Artificial Intelligence (AI) is poised to transform the business world. It is not the first time that technologies have done so. Decision support systems, expert systems (already integrating AI) and knowledge management systems, which were promoted by the largest consulting firms and IT manufacturers, monopolised the interest of companies in the late 1990s. Yet, despite the undeniable benefits they offered, the interest in these technologies waned.

Even if history does not repeat itself, it holds valuable lessons relevant to the arrival of AI in business. In particular, the role humans will take in the success or failure of these technologies and how companies should involve everyone across the organisation in rethinking their roles.

Amazon Go and mass retailing

In January 2018, Amazon Go, the very first AI-powered supermarket, opened to the Seattle public on the ground floor of Amazon’s headquarters. Customers are identified as they enter the store by scanning their Amazon Go smartphone application codes, but there are no checkouts or security guards. Hundreds of devices installed on the ceilings and display shelves recognise customers when they take a product from a shelf and then add it to their list of purchases. Once shoppers leave the store they are billed for all items on their list.

This technology is not yet perfect. For example, no products by weight are sold in these stores and initially, only Amazon employees can shop there. By providing feedback about the initial flaws, employees have spent months helping to iron out problems with the AI devices in this store.

Although there are no checkouts, new jobs have emerged. For example, employees examine camera images to ensure that customers are identified even if their face is partially obscured. This role is not about “monitoring” the system but rather “training” it by telling it whether a face is being hidden by a scarf or a hat pulled down low, for example. Over time, this performance of this system, equipped with a “deep learning” architecture, will improve.

Amazon Go employees also act as “sustainers” to prevent the AI systems from acting inappropriately. For example, if you are tall and help someone by reaching for a product on a high shelf, your generosity could lead to the product being placed on your list of purchases. The role of the “sustainers” behind the screens is to prevent precisely this sort of mistake.

Reimagining work

These are just two of the new roles in this store which, according to Amazon, has as many employees as a traditional store. Many other occupations are emerging due to the adoption of AI by companies, including “explainers” who are responsible for clarifying the actions of AI systems to decision-makers, e.g. proposals to refuse credit to customers.

Trainers, sustainers and explainers… In these new roles, it is people who are helping AI. Conversely, AI can help people carry out their work by augmenting their cognitive abilities (such as WAZE which detects, analyses and decides on the route to be taken at all times), by interacting on their behalf (e.g. “chatbots” which answer customers’ questions on websites) or by physically helping them (e.g. robots working alongside mechanics in garages).

All this has an impact on the work carried out in companies. If it is a question of new jobs in support of AI systems, employees such as checkout staff must accept them and be trained to perform them. Companies must also allow unwilling staff to move into more traditional jobs. If it is a question of achieving capacity gains — therefore time savings — through AI, there must be a rethink of the additional functions staff might wish to take on.

Successfully reconsidering ways of working

All this has an impact on the work carried out in companies. According to Oxford University, 45 per cent of today’s jobs will no longer exist in 25 years’ time. They will be replaced by others, linked to AI, but not at the same pace, and they will not require the same competencies: the old jobs, performed by poorly trained staff, will disappear at a faster rate than the new jobs, performed by highly trained staff, will be created. To avoid major social and personal tragedies, companies should immediately start reconsidering the ways in which the people whose jobs will disappear will work in the future.

Indeed, one of the lessons learned from the failure of the knowledge management systems mentioned earlier is that this reconsideration will not happen on its own. Back in 2001, managers and consultants were deploring the fact that “the first knowledge management projects were not taken very seriously, with people thinking that attitudes could be changed simply by introducing a sophisticated tool”. In fact, 65 per cent of knowledge management systems were implemented by “experts” rather than jointly developed by senior management and employees with the assistance of experts.

Fortunately, there were exceptions. At STMicroelectronics, for example, “rather than rolling out a cumbersome and restrictive system, the firm opted to create a small advisory unit to assist with the development of spontaneous experiments”. At Cofinoga, a customer credit company, a senior manager asserted that “I am not at all in favour of a top-down approach; we started out in the field by seeking to simplify employees’ daily activities”, and a leading consultant summed up the situation by stating that “it’s all about voluntary involvement and motivation: you cannot impose sharing”.

Involve employees to ensure AI is a success

I am enthusiastic about AI, having experimented with it as an engineering student and then as a researcher in cognitive sciences. But that is not sufficient. If a company proceeds in a traditional top-down manner, by “outsourcing” technologically oriented projects to experts and by failing to involve the people whose activities will be predominantly affected by them, it is destined to fail.

A company is much more than the sum of its technologies, tools and resources. It is first and foremost the people who work for it. They must not come to work reluctantly, driven solely by the thought of making it through to the end of the day — 89 per cent of British employees are disengaged from their companies, according to a Gallup survey in 2017 — but driven by the desire to accomplish a vision that they share  and which inspires them.

In other words, if business leaders do not transform their organisational environments to secure their employees’ engagement, which includes involving employees to rethink their jobs in light of AI, then this technology will fail. Rather than enchanting businesses and their customers, AI will disenchant them or worse: companies and the jobs of their employees who fail to make this transformation will disappear, to the benefit of those – Google, Apple, Facebook and Amazon (GAFA) – who are driving it. And the latter will not be to blame.



  • This blog post gives the views of its author(s), not the position of LSE Business Review or the London School of Economics.
  • Featured image by Sikander Iqbal, under a CC-BY-SA-4.0 licence
  • Before commenting, please read our Comment Policy

Isaac Getz is professor of leadership and innovation at ESCP Europe, Paris/London, author of Freedom, Inc. and Leadership without ego. His forthcoming book is The altruistic corporation.