once an organisation has realised the need to go all-in on AI, many of them, small or large, simply aren’t ready yet. Readiness isn’t only about infrastructure; it’s also about being able to spend on it, which might mean optimising or cutting costs somewhere else.

Cost is readiness. Infrastructure is readiness. People are part of readiness. The building blocks might just not be there yet.

As consultants, in the kind of firms we’re with, that’s our opportunity: to help organisations build those blocks and move towards the goal, whether that’s related to data or to people. It’s about transformation.

A few years back, we were talking about cloud transformation, moving from legacy infrastructure to the cloud. Getting ready for AI usage is similar. It’s also a transformation journey.

Do you see a clear shift in the industry, companies building more AI capability in-house, or is it still mostly about leaning on third-party solutions and building on top of them?

I haven’t seen a particular trend yet. Everyone is talking about AI. From my organisation’s standpoint, I’ve had the experience of talking to large enterprises across software, platform and other industries, and some of them do have the capability and are trying it themselves.

Some are going for a mixed approach, trying some things in-house and outsourcing others. And some will simply say, hey Accenture, or any other partner, can you build this for us from scratch?

There’s no particular trend yet. It’s very new and recent; maybe a year or two from now we’ll be able to see one clearly.

You’ve spoken about innovation happening at the bottom of the pyramid. But as AI takes over more of that layer and teams get leaner and more spread out, doesn’t that create a real tension?

There used to be a concept of a triangular pyramid, and there’s also a concept of a diamond-shaped pyramid. There will always be some kind of bottom layer.

If we’re eroding the bottom-most layer through AI bots, for example, there’s still somebody governing the day-to-day observability, a team looking at how the AI is actually functioning. They know what’s working well and what isn’t, and out of that, they identify the need for innovation. That’s still a bottom layer, from a human perspective.

So there will always be a layer that’s regularly using or orchestrating the service. Even with AI, there’s a governance layer for AI: somebody watching how the service is performing and what areas need innovation and improvement.

That’s why it may not be visible at the topmost layer, but innovation is still happening there. That’s my belief.

Where does the human in the loop actually come in, when you’re deciding on a solution for an AI layer?

It depends. It varies from one solution or situation to another, depending on the risk you’re willing to take. A risk for a medical institution, a hospital, is very different from a risk when you’re dealing with a wagering organisation, and very different again from the risk at a financial institution.

Depending on that, when you’re building the solution, you have to take a call on exactly what you do not want the AI to try out and hallucinate on.

We heard an example earlier: users kept asking for a hole, and it created a hole. It’s like that. So, again, it’s judgement.

In the large projects and deliveries you work on, how has AI actually helped you move faster, in ideation and in requirement gathering specifically?

Research and prototyping have become very easy, whether that’s using internal data, external data or a combination of both. That’s research, and it’s become extremely easy.

Prototyping has become easy too. Even if I’m not a developer, I can create a prototype overnight and check with dummy data whether it’s working well. With live data, it may behave a little differently, but not poles apart.

So, from idea generation right through to execution, the nuances have become easier, and that leads to the regular question of what happens to the workforce if they were doing these things manually and you’re now compressing the execution timeline using AI.

It’s definitely our responsibility to reskill them, redeploy them and use their skills somewhere else, where human judgement and human thinking are what’s actually needed.

And how has that shifted client expectations of consulting, now that they know you can be so much more efficient with AI?

Reducing cost has been the first impact. One is cost, one is quality. The expectation is: go error-free and reduce cost. That’s really how it’s changed.

Building on your own experience, what’s one piece of advice you’d give someone starting out in consulting today, something you may have received yourself when you were starting out?

Being human is very important, whether in the AI age, before the AI age or in some future super-AI age. Being human is very important whether you’re an entrepreneur, an employee of a large or small organisation, helping customers or working with people. It’s important to think humanly.

I strongly believe in execution. Strategy is an integral and very important part of it, something you build, but outcome comes through execution. Why am I talking about execution when you asked for one piece of advice? Because it’s about working with people.

I strongly believe in participative leadership. Influence is key in today’s world. You may not have people reporting to you; it’s diverse across geography and function, but you still have to get things done, you still have to show outcomes, whether you’re using AI now or weren’t using it earlier.

The principle of execution, of working with people, remains the same. Even if the workforce size shrinks, at the end of the day, you’ll still be working with people.

For someone just starting out on a one-year MBA, grappling with the constant push and pull of everything changing in the industry, the uncertainty, a new, more advanced model seemingly every other day, what would you suggest?

Hold your stand. I mean, we don’t need to be overwhelmed by whatever is changing, because it’s going to keep changing for the next 10, 20 years. We don’t know the scale of the change that’s coming.

Somebody once said, who will live on this planet if a human is replaced by machines? It’s just not possible.

Technology will keep changing, and we’ll have to keep adapting to its pace. That’s key. If you’re able to pick up the key messages, the key learning from each technology, that’s more than enough.

We don’t always need to look for depth. In a changing world, we have to look for breadth to stay relevant. And overwhelm is something we should consciously try to avoid, so that we don’t feel helpless.

We should feel empowered that whatever the technology is, it was created by humans. We are human. We can master it, we can use it, and we can do what’s best done by a human. That’s how we should think about it.

Judgement is a combination of science and art. Science is the data you’re able to see. But there’s an art to it too, which is entirely your own

How much information I need to look for, that’s judgement. How I take a decision based on the information I have, that’s judgement. Going after the right problem in the first place, that’s judgement

Readiness isn’t only about infrastructure; it’s also about being able to spend on it, which might mean optimising or cutting costs somewhere else. Cost is readiness. Infrastructure is readiness

It’s definitely our responsibility to reskill them, redeploy them, and use their skills somewhere else, where human judgement and human thinking are what’s actually needed