ai, the next digital frontier ?
Artificial Intelligence is becoming increasingly powerful in the business world, as a McKinsey report entitled Artificial Intelligence - The next digital frontier?, which was unveiled during the 2017 Viva Technology fair in Paris, clearly shows. There is no longer any business field or company department that would not be able to benefit from AI-based advances. But the question of just how a company ought to handle this new industrial revolution has not yet found a definitive answer.
So far, company department management have usually taken the lead on introducing the use of AI algorithms. Marketing departments are a good example. They started very early on to use machine learning to feed into the product recommendation engines on their websites and optimise their advertising message content. Other business fields subsequently became interested in AI technology. For example, banks and insurance companies began deploying AI technology to detect fraud, and logistics firms started to use predictive algorithms to optimise stock rotation. However, AI technology, which is proving able meet the specific needs of so many areas of business, has usually been rolled out by startups that have developed Cloud services, and in such cases the expertise remains with the company’s startup partner. In 2016, there were over 1000 startups worldwide working in this new area of AI applications.
The advantage of this approach for companies is that they can benefit very quickly from AI-driven gains without having to invest heavily in building their own technology platform and teaching employees a new set of skills. In fact, the job of the Chief Information Officer (CIO) often consists of little more than channelling company data to feed the algorithms that make their calculations in the Cloud. The disadvantage here is that the service is used in ‘black box’, silo mode, which prevents synergies developing between different areas.
Who should manage company AI strategy?
How long before appointed VP for AI?
Now that Artificial Intelligence is seen as a key competitive factor for companies in the future, they will need to go beyond tactical use of the technology and work out an overall AI strategy. Some firms have incorporated AI aspects into their Big Data strategy and we are starting to see high-flown titles for people in charge of Big Data and Artificial Intelligence. The most common current practice is still to give AI to the Chief Digital Officer (CDO), i.e. the person who has already been entrusted with the responsibility of managing the firm’s digital transformation. However, some companies have gone further, adding the post of Vice President, AI to their organisation chart. This happens very often in companies where AI is part of the product and service range. One such case is Capgemini, which has a Vice President for Big Data and Artificial Intelligence. Meanwhile Visteon, a supplier of electronic products to vehicle manufacturers has created the post of VP Artificial Intelligence to address the challenges of the autonomous car. The job has in fact been given to a serial entrepreneur, Vijay Nadkarni, who is now based at the technology centre Visteon has built in Silicon Valley. His job is to design autonomous driving systems which Visteon will then sell to automobile manufacturers.
It’s better to start from problems you need to solve and then move towards AI to do so, rather than appoint a senior person who will comb the organisation looking for places to apply AI technology.
Despite these rare examples, a debate has now begun on whether it is necessary to create the role of a VP for AI or a Chief AI Officer. In an article entitled Please Don’t Hire a Chief Artificial Intelligence Officer, published in the Harvard Business Review, Kristian J. Hammond expresses his conviction that companies should definitely not appoint a specific Vice President to run their AI strategy. The Scientific Director at artificial intelligence technology company Narrative Science, who also has tenure as a professor at Northwestern University in the United States, argues that it is better to start from problems you need to solve and move towards AI to do so, rather than appoint a senior person who will “comb the organisation looking for places to apply AI technology.” He believes that a bottom-up, rather than top-down approach is the right way to go.
Orange Business Services
We should not repeat the thinking behind the idea of the CDO (...). The weakness of this approach is that you give a title to someone and then automatically suppose that s/he has the right skills in this field.
"The main issue is digital transformation and, above all, the company culture," underlines Robin Ferrière, Digital Intrapreneur at Orange Business Services. Responsible for working on optimising customer relations and conversational commerce (i.e. chatbots), this intrapreneur strongly encourages the decentralised use of AI throughout Orange Business Services. He points out: “Each business line needs to carry through this transformation according to its own practices, the issues it encounters in its own business. We should not repeat the thinking behind the idea of the Chief Digital Officer (...). The weakness of this approach is that you give a title to someone and then automatically suppose that s/he has the right skills in this field. The appointee will have neither the credibility nor the right hierarchical relationships vis-à-vis the individual Business Owners, who have to manage their own P&L and run their own budgets." Robin Ferrière is more in favour of a pragmatic approach where each management team runs its own AI projects independently of the central authority. "French companies are currently trying out this approach over periods lasting 18 to 24 months. Only during the 2nd phase will they be able to switch over to a rationalised approach. At that stage, the CIO will be able to act as coordinator between the different management teams and ensure that the firm’s AI resources are shared across a number of platforms,” he predicts.
CIO looking to take back control
It has quite often been the case that CIOs have been kept at arm’s length from digital transformation projects, especially by the Marketing department, which has a bigger budget that it does not have to spend on managing legacy IT systems. Cloud Computing has helped to drive this approach as it is no longer necessary to have IT infrastructure in place in order to make use of applications based on artificial intelligence.
While Information Systems management may wish to take responsibility for digital transformation projects into their own hands, they don’t necessarily have full credibility.
