Is generative AI worth the hype?

Generative Artificial Intelligence (GenAI) is all the rage, but is it worth the excitement? Are we losing all sense of proportion, as we did two years ago with the grandiose if elusive promises of the metaverse? Or is this the beginning of a wave that will sweep through the economy, leading to new winners and losers? And, what does this mean for companies and policy makers alike?

The problem, Professor Michael G Jacobides argues, is that GenAI’s impact is both over-hyped (in the sense that it is unlikely to boost corporate profits as some consultants like McKinsey suggest), and under-appreciated (as it will radically disrupt some sectors such as advertising and some business models like advisory more than others).

The challenge is that there is little research that looks into the ways in which GenAI will reshape the competitive impact. Advisors and technology vendors are excited to make the case for new products and services spurred by GenAI, and academics cannot yet assess how GenAI is changing the competitive landscape. Yet we urgently need to have an initial map, as businesses cannot afford to sit back and wait for the noise to die down and the air to start to clear, since technology is developing fast. 

The upshot is that not only business leaders and politicians, but also academics and consultants will need to dig deeper. They should be looking at their own models and methodologies, examining their products and markets and challenging the assumptions on which those models and methodologies have been based in the past.

That is the background to a new policy paper, published by the School and the Institute of Directors (IoD), assessing the expected impact of GenAI on the U.K. competitive landscape. The study was produced by a London Business School team led by Michael and doctoral student Mingyu Dalbert Ma, in cooperation with Faisal Khan, Chair of the IoD’s Expert Advisory Group on Science, Innovation and Technology. It was funded by the U.K Research and Innovation Fund and received further support from the London Business School’s Knowledge Exchange Fund and Micheal’s own digital advisory business, Evolution Ltd., on which Mingyu is also a consultant, supported by a team of advisors including Yuri Romanenkov, Justinas Sukuys, George Alevras and others.

Armed with 277 questionnaire responses as well as engagement with 101 people in 15 workshops and numerous in-depth additional interviews, the researchers found that not all sectors expected to be disrupted in the same way or as dramatically. In the absence of hard data on the technology’s impact to date, the researchers concentrated instead on the way GenAI was used by firms (which self-assessed the impact it had on their firm – ranging from minimal to substantial), as well as the executives’ expectation of industry transformation. Questions covered both individual companies and their business models, and the sectors and ecosystems they belonged to. 

This project, which was presented in the IoD with opening remarks by the Minister of State for AI, Jonathan Berry, Viscount Camrose, the Chair of the APPG for AI, Lord Clement Jones, and the proposer of the AI regulation Agency Lord Holmes, looked both at GenAI adoption and its impact. In terms of adoption a key indicator was the extent to which company leaderships were ready to engage with the technology, which tended to vary according to the size of the business. While leadership engagement had a profound influence on how firms integrated and capitalised on GenAI, this was likely to be less of a barrier to adoption in the smallest companies than in larger businesses. As an intriguing aside, the paper suggested that larger organisations that aspired to grow quickly, “such as those owned by private equity,” had to deal with multiple competing claims on their time of energy, so that lack of leadership initiative could be a barrier to adoption.

More interesting yet were the factors that helped gauge when GenAI would have a transformative impact. The researchers found that the importance of pattern recognition to a particular sector or profession and of proprietary data to individual businesses were key to the usefulness of the technology. Concerns about the accuracy and bias of GenAI’s output, questions of regulatory compliance and security and about the availability of professional skills and expertise were among the key barriers to its adoption.

In the longer-term, however, what will count are the characteristics of the technology itself and how these characteristics will play out in its usefulness to particular industries or sectors and whether it is perceived as an asset or a threat.

Anecdotally, for example, the advertising industry is already facing significant disruption, with agencies making layoffs and their clients taking a few operatives in-house to deliver GenAI results directly. The paper duly noted that the entire industry may be “disintermediated or bypassed,” because “a significant proportion of the activity may be performed directly by clients of advertising agencies, who could simply use GenAI tools as opposed to relying on an external provider.” At the same time, however, advertising was also among those sectors, where respondents expected that GenAI might provide “the opportunity to unlock distinctiveness and as such gain a competitive edge”. The importance of pattern recognition in particular was seen as a driver of competitive success. This was also true of a range of sectors from healthcare to law to education. 

Another key enabler and differentiating factor among individual companies across many sectors will be the importance each firm places on its proprietary data. The survey revealed that respondents who regarded proprietary data as important “were 63% more likely to believe that GenAI could unlock significant levels of distinctiveness”. The bad news is that GenAI may also lead to greater inequalities between those with existing proprietary data and the resources to gather more and those without the resources to build and digitalise a database. Similarly, those with the resources to “hyper-customise” their products will outpace those less well placed to recognise and exploit client needs and behaviours.

However, there are some sectors which enjoy a considerable amount of protection though either their own professional statutes or public regulation. Michael quipped that lawyers should perhaps in theory be “obliterated” or “blown up” by GenAI, because their work is so heavily based on pattern recognition through case law and standardised contracts that it could easily be replaced by a machine. Yet the reality is they are protected by the regulatory frameworks that they themselves have designed.

As Michael said in a recent speech, “We as a society place some requests and expectations on where we get our advice. And, ironically, it is this requirement for regulation that provides the organisations that run it with the upper hand.”

With these disparities in mind, the paper also calls on policymakers to do their part in setting frameworks for the rollout and adoption of AI among various sectors, and take into account the effects of regulation both to protect and support businesses and the public interest.

As well as providing targeted support to ensure adoption of GenAI in sectors that are most at risk of widening corporate inequalities, policy makers are asked to look at ways of “democratising” proprietary data where it could lead to broader societal benefits. This could include data sharing that protects individual privacy “while allowing for collective advancements in research and development.” Other policy recommendations include encouraging firms to look beyond cost-cutting to focus on “the new monetisation and revenue opportunities that GenAI can unlock;” redesigning the education system for “a post-GenAI world” by preparing the workforce for new demands; and encouraging regulators, such as the Competition and Markets Authority, both to prevent excessive market concentration and to examine one industry’s use of GenAI on other sectors and their business models.

Finally, the paper argues that the UK lags behind the European Union, the EU and even China in establishing an overarching regulatory framework for GenAI, leaving responsibility instead to existing regulators with narrower remits, in an “approach that risks losing the forest for the trees.” It calls for the creation of an overarching regulator along the lines of – but likely going further than – a private bill under consideration in the House of Lords for the creation of an AI Authority.

The paper was published on the eve of the surprise announcement of a general election in Britain. It remains to be seen which, if any, of its recommendations will be followed through, when the new government inescapably considers how to harness, or partly tame, this significant if ill-understood agent of change. Whichever the case, the study suggests that we still have some way to go to get a better understanding of GenAI – and how to manage its impact.

Michael Jacobides is Sir Donald Gordon Professor of Entrepreneurship and Innovation, and Professor of Strategy and Entrepreneurship at London Business School