The AI Amplifier: Moving from Tool Adoption to Work Design with Dr. Laura Weis, WPP

Richard Breeden: My guest today is Dr Laura Weis, who leads human strategy at WPP. Laura holds a PhD in social and organisational psychology from UCL. Before joining WPP, she led Future of Work Strategy at Sertalia, a UCL-born AI company that WPP later acquired. She sits right at the intersection of AI and people strategy, helping organisations figure out how humans and AI actually work together in practice. Welcome, Laura.

Laura Weis: Thank you so much, delighted to be here.

Richard Breeden: You’re the Global Human AI Strategy Lead at WPP. What does that actually involve day to day?

Laura Weis: My work at Sertalia was very much in creating tech assets in the future of work space. When we were acquired by WPP as the AI arm of WPP, I moved centrally into the HR department, because my stakeholders were predominantly in the HR space. I took on a role as Talent Innovation Director — a title specifically created for me — and for about a year and a half I was focused on how we needed to future-proof our talent and rethink how we view our humans in light of AI.

Laura Weis: I then really missed working with clients and moved into the CTO team, still bridging heavily to the HR space as Human AI Strategy and Transformation Lead. The observation that drove that move was that most organisations have a talent strategy or a people strategy, and most organisations have an AI strategy or a tech strategy. But a lot of the time these strategies are sitting in parallel and they’re not really aligned. Everything I do is at that intersection. You can have the best people and you can have the best AI. If you’re not connecting those in a meaningful and intentional way, you still won’t get much value. It’s all about how we bring human and AI together so that each can live up to their full potential.

Richard Breeden: You’ve talked about AI as an escalator towards average — it lifts everyone’s baseline, but if your skills and culture aren’t at that baseline, it exposes that very quickly. What does that actually look like within an organisation?

Laura Weis: A lot of the time AI is an amplifier. It shows how good you’re already working rather than fixing you. I think that’s where the problem starts — people think AI is a silver bullet, that you drop it into a system and it will make things better. But what it actually does is multiply value in good working systems. If you have an organisation that structurally isn’t clear, where there’s a lot of ambiguity, where the culture isn’t what it should be, where people are super stretched — if you put AI on top of those systems, you’ll just increase noise. They kind of get worse and worse and worse. It’s an amplifier. So a lot of the time AI exposes what works and what doesn’t work rather than fixing it.

Laura Weis: What that looks like practically, if you put AI on top of dysfunctional or outdated work structures: you get more output, but it’s less differentiated. You get polished output, but quite shallow thinking. You see a lot of what I call shadow work — a lot of reworking of content that’s quite quiet, because the first passes aren’t really strong enough. People get these wonderfully AI-polished outputs and then you look under the hood and have to rework a lot. Usually your strong performers get tied into that correction and editing, and that’s not where they’re positioned best. Confidence rises much faster than capability in organisations, and that bears a lot of risks.

Richard Breeden: You’ve said most businesses don’t have an adoption problem — they have a work design problem. How does that manifest specifically around culture?

Laura Weis: A lot of the problems we’re having with AI actually start with how we frame AI and the value of AI. AI gives us speed — and we’ve all been very excited about that for the last two years. But much more importantly, it gives us space: space to think, to challenge and to connect things that weren’t really connected before, to create new things and to innovate. The mistake that most organisations are making is that they convert a lot of that space straight into more output, more noise, more of whatever they’re already doing. As Michael Porter already said, speed is not a strategy. That’s why I don’t really think it’s an adoption problem. It’s a work design problem. We need to design for this space to be used to differentiate, to continuously reimagine and to innovate.

Laura Weis: Most organisations don’t just need more AI. They need better conditions to use AI well. You need to think about how to redesign workflows so you don’t just bolt tools onto broken processes — that’s like putting a Ferrari engine into an old car. And you’ve got to protect that space, because what I find a lot of times is that people are actually more stressed and more burnt out than they’ve ever been, because they’re just doing more and more. There’s also the cognitive load of continuously reviewing AI output and having all that optionality. If you put a brief into an AI tool, you suddenly have 20 different answers — how on earth are you going to make a decision? That compounds at team level and leads to what we call a paradox of choice. If you have all these options and AI is very confident about all of them, we get overwhelmed and don’t really know how to make decisions. Yes, AI creates capability — but work design is the factor that determines whether it goes into value or just more activity.

Richard Breeden: Is that time saving actually manifesting itself in practice?

