How disruptive will AI really be?

How disruptive will AI really be?

There’s one thing almost every expert can agree on, and that is the extraordinary speed at which Generative (Gen) AI has emerged and evolved. How much impact it’s really having, versus how much hype, is harder to pin down. This leaves us with economic uncertainty, questions about where the real value lies, and a scramble among business leaders to appear ahead of the curve. Privately, many admit they’ve yet to see significant gains.

It is against this backdrop that Michael Jacobides, Sir Donald Gordon Professor of Entrepreneurship and Innovation, spearheaded London Business School’s first AI Masterclass, in partnership with the Financial Times. Across two sessions, he unpacked how disruptive AI is likely to be, what it will take to harness its power, and how organisations should respond.

The hype

“GenAI is strange,” Michael begins, “because it was developed in a different way from almost every other technology in history.” Rather than being built to solve a defined problem, he argues, GenAI was developed to see what might be possible. This has led to a race for use cases, as well as a surge in those trying to get rich riding the technology wave.

Meanwhile, much of the money behind AI still flows from the capital markets, rather than returns from the tech itself, with some of the biggest players buoyed by what Michael describes as circular valuations. For now, he argues, the people making the money are “those selling the shovels and the spades,” invoking the gold rush, where it was often those supplying the tools – rather than those digging for gold – who profited most. In today’s case, that means companies like Nvidia, selling the chips that power the ecosystem, and cloud providers like Amazon Web Services and Microsoft Azure.

“We don’t even have comprehensive data on the benefits,” Michael says. “People can’t really explain the upsides. Hype can mobilise action, but it can also blur judgment.”

There is a generational challenge, too. Many of the people sitting on boards are, statistically speaking, among those least likely to have engaged directly with the technology. “They’ve reached a stage where they don’t need to get their hands dirty,” Michael observes. “Their agenda is full, and many are balancing something without really knowing what they’re talking about.”

And yet the technology is real – and it presents real opportunities. So how do we harness it?

The system

If there’s one central message, it is this: AI’s real impact will not come from adoption alone, but from organisational transformation.

Michael draws a parallel with the shift from steam to electricity. Early adopters saw limited improvements because they simply replaced one power source with another. But the real advantage of the electric engines was how much smaller they were – so instead of having one huge engine, manufacturers could have multiple electric engines to serve the localised needs of each part of the production process.

This required an entire systemic rethink. “The problem is, we don’t want to rethink the entire process,” Michael notes, “we just take one thing and plug it in.” And this is exactly where many organisations find themselves today. GenAI is layered onto existing workflows, improving efficiency at the margins but rarely changing outcomes in a meaningful way.

There are early signs of deeper impact – most notably in software development, where coding productivity has increased significantly. Elsewhere, gains are slower to materialise. That lag is not surprising: new technologies only prove valuable when we reconsider what we do.

The divide

AI will not affect all work equally, Michael explains. Tasks that are modular, repeatable and pattern-based are the most suitable for automation – for the time being at least. This includes areas such as basic legal drafting, reporting and parts of marketing – domains where GenAI can already perform effectively.

By contrast, tasks that are relational, contextual or heavily regulated are less likely to get replaced. Work in which clients do not just pay for outputs, they pay for judgement, accountability and trust – understanding client needs, and navigating complex regulation, for instance – is harder for AI to replicate.

But there’s still a lot of uncertainty – which explains why perceptions of disruption vary so widely – even within the same profession. Research at London Business School found that some professionals expect to be heavily disrupted, while others feel relatively secure.

The context

GenAI’s outputs depend entirely on the data and context it is given. This makes proprietary data, customer relationships and domain expertise crucial if you want to get ahead.

Michael cites how companies like L’Oréal are leveraging rich datasets to deliver highly personalised services. Others are combining AI with regulatory or operational expertise to create new products and services that are hard to replicate. To make the most of the technology, you need to combine it with unique context.

The payoff

Many organisations are currently focused on small, visible use cases – automating emails, generating slides or experimenting with chatbots. These “white rabbits”, as Michael calls them, help to get people more comfortable with the technology, but they rarely make a meaningful impact.

To deliver real change, you need to find what Michael terms the “big elephants” – initiatives tied directly to what matters most – growing revenue, reducing operating costs and improving customer value.

This means prioritising. Rather than pursuing lots of disconnected experiments, organisations should focus on a small number of high-impact transformations – while still encouraging experimentation at all levels to surface new ideas.

The challenge here is not technical, but managerial: getting leaders on the same page, redesigning workflows and committing to change.

The people

If technology is advancing quickly, organisational change – particularly when it comes to people – is lagging behind.

One thing that’s often missing in the workforce is the ability to critically evaluate the often highly polished outputs that AI generates. GenAI is designed primarily to be persuasive, and not necessarily accurate. This creates real risks if people don’t know how to spot errors.

There’s also a structural challenge – if AI increasingly reduces the need for junior roles, how will organisations develop future talent? So far, there are no clear solutions, only emerging experiments.

The value

Ultimately, GenAI is not just changing how work is done – it’s reshaping what organisations are for.

Michael offers another analogy: doormen were not replaced by automatic doors, because their value extended far beyond opening doors. They provided reassurance, service and status.

So as AI takes over certain tasks, organisations must reconsider their value proposition – rather than just focusing on what they’re already doing.

This shift is underway in education. Business schools are rethinking what they teach, how they assess and their broader role in a world where knowledge is widely accessible.

The remaining questions

For leaders, the challenge is not to chase every new development, but to focus on the fundamentals:

• Where does AI create real value – not just efficiency, but differentiation?

• What role should we play in the AI ecosystem?

• What unique data or capabilities can we leverage?

• How do we redesign workflows to unlock impact?

• And how does our value proposition need to evolve?

The organisations that succeed will not be those quickest off the mark to adopt AI, but the ones that rethink themselves most deeply. As Michael concludes: “keep your eyes on the stars – and your feet firmly on the ground."