You don’t even have to know how to program to play an active role in Artificial Intelligence in the enterprise. It’s time for the “business” to acquire the necessary skills and work on AI issues on an equal footing with data science teams and IT, writes our author. “In God we trust, all others must bring data.” Data has become the most valuable resource of the digital age. However, when you ask leaders a little more specifically how exactly they plan to benefit from data, it often goes quiet or is followed by vague statements such as, “We recently hired a data scientist” or “First we need to get our new data lake up and running” – these are early warning signs that leadership lacks understanding of artificial intelligence (AI) and may lack a clear vision for an AI initiative[1]. Why do the detailed questions about concrete use cases of AI in their area of responsibility embarrass many leaders? The answer is obvious: seen from the outside, the barrier to entry into the topic appears to be very high, especially for managers and subject matter experts without technical knowledge. There is a widespread idea in many companies that you can only understand artificial intelligence if you know how to program (e.g. R or Python) and also have advanced knowledge of mathematics and statistics (see figure). Fig.: How the different aspects of Data Science are interrelated. (Source: https://blogs.gartner.com/christi-eubanks/three-lessons-crossfit-taught-data-science/) This opinion is often given additional impetus by a “loud minority” of the data science community, which publicly celebrates a romantic image of “data science superheroes”. The topic is presented as complicated and only understandable for AI experts. This often leads to executives adopting an emergency vocabulary of AI buzzwords, outsourcing the whole topic to IT and data analytics teams, but otherwise retreating into their own comfort zone, firmly believing that the “business”, i.e. the professionals in marketing, supply management, production, risk and data scientists will understand each other and solve the right problems with AI. But IT is understood as an “enabler” in business management. Without clear goals and a concrete application orientation of managers to economic issues, AI offers no added value. It is an instrument to be used and not an end in itself.

AI yes, but do the right thing with it

Carl Hillier, a technology evangelist says: “You don’t make your company successful by buying a bucket of AI. It’s not about waving an AI wand – that doesn’t generate any actual value in itself. It’s about doing the right thing with AI.” And the evidence is there that there has been far too much wand-waving so far. According to a study by Dimensional Research, 8 out of 10 AI projects fail before they reach the deployment phase, and a third of all projects fail in the proof-of-concept phase.^[2] Just how do you get it right with AI? This question can be answered much more easily from a manager’s perspective once there is clarity about what AI is, how it works, what it can do and under what conditions it delivers good results. There is no reason to be afraid of AI technology. Today’s AI tools allow users to manipulate data, experiment with it and create predictive models without any programming knowledge. The goal in using such tools is not to become a Data Scientist, but to gain a better understanding of the workflows and common problems faced by AI teams. Terms like “Model Accuracy”, “Precision, Recall” or “Lost Function” should already be part of every manager’s toolkit. And not just as vocabulary, but as application-oriented understanding and professional competence.

Generalists interpret between internal stakeholders

One of the first signs that the problem has been recognised at the management level and that the gap between the “business” and the “data science teams” is being closed is the emergence of the role of “Analytics Translator”. Analytics Translator is a generalist role that is equally at home in data analytics as in business topics. It translates between the two areas in a similar way to the role of the business information scientist. With the support of the Analytics Translator, leaders identify and prioritise the business problems with the highest positive impact for the organisation that can be solved with data.^[3] However, an analytics translator is only an intermediate stage on the way to the “organisation of the future”, in which every manager has basic AI skills and does not need a translator to get the most out of AI. Ideally, leaders have these skills:

• You know the potential of different AI/ML techniques for business process optimisation in different business areas.
• They understand how to describe their own AI use case based on the methods such as the “ML Canvas” and how to initiate an AI project.
• You have a broad understanding of different roles within an AI team and know how to build such heterogeneous teams from business and technology and how to work with them in a goal-oriented way.
• You understand regulatory and ethical challenges of AI technologies and can assess these for your company.
• You will be able to foresee future changes in the business environment due to AI technologies, react adequately to them and use them accordingly.

According to a study by PWC, the greatest positive effects of AI on companies are expected in the following areas:

1. Business process automation,
2. Increasing productivity and
3. Increase demand through AI-assisted supply personalisation.

PWC estimates that in the next ten years, the contribution of AI to global GDP will grow to 14% or nearly $16 trillion. 4] The transition into the AI era and the realisation of this immense potential can only happen with AI-competent leaders and subject matter experts. Therefore, it is now time to put down the AI magic wands. AI is not a magic wand, but a tool, a powerful tool that can only unfold its power in skilled hands.

AI in continuing education at BFH Wirtschaft

Starting next year, BFH Wirtschaft will be offering the CAS Artificial Intelligence for Business. Thanks to the balanced mix of proven theoretical and application-related concepts, systems and tools (frameworks), active, practical ML exercises and current AI case studies, participants will be able to participate in discussions at the same level as data scientists in their organisation, actively initiate and lead AI initiatives and projects and maximise the potential benefits of data for their organisation. They will become fluent in the language of Data Scientists and AI infrastructure managers, enabling them to represent and advance the vital interests of their business unit. For more information on the CAS, click here

References

• 1] Ten red flags signalling your analytics program will fail – McKinsey & Company (2018)
• ^[2] “Artificial Intelligence and Machine Learning Projects Obstructed by Data Issues” Dimensional Research (2019)
• ^[3] “Analytics translator: the new must-have role” HBR (2018)
• [4] Sizing the prize What’s the real value of AI for your business and how can you capitalise? PWC (2017)