Laura Weis: I think there is broad consensus, including from recent research, that we totally overhyped the time-saving rate. Very few organisations can prove they are really saving a significant amount of time while holding quality steady or even improving quality. That’s partly because we haven’t actually redesigned work. We’ve put AI onto old ways of working, old ways of decision-making, old ways of viewing people. That doesn’t really allow us to get the full potential of AI.

Laura Weis: At WPP, and we’re not better or worse than any other company, I think everyone has fallen to some degree to the AI hype around focusing a little too much on the efficiency side of things. In times of ambiguity, from a psychological perspective, we tend to focus on what is going to go away — where can we save costs, what roles are we not going to need anymore — rather than what is going to emerge. Over the last year, that narrative within WPP — and I think more broadly — has shifted. AI for efficiency: yes, we’ll get there somehow, and on an individual level a lot of people are feeling that efficiency. But we can’t scale it a lot of the time because the system around us — the decision-making at scale, the infrastructure — isn’t following as quickly. We’ve started very much focusing on what is going to really differentiate us, what is the role of the human there, and how do we collaborate with AI to create not just more output, but more differentiated output.

Richard Breeden: There’s a huge amount of institutional knowledge and creative thinking within WPP. How do you go about capturing that so it becomes useful for people?

Laura Weis: With Agent Hub — and I love Agent Hub — Amelia Gander, who leads it, very much set from the start a strategy of humans at the helm. We build systems that fit humans, rather than shaping our people to fit the system. She talks about the concept of human in the loop being too passive. It’s not just that humans are an approval gate. They need to be above the loop orchestrating. They need to be across the loop translating from strategy to execution. They need to be focusing on quick iteration and creation with AI. It’s a much more active role than simply being in the loop and approving things.

Laura Weis: On capturing institutional knowledge: expertise isn’t just knowledge. It’s how decisions get made in context — what to prioritise in a certain scenario, what to ignore, what to push. Capturing that for us means not only incorporating best practices and knowledge, but encoding certain decision logics. Decision-making is often the constraint to moving faster with execution, because we don’t know how to make good decisions at scale. So we’re encoding decision logic, not just information — using real examples, edge cases and trade-offs, and learning from those in the moment. We also need to make sure that when one part of the organisation is learning from experimenting with a problem, that knowledge scales rather than staying isolated. And making it specific to context: a lot of times people go to agents or AI tools that are too far removed from their actual workflows where the decisions actually happen. If it doesn’t change decision-making, it’s not useful knowledge — it’s just documentation.

Richard Breeden: What about junior roles? There’s been a lot in the news about AI’s impact on entry-level positions and how organisations future-proof themselves when juniors can produce answers fast without necessarily understanding how those answers were formed.

Laura Weis: Juniors are insanely important and I think we’ve moved past the inverted pyramid narrative of a year or two ago. Juniors are incredibly literate with these tools — they don’t need training, it comes as table stakes. They come in with the newest tools, new ways of doing things. I learn a lot from our juniors. Reverse mentorship is a real opportunity here. But at the same time, there’s a real risk: people can get to an answer super fast, but they don’t really understand how those answers were formed, when to trust them and how to really improve them, because they’ve never done the craft. They’ve never been in the arena and actually doing the work and learning from what works, what doesn’t, what fits together, what good looks like, and experiencing the consequences of when you don’t do good work.

Laura Weis: That’s where we need to design for that. We need to design an arena where they can actually do the work themselves and experience the consequences. I’m a big fan of thinking about an AI driving licence: unless you understand the craft properly and have the knowledge to know how to really interrogate things, you need to pair generating stuff with being able to interrogate stuff properly. If you can’t demonstrate that you can do that — that you know how to actually look at AI output and use AI for interrogation as well — then maybe you shouldn’t be using AI yet. We’re experimenting with that to see what works at scale.

Richard Breeden: Tell me more about the AI driving licence concept.

Laura Weis: I have to say this is probably too flat as a concept to just be a thing you pass and then you’re fine. The idea is that there’s no one way to use AI. Depending on how junior or senior you are, whether you’re a generalist or specialist, you have certain strengths and weaknesses when it comes to interacting with AI. Juniors often lack understanding of what really matters, or the consequences of making bad decisions, because a lot of times they don’t know the process. On the other hand, they can use AI to learn at a speed we were never able to before. So the idea is: for a particular group in the business, what are the risks that group will encounter when using AI, and then design some sort of mitigation process — whether it’s a driving licence or something else — that helps them continuously ensure they have the right thinking to engage with that tool. That can be tool-specific as well. If a tool skips what was previously a whole chain of thinking, there might be certain caveats to be able to use it, or stop points within the tool itself where you ask: have you actually done the cognitive work to proceed?

Richard Breeden: You talk about M-shaped talent within organisations. What does that mean and how can organisations develop it?

