How People and AI Are Now Shaping the Future of Business Decision-Making

Written by: Jasmina Očko, consultant for controlling at Kontroling Kognosko

When we talk about managing companies, about making operational or strategic decisions, there have been high hopes placed on artificial intelligence for a long time. No one doubts that today’s reporting tools will become increasingly better over time. We expect a lot from the analysis of large amounts of business data, but even more from recognizing patterns and suggesting concrete solutions to managers. Only controllers who compile reports daily using current tools can truly understand what kind of help this can be. Both in quantitative and qualitative terms.

Today’s tools follow given instructions but do not think for themselves, so we can say they are actually quite ‘dumb’. Artificial intelligence – in contrast – learns and becomes smarter. It learns based on repetitive data, recognizes patterns, and adapts them to new situations, thus providing insights into future scenarios. Of course, we are talking about prediction, not magic. Therefore, traditional analyses provide answers to the question ‘what happened’, we can, for example, observe correlations among phenomena, statistically calculate trends or deviations, and partially understand why the result is what it is.

AI is expected to focus on predicting future trends much more precisely than mere statistics and to provide recommendations for the right decisions. We are moving from the question ‘What happened?’ to a space where we discuss ‘What could happen and what do we need to do now in that context?’. However, the path to that is much longer and harder than it seems. Many think that AI will be a magic wand that will improve our decision-making on its own, patch up our imperfections, or do what we did not because we did not know, could not, or simply did not care. And the truth is far from that. An exciting future full of improvements awaits us, but also challenges and limitations.

Advantages and Challenges

What advantages can AI bring to controlling for creating smart reports, and then to management for making smart decisions?

• It is expected that data from various sources, both internal and external, can be analyzed simultaneously (in real-time). A trivial example: most of today’s tools have great difficulty linking data from the same IT system into a single table, for example, for one customer – sales data and profit achieved with the customer’s debt data.

And where are we when we want external connections with their publicly published GFI reports? We currently obtain such data manually, which is extremely slow, so such endeavors are only carried out periodically and only for the most problematic customers.

• We want AI to also provide us with concrete action proposals. Not only should AI connect the desired internal and external data in seconds, but it should also be able to answer the question, for example, what sales conditions we should grant to a customer, whether it is worthwhile for our field sales representatives to continue visiting them, and which segment of our offer we need to change and how. Of course, in addition to operational proposals, AI is expected to think at a higher level, providing strategic proposals. In our example, this could be, for instance, a much more effective larger investment in market A than in market B because it has a greater overall potential of customers (based on all internal and external data).

• We are confident that AI will bring us (to a large extent) automation in data collection. Many of today’s business data is actually unusable because it is prepared and managed by people, and people sometimes make mistakes, forget, or do not care. In many companies, the rules based on which data would be managed in a safe, consistent, and efficient manner to be accurate and usable are still not clearly defined. In this sense, AI is necessary because if data is not managed well, no subsequent step in the analysis can be good (garbage in – garbage out). • Automated reporting is expected. This specifically means that the tool answers specific questions from controllers or managers in natural language instead of programming languages or formulas and immediately creates tables, graphs, and comments. Questions such as ‘What are the main reasons for deviations from the revenue plan in the last quarter?’, ‘Which products achieved the highest sales growth compared to last year?’ or commands such as ‘Analyze the increase in operating costs in region X’ should receive answers in seconds. Similarly, it could be expected that these answers appear in the form of tables and graphs that immediately take the format of reports.

• We hope that forecasts will be good (as much as possible). This would significantly improve today’s planning and reduce uncertainty. Of course, no one can know exactly what will happen tomorrow, but AI is expected to be able to estimate much more accurately based on recognizing certain patterns and again based on linking internal and external data. If some estimate does not turn out to be accurate due to unpredictable changes, adjustments could be very quick and agile. Today we have planning tools, we have various scenarios, but any changes to plans and scenarios still require a long and laborious job for controllers and immense knowledge in navigating a large number of interconnected tables. Ultimately, the goal is for artificial intelligence as a management support tool to provide the company with a competitive advantage and facilitate the identification of opportunities and threats, strengths and weaknesses. In a time when business opportunities are born and die overnight, in an era when nothing is certain, such support could indeed help. However, whenever we talk about the potential advantages that AI will bring, we must not forget to emphasize the possible negatives. And we already know that it can also bring many challenges and limitations.