While Information Systems management may wish to take responsibility for digital transformation projects into their own hands, they don’t necessarily have full credibility. "Over the next five years AI could grow incrementally at companies or it could turn out to be completely disruptive,” concludes Frédéric Charles, head of digital strategy at Suez Smart Solutions. His argument is: "If it’s growing incrementally, and the firm’s IT managers have already taken on the role of incorporating digital technology and have helped to create an online relationship platform with the company’s stakeholders, and have increased the amount of data collected via these relationships or via connected objects, it follows logically that the IT management can help the firm’s various departments to get the best out of the data. But if, on the other hand, IT management hasn’t been involved in these various stages of digitising relationships and sourcing data from them, then these people will certainly be able to add very little value to AI, which is after all an additional step in the firm’s transformation. So it would be better to give the job to the managers who have already been involved in the company’s digital transformation."
AI is a vast topic, and that’s why it should be Innovation management that introduces AI to the company’s various business departments. They can then identify the areas most likely to benefit from the technology
Frédéric Charles believes that Innovation management should be involved when new paradigms made possible by AI technologies are being tested upstream. "Testing could be part of a project run by an internal Lab or it could be done in an ecosystem of industry players,” he suggests, revealing: “At Suez Smart Solutions, Innovation management has undertaken a study in conjunction with the group’s network of data scientists in order to identify the opportunities, explore the tools, and test out a number of scenarios." Patrick Constant, one of the founders of French search engine Qwant, who is an expert in AI and automated language processing, also feels that Innovation management should play a key role in a company’s adoption of Artificial Intelligence solutions. He points out that "AI is a vast topic, and that’s why it should be Innovation management that introduces AI to the company’s various business departments. They can then identify the areas most likely to benefit from the technology (and there are many in every domain). Only once people have discovered these applications and after several Proofs of Concept have been implemented should a dedicated AI team be recruited, reporting to Information Systems management and ultimately to a VP for AI." French companies such as EDF and Airbus have set up AI departments as areas of expertise which other departments can call on to help run their projects. Our experts believe, however, that this is only a transitional step.
What if HR management took charge of AI?
HR management ought to be rapidly fused with Information Systems management. By 2020, separating the management of silicon brains from that of human brains will be seen as a highly out-dated approach…
In the keynote speech he gave (in French) on 30 August to the ‘summer university’ run by French employer federation Medef, entitled Artificial intelligence: how is it going to shake up the economy?, Laurent Alexandre, President of molecular biology company DNAVision and of NBIC Finance, underlined the fact that Artificial Intelligence is becoming increasingly important in society as a whole. He urged French and European companies to incorporate AI into their strategies, or risk rapid demise. Regarding the way companies are structured, Laurent Alexandre gave some very clear advice to the business leaders in attendance: “Human Resources management ought to be rapidly fused with Information Systems management. By 2020, separating the management of silicon brains from that of human brains will be seen as a highly out-dated approach.”
The idea of merging HR and IT systems management is nothing new. It is the direct result of a socio-technical approach that many corporations began to adopt back in the 1950s. Many HR directors in France support this approach – which focuses on ensuring harmonious interactions between the workforce and the technology in use – as does AI expert Philippe Gautier. Today he is head of a startup called Glookal, but a decade ago he tried unsuccessfully to offer AI solutions to Information Systems managers in France. He tells us: "If we place ourselves in the historical, epistemological context, the socio-technical school of the 1950s advocated an approach that brought together the social and technical aspects of organisations, the basic idea being that if technology was viewed as a peripheral aspect or even managed as a parallel aspect to the social organisation of the company there was a serious risk of aberrations and conflict in the way companies functioned,” arguing: “The last thirty years have not been very favourable as regards bringing in AI techniques, but now companies are really having to think about how they incorporate the technology. A company needs to be run as a whole entity and there’s no longer any reason to hive off people who work on the human resources side of an organisation from those who work with technology.” Gautier points out that companies have nevertheless largely rejected this way of thinking and have set up information systems that work independently from the human resources side of the organisation. The time has now come for the machines to be brought alongside the human resources within a single ‘socio-technical’ organisation, he argues.
This post of Socio-Technical Resources Director – created by merging the HR and Information Systems departments – will be responsible for processes and resources, whether we’re talking about AI or people
At the present time, there are very few companies that are prepared to merge their Human Resources and Information Systems departments. “Nowadays it’s usually the Chief Digital Officer who takes care of AI, much to the chagrin of the Information Systems department", says Gautier, adding: “lS departments are first and foremost producers, they have to manage cost centres and they don’t have a lot of persuasive power; others in the company don’t really listen to them.” However, he believes, the CDO being in control of AI strategy is only a transitional stage, a step towards the overall management of human and technical resources which is a direct descendant of the ‘socio-technical’ school of the 1950s. Gautier predicts that “this super-post of Socio-Technical Resources Director – created by merging the Human Resources and Information Systems departments – will be responsible for processes and resources, whether we’re talking about machines, artificial intelligence or people. This high-level management role should report directly to General Management as it’s a strategic role for the firm and will be a determining factor in its future.”
While, from a technical point of view, first-level Artificial Intelligence – basically machine learning and deep learning algorithms which extract value from large quantities of data – is starting to gain ground inside companies, the advent of ‘strong AI’, as defined by Laurent Alexandre during his talk at Medef, may lead to a complete reshuffling of the cards. Fourth-generation AI systems, which are likely to possess self-awareness and an IQ greatly superior to that of human beings, could even take over the job of CEO! Alexandre predicts that AI systems capable of developing and running their own projects will appear somewhere between 2035 and 2050, if human beings do not put a stop to the process before then.