Laura Weis: When thinking about the future of talent, you first have to think about how work is going to change. The shift will be from relatively hierarchical, quite rigid, often siloed, waterfall ways of working, towards something much more agile — an organisational ecosystem and network, much less focused on traditional hierarchies and much more merit-based, because organisations including WPP are moving more and more to outcome-based models rather than time-based models. Smaller pod-like structures where people work in smaller groups, surrounded by or working together with this army of agents and AI tools, getting the work done like that.

Laura Weis: From a talent perspective, the characteristic that really stood out in our research is that people need to be absolutely amazing at integrating — connecting the unconnected. Connecting the unconnected across processes, across actors within the organisation, between people and people, between people and AI, between AI and AI, as agent-to-agent work grows. And from an innovation perspective: truly novel ideas come from connecting concepts that have never been connected before, and you do that best with diverse teams.

Laura Weis: M-shaped is a progression of the T-shaped model. You have your deep specialisation — the craft — which is so important, because without it you don’t know what really matters. You also have that broader knowledge that allows you to connect the unconnected and understand end-to-end business flow. And then you have one additional area of growing expertise that helps connect the unconnected across processes, actors or from an innovation perspective. Within WPP we’re not saying everyone now has to have three areas of deep expertise — that’s not doable. But we are building that generalist axis in a very deliberate way, with certain characteristics we believe are super important for working successfully in an AI-enabled world. Plus, everyone needs to find their additional ‘M’. If you’re a media planner, you need to find that other area — which in media planning is often in the more strategy space — to help connect the unconnected while still building deep craft in your core specialism.

Richard Breeden: You’ve mentioned critical thinking several times. Is that the primary skill, and what does it actually mean to operationalise it?

Laura Weis: We did a huge piece of research at WPP and the first question we asked was: what is going to be the most important skill in a more AI-driven world? Something like 80 per cent of people said critical thinking. So the next step was to look at: do we hire based on critical thinking? Do we promote based on critical thinking? Are our key leaders our best people at critical thinking? And I have to say: while it was woven into some of those mechanisms, it wasn’t made as explicit as it should be, given that it’s supposed to be the most important skill in this world. Also, when we asked what people meant by critical thinking, we got wildly different responses. You can’t really expect people to bring critical thinking to the table if it’s not explicitly rewarded, and if we’re not very clear on what we actually mean by it.

Laura Weis: Do we have a culture that rewards leaders who create a culture where critical thinking is not only accepted but actively encouraged? And also, if everyone starts thinking critically, how do we then make decisions? You can’t just take everything apart without then moving on. So we’re starting to really think through this: yes, critical thinking is wonderful, but how do we actually operationalise it? How do we ensure our talent mechanisms reflect it? How does it translate into how humans and AI interact? How do I actually critically evaluate AI output? Everyone has some ideas around this, but really building standards and protocols around it is something we’re starting to work on seriously.

Richard Breeden: You’ve highlighted research about disclosure — if you tell clients you’re using AI, they trust you a bit less; if they find out you didn’t tell them, they trust you even less. How should businesses manage that?

Laura Weis: I think it isn’t really a disclosure problem. It’s a trust design problem. If you make it about whether AI was used or not, you kind of lose either way. I think the trust we ultimately want from our clients comes from confidence in how the work was produced and assured. We need to lead with standards, not tools. We need to make the human layer visible. We’re very quick to say: we produced this wonderful thing using these wonderful AI tools. But what our clients really want to know is: what was the role of the human there? Was the human just in the loop or did the human orchestrate? How did we make sure there was judgement? Where does accountability lie within the system? I don’t think clients need reassurance that AI was or wasn’t used. I think they need reassurance that the work is sound. We need to lead with standards and outcomes, and then let the tool use be secondary to the value and the quality layer. Tie trust to outcomes, not inputs.

Richard Breeden: Final question: you’re in the shoes of a CEO of a mid-market business who’s just hired someone to lead AI across the organisation. What should the priorities be for the first six months?

Laura Weis: First of all, appreciating that AI is no silver bullet, and really figuring out what are the conditions you need to put in place for AI to be successful in the first place. Then figuring out what AI success really means and where it actually creates value within the business. Start by mapping where work currently breaks across key workflows, because if you just put AI on top of that, it’s not going to fix it — it’s going to make it worse. Identify high-friction, high-value points.

