We all face mental biases when it comes to decision-making – especially those at the top. Thanks to AI, business leaders have an unprecedented opportunity to look at their brand’s data from an unbiased, third party perspective, and benchmark their company against the competition to get an accurate perspective of their place in the market.
The human biases faced by executives in our fast-paced world
Overconfidence bias occurs when business executives become overly dependent on their team’s processes and analyses, or when they become too reliant on their own ‘gut’ in the decision-making process.
In the former situation, the executive may make the wrong decision because he either relies on flawed analysis containing human error or misses critical elements of the business issue. In the latter situation, the executive might make an incorrect judgement because they did not leverage sufficient data and information to their decision-making process.
For example, Richard Fuld, Chairman of the famous Lehman Brothers, decided to refuse an offer from the Korean Development Bank to buy a stake in the troubled company during the recession. Dick did not listen to his advisors and financial analysts in terms of the valuation of the company and relied solely on his personal view that the offer was too low and that there would be other suitors. He was also overconfident in his view that the U.S. Government would bail out Lehman. Unfortunately, none of the scenarios he projected materialized and Lehman Brothers was forced into bankruptcy. Some still attribute this downfall to Fuld’s own hubris.
Confirmation bias ensues when a business leader has already made a predetermined decision which can be driven by a desired outcome, preference or emotion. This issue may be exacerbated by time and resource constraints but becomes problematic when the executive is no longer open to dialogue and analyses or to re-evaluating the situation. Unfortunately, the critical facts or nuances that were either intentionally omitted or ignored are material and should have resulted in a different decision.
Looking at international expansion in the retail market as a prime example, Best Buy executives demonstrated overconfidence bias when they decided to enter the European market with a ‘Big Box’ strategy. They wrongfully assumed that Europeans, like Americans, want large superstores with a broad selection of items at discounted prices, demonstrating a severe misunderstanding of their European customer. They believed this approach would give them a competitive advantage over the smaller, local European stores. However, this was not the case and Best Buy discovered this too late. The company closed all of its European stores in 2013.
Loss aversion occurs when a business decision or investment results in losses, and as a result there is a predisposition by the executive to undertake more aggressive risks in order to turn the loss into a profit. These risks can result in higher losses that can put the business unit, department or even company at risk. In these situations, the best decision may be to exit, close or sell the investment or operation to minimize losses.
A prime example is the historical collapse of Barings Bank, which occurred when head derivatives trader Nick Leeson took larger speculative bets on the Nikkei Stock Exchange after accumulating £208 million in losses in 1994. While attempting to make up for this mistake, he ended up incurring over £800 million in losses – more than double the bank’s capital. As a result of his actions, Barings Bank became insolvent.
‘What you see is all there is’
This sort of bias takes place when executives assume that the information they have and know is all that needs to be considered to make the best decision. This type of ‘tunnel vision’ can result in the executive missing something critical such as an important risk or opportunity which leads to the wrong decision and action.
A well-known example of this risk is illustrated by the fall of Research in Motion, the provider of the Blackberry smartphone. The company had an enviable position as global leader in the corporate smartphone market up until 2008. However, they focused solely on a single market segment, ignoring the larger consumer smartphone market which Apple and Google began to target.
Research in Motion was convinced their hold on the corporate and business market was ‘all there is’ and failed to anticipate that the immense popularity of the iPhone and Android products for the consumer market would cross over to the business segment. By 2011, the company was struggling as most of its customers had switched over to competitor products.
AI helps to counteract human mental biases
Every day, business executives face challenges such as dealing with complex issues, severe time constraints, processing multiple opinions and perspectives and managing political pressure.
These challenges can make it very difficult to overcome mental biases. Artificial intelligence can assist executives in making better decisions by providing unbiased, fact-based analyses and recommendations. It has also demonstrated an ability to uncover previously undetected drivers and relationships in the data which has led to better business solutions and outcomes.
In the case of overconfidence and confirmation biases, for instance, executives can deploy AI to ensure that all critical data available is analyzed and all underlying business factors have been evaluated appropriately. With a robust process for analysis, the tool will ensure that there are no hidden relationships in the data.
By utilizing AI to provide an objective, fact-based approach, business leaders are able to avoid taking excessive risks and neutralize this tendency through empirically driven analyses of the current state. The ability of AI to uncover hidden relationships in the data counterbalances the ‘what you see is all there is’ bias. Note that companies must have all of the relevant data available to effectively utilize AI for a business problem.
External data sources inform AI to offer business executives unbiased decisions
The value of AI is predicated on having large quantities of relevant and accurate data to train the neural network. There are a variety of data sources both within and outside the company that can fuel these algorithms.
With the growth of IoT, companies are facing a deluge of data, so the pressing questions have become more about ‘which data do I use?’ rather than ‘where can I get the data?’ However, sources of external data like partner firms, social media, research firms, consulting firms and other organizations can offer a variety of critical data on your products/services, equipment, customers, suppliers, competitors etc. This can be used by AI to assist in making better decisions.
One example is using AI to help measure and monitor the market’s sentiment toward a product line. An AI solution that analyzes external data can be used to search social media for consumer insights on the company’s product line. Natural Language Processing and a scoring algorithm would then be used to gauge the public’s attitude toward the product at any point in time. Business leaders should then use this information to inform their strategy.
Why it’s vital for businesses to incorporate an AI strategy
AI can benefit companies by reducing human error, increasing staff productivity, decreasing departmental costs and identifying risks and opportunities that were not previously detected.
Through objective analysis, it has the potential to counter emotionally-driven perspectives and political viewpoints that can bias executives.
With highly dynamic business landscapes and the increasing proliferation of data, it becomes impossible for companies to have certain business activities be performed predominantly by people without accepting a growing risk of human error. As the business environment becomes more complex, executives risk making business decisions based on models that do not fully capture the underlying business reality.
However, it should be noted that AI can be biased if the data that is fed into the system for pattern recognition and solution optimization contains human biases.
One example of potential bias is in the area of hiring and recruitment. If historically a company has hired males for a specific position, even if this is only due to males traditionally applying for the role, the AI tool would detect ‘being male’ as one criteria for candidate hiring, placing females at a complete disadvantage.
To ensure that the algorithms are not plagued with bias, companies need to make certain that their data is properly vetted and cleansed. In this case, we can address the issue by removing the gender type from the data set and replacing individuals’ names with sequential numbers. However, companies are finding that by migrating to an AI solution, they can better understand their current criteria for decision-making which can help uncover biases that they may not be aware of themselves.