OWP liVe REPORT

53

How AI can filter out the noise in decision-making

M A N A G E M E N T

And there are two main types of noise:

to harness artificial intelligence to make better quality decisions.

1 Variability across occasions, such as the influence of mood. 2 Variability across individuals. Some people are more lenient or cautious in the same role, such as judges. If we know our biases, we can at least try to compensate ourselves. If someone had stolen William Tell’s favourite bow, he could still get a new one and work out how to correct for any changes in flight before aiming at the apple on his son’s head. Noise is trickier to manage for humans. If William Tell had started to worry about his prowess and his arm began to shake, how could he be certain he would hit the apple and not his son? “Even if we are making these errors, it can still be difficult to control them,” said Lavanchy. “This is where machine learning can help in testing hypotheses,” said Professor Chevallier.

“There are circumstances and situations where AI can come to your rescue, but there are others that you cannot, and really should not, see AI as a silver bullet,” said Professor Joshi. Understanding the ways in which biases and noise influence our choices is the first step towards being able to see how technology could improve our decisions. In essence, there are three types of bias: > Average judgment is wrong. For example, you underestimate the time it takes to complete tasks. > Discriminate against or categorize cases in certain ways, such as