Laura Weis: Redesign a small number of workflows end to end, thinking about how you properly embed AI into those workflows. Then automate for relief and not for applause — because I see a lot that people are trying to automate the things that are super glamorous, the things that look impressive. Why are you doing that? Why are you not doing all the grunt work stuff first? Automate for relief. That creates capacity. Capacity creates better decisions. Better decisions create real value. Define quality standards — that quality layer, what really matters, what the crucial decision points are, and how you use AI not only to automate and to be faster, but to make better decisions at those points. And lastly, scale what works well, not what demos well. Scale from proof of value, not proof of concept.

Why do most organisations have a work design problem rather than an AI adoption problem?

Dr Laura Weis argues that the mistake most organisations make is converting the space created by AI directly into more output — more of whatever they are already doing. AI gives speed, but the more important gift is space: space to think, challenge, connect ideas and innovate. When that space is immediately filled with additional activity, organisations end up with more output that is less differentiated, polished output with shallow thinking underneath, and people who are more burnt out than ever. The fix is not a new tool. It is redesigning the underlying work so that AI capacity goes into differentiation and better decisions, not just higher volume.

What does it mean to say AI is an amplifier, not a fix?

Dr Laura Weis uses the amplifier framing to describe how AI multiplies what already exists in an organisation. If the working system is clear, the culture is healthy, and the processes are sound, AI makes it better. If the organisation is structurally ambiguous, the culture is poor, or people are already overstretched, AI makes that worse too — faster. You get more output, but it is less differentiated. You get polished outputs with shallow thinking underneath. Strong performers get pulled into correcting and reworking AI-assisted drafts rather than doing the work only they can do. Confidence rises faster than capability, and that creates significant risk.

What is the paradox of choice in AI-enabled work and how does it cause burnout?

When AI tools generate 20 possible answers to a single brief or question, teams face a decision-making problem rather than an output problem. The cognitive load of evaluating all those options, making a judgment between them, and then reviewing and reworking the result is enormous — and it compounds at team level. Dr Laura Weis calls this the paradox of choice: the more options AI provides, the harder it becomes to make decisions, and the less satisfied people tend to be with the decisions they do make. This is one of the reasons why AI adoption is creating more stress and burnout rather than less in many organisations, and why work design — including clear decision-making protocols — is essential.

What is an AI driving licence and how does it work in practice?

The AI driving licence is Dr Laura Weis’s concept for ensuring people have the craft knowledge and interrogation skills needed before they are given access to high-impact AI tools. The principle is that generating AI output and being able to critically evaluate that output are two different skills — and that using AI without the second skill leads to fast but shallow and unreliable work. In practice, this means designing tool-specific checkpoints where users must demonstrate they have done the underlying cognitive work before proceeding, rather than simply accepting AI output at face value. The concept is still being developed and tested within WPP, but the principle is: pair generation with interrogation, and ensure that access to powerful tools is matched by the ability to judge what those tools produce.

What is M-shaped talent and why does it matter for AI-enabled organisations?

M-shaped talent is Dr Laura Weis’s progression of the T-shaped model. T-shaped talent combines deep specialisation in one area with broader general knowledge. M-shaped adds a second area of growing expertise specifically chosen to help connect the unconnected — across processes, people, disciplines or from an innovation perspective. The rationale is that as organisations move towards smaller, more agile pod-like structures working alongside AI agents, the ability to integrate — to connect what has not been connected before — becomes the differentiating human skill. Within WPP, this means deliberately building the generalist axis of each person’s development while protecting and deepening their core craft, and helping everyone find their second area of expertise that helps bridge across silos.

How should organisations think about trust and AI disclosure with clients?

Dr Laura Weis reframes the disclosure problem as a trust design problem. If the question is whether AI was or was not used, organisations lose either way — disclosure reduces trust slightly, while concealment that is later discovered reduces it significantly more. Her argument is that clients do not actually need reassurance about which tools were used. They need reassurance that the work is sound, that a human was actively involved in orchestrating and judging the output, and that accountability sits clearly somewhere. The practical implication is: lead with standards and outcomes, not tool transparency. Make the human role visible — the judgement, the orchestration, the quality layer — and let the tool use be secondary to the value delivered.

What should a mid-market business do in the first six months of an AI leadership hire?

Dr Laura Weis’s priorities are: first, establish that AI is not a silver bullet and define what AI success actually means for the business. Then map where work currently breaks across key workflows, because AI applied to broken processes amplifies the breakage. Identify high-friction, high-value points and redesign a small number of workflows end to end with AI properly embedded. Automate for relief rather than for applause — the most glamorous automations are rarely the highest value ones; the grunt work is what frees teams for judgment. Define quality standards and the crucial decision points where AI should improve decision-making, not just speed. And scale what works well, not what demos well. Scale from proof of value, not proof of concept.