• The quality of data and the reliability of systems above all. We have a huge amount of data, but debatable and in silos, which causes irregularities in reasoning. The quality of data is likely to improve, but it will never be perfect because its collection will likely never be fully automated. Human intervention and supplementation will certainly be needed for a long time. Inconsistent data quality can undermine management’s trust or result in wrong decisions, which calls into question the justification of the entire investment in the development of artificial intelligence.

• For AI to be used to assist management, people need to have concrete competencies and knowledge. And they need to know what they want. AI is as smart as the person using it. It cannot bring knowledge and skill; it can only enhance the effect of the knowledge we already have. Developing competent multifunctional teams takes years, and such a leap cannot be mastered overnight. Technology is advancing extremely quickly, and the question is how quickly employees, not just controllers and managers, can keep up.

• The next challenge is resistance to change. It is natural for a person to feel most comfortable in a familiar environment and comfort zone. However, in the AI environment, everything is anything but comfortable. Continuous learning is required, teams need to be flexible, agile, and ready for constant changes without fear of being replaced. The key to success is the concept of augmented intelligence (engl. augmented intelligence). This is the joint action of people and artificial intelligence in which AI learns from people and people learn from it. Or, in other words, AI helps people, not replaces them; it frees them from routine tasks and allows them time to focus on acquiring new knowledge and generally developing a learning culture.

• One of the bigger challenges is the black box algorithm, that is, the fact that it is difficult to understand how the machine thinks. While it is nice to imagine that AI gives us suggestions and action proposals in an instant, it is indisputable that a person must be able to understand and interpret those suggestions. AI means nothing if the logic by which the machine thinks is not understood: ‘If I can’t trace the logic, I can’t trust the result.’ If we know nothing about the situation, we uncritically accept everything the tool provides us, which can certainly be detrimental. The more expert we are, the more we question the results; the better we master the situation, the more critical we are, but then we can also use AI more wisely.

• Investments in the development of artificial intelligence systems should truly be considered a strategic task, but this development cannot be placed in concrete time frames because there is no defined scope or concrete end. Therefore, the fundamental conditions are a clear strategic plan (what comes first, what second…), a defined personnel policy (what kind of staff we want to develop), and a defined long-term source of funding (which is a limiting factor for many companies). The greatest threat is the indiscriminate and uninformed rush into projects just because they promise some involvement of artificial intelligence. This must be avoided at all costs.

• Decision-making can be easier than ever, but everything can easily get out of control. Artificial intelligence that is being developed to support management starts with data and ends with a decision. The goal is to help management conclude faster and easier in the future. But behind every decision will still stand a flesh-and-blood person. And it is human nature to take the path of least resistance. When AI tools offer ready-made solutions or recommendations ‘with one click’, it is easy to accept the result without thinking. One should not even think about where that might lead us.

27885

Both Data and Intuition

And in the end, despite all current and future artificial intelligence tools, we must not forget that managers in real situations often make decisions based on some of their completely unrelated beliefs and biases. This decision-making concept is called ‘behavioral controlling’. In fact, it is easy to become aware of it – it is a situation in which managers only look at data that confirms their views and ignore those that contradict them; when they give greater importance to recent data in decision-making even though they have no long-term impact at all, or when the opinion of an authority that is important to the manager significantly influences the decision more than actual data.

Strong intuition is the main tool for a large number of managers. Many feel overwhelmed by a pile of information that they do not need for decision-making at all, and some do not have a great desire for speed and actually find it difficult to make decisions. Some are driven by ego, others by pure emotion, which probably no AI will change.

The Key to Success is in People

The beginning of our future benefits from artificial intelligence lies in the quality development of data infrastructure, accountability, and a highly developed awareness of data management. There is no business management without data, and there is no data without data management. Artificial intelligence is significantly changing controlling and management, enabling faster, more precise, and informed decisions. However, the key to success lies in people, that is, in a strong synergy between technology and human expertise.

AI brings many advantages that a person could not reach alone, but a person’s concrete and practical knowledge and experience guarantee quality data interpretation. Companies that recognize this combination and are now developing solid strategies for artificial intelligence development, investing significant resources in both technology and employee education, especially managers, will undoubtedly have a significant competitive advantage in the future.