VIEWPOINT
  • Innovation

Championing AI for Business

Professor Phanish Puranam on how organizations can develop the mindsets and skills to win with Artificial Intelligence



Monday 11 March 2019

 

By downloading this resource your information will be shared with its authors. Full privacy statement.

AI is starting to transform how organizations operate and how they take decisions. These effects will intensify over the coming years and their scale will shake up the structure of most industries. The winners are likely to be those who can bridge the operational gap between the technology and its applications.

We already know two of the pools from where these winners will come: tech giants who write the algorithms and start-ups who find innovative applications. The third and possibly largest pool will be of those companies who learn fast how to deploy their data and develop AI applications.

Where organizations are with AI

AI technology – and in particular machine learning – is already real in marketing. Algorithms are predicting what customers will buy at what price point. These patterns are often too complex for us to follow, even though we can see the superior returns being made.

The wider potential of such black boxes was brought into focus in 2016. Up until then, the big AI victories the public knew about, such as Kasparov’s defeat at the hands of IBM’s Deep Blue had been the result of teams of smart coders who had built programmes that could beat a world chess champion. Alpha Go’s AI-fuelled victory at the game of Go, the ancient Chinese game of strategy, was different. It found patterns in millions of previous games, many of which it played with itself, to defeat the world champion Lee Sedol. This time, it was a triumph for bottom-up data driven intelligence.

……………………………………………………………………………………………………….

INSEAD launched its first open-enrolment program on AI for Business over three days at the beginning of March 2019 on its Fontainebleau campus.

Further programs will be rolled out in Europe and Asia. See details HERE
……………………………………………………………………………………………………….

Over the last three years, many organizations around the world have ramped up their involvement in AI, recognizing that it changes the landscape for making predictions and therefore good business decisions. The number of adopters keeps growing each month, though for now, most are still figuring out how to integrate data, business knowledge and expertise with algorithms and data science to create winning strategies. The good news is they are on that first rung of sophistication – evidence-based management.  

Competitive dynamics

In AI, we are now finding ourselves at the confluence of three powerful currents: the ease of using algorithms is exploding, thanks in large part to open source software; the hardware for processing these algorithms has increased dramatically in terms of capacity and speed; and huge amounts of data about how customers, suppliers and employees behave is becoming available.

The winners will be found among those who can develop smart managers to champion AI within their organization.

These developments are on such a scale that the consequences may only truly be understood over the next decade. Multiple functions and multiple investments are going to be touched. Wherever you have large amounts of data that could feed into key decisions, the odds are high that you will see a high level of adoption within that industry.

The winners however will be found among those who can develop smart managers to champion AI within their organization, testing out applications and making the case for investment. The skills to do that sit at the intersection of business, organization and data science. That’s what INSEAD excels at bringing together. Our faculty routinely use cutting edge data science in their research, and their teaching and consulting gives them deep exposure to management problems and solutions across a range of companies and sectors.

Our new three-day course is a first in allowing executives to work directly with algorithms (with hands on support, of course) and then interpreting the results for themselves. The emphasis is less on the details of how the algorithms work. It is more on what they can do – and what they can’t. The aim of the program is to create the skills for managers to champion AI projects within their organizations, leading teams drawn from business and data science.

Limitations, risks and biases

The field is also becoming smarter about the limits of machine learning based decisions. Consider for instance an application that is becoming popular: using data within the organization to predict who is likely to be a successful manager. Suppose the analysis suggests that men are more likely to succeed in senior managerial roles in the organization – what does this mean? Is the algorithm biased? Not really, because the algorithms produce the best estimate given the data. Are the data biased? Not necessarily, if they capture past instances of success and failure at senior roles in a representative and accurate manner. Instead, what is very likely the case is that organizational processes such as selection, promotion, mentoring and succession planning – are biased against women.

Blind use of the algorithm can lock us into a self-perpetuating cycle of discrimination. Fortunately, we now know how to spot the tell-tale signs of such problems and are developing a range of possible responses.

We must also become more sensitive to issues of data privacy and new regulation like the General Data Protection Regulation is playing an important role in structuring that.

How organizations will change with AI

Within organizations, we are used to making decisions based on experience, or intuition. That is of course data-based decision making too, filtered through the algorithms of our mind. The machine learning revolution is giving us an alternative, leaving us to choose which set of algorithms is most useful for which problems. Yes, different companies are at different levels of sophistication around data quality and integration and it may take years for some to even get their data ready to start analysing it for insights; but the basic choice created by the arrival of machine learning and big data is now a fait accompli.

AI will improve the quality of decisions and, in some cases, it will automate them altogether.

In the end, AI is shaping up to have two significant effects on how we design organizations: it will improve the quality of decisions and, in some cases, it will automate them altogether.

Many routine tasks will disappear. But managerial jobs are bundles of tasks, so this does not mean jobs will necessarily disappear – but it does imply they will be redefined, and there may well be fewer of them, of the kind we know today.

Instead, managerial jobs may focus on problems where data is sparse. It’s an area where humans excel. We make reasonably good decisions even where there is no data, in part because the standard of what is a good decision is judged by other humans, and in part because we are good at making analogies. Through analogy, we draw enough similarities from other situations to apply to the current context.

According to a recent article in INSEAD’s Future of Management series, demand for these skills in unstructured decision-making will remain high, as it will for those who work with their hands. Surprisingly, basic tactile tasks that require picking and grasping have been very hard for AI and robotics to master to date. 

Steep multi-layered hierarchies could be reaching the end of their shelf-life.

But one thing we can say with confidence – to the extent that an organization replaces humans with algorithms for decision making, this will necessarily make the organization smaller and flatter. Steep multi-layered hierarchies could be reaching the end of their shelf-life.

These are the scenarios that we are preparing for at INSEAD. We already know across a range of domains in marketing, finance, organizations and strategy that our teaching will be less about what decisions to take. It will be more about how to take decisions using your own data. It’s a complete change. We also know that this raises some interesting questions about the role of theory, whose purpose is to give us a simple framework for making the future comprehensible. If algorithms, however incomprehensible their working, can systematically make better predictions, we may well have to re-define what theory is for and what theorists do.

These are important questions for the future. For now, our focus at INSEAD is on giving managers an understanding of how they can use the power of algorithms and how they can apply them within the context of their business.

Further reading: The Future of Management: an INSEAD Knowledge series


As one of the world’s leading and largest graduate business schools, INSEAD brings together people, cultures and ideas from around the world to change lives and to transform organisations.



ARTICLES YOU MIGHT LIKE

BOOK REVIEW

Defending Human Centred Innovation

WEBINAR

Five Core Principles of Managing Innovation

View a recording of this workshop with Georgetown McDonough’s Eric Koester, to help you manage internal innovation, lead change and activate intrapreneurship inside your enterprise

VIEWPOINT

AI and Big Data: Reality Beyond the Hype

VIEWPOINT

What’s Your Problem?

VIEWPOINT

The Disciplined Entrepreneur



 
Close
Google Analytics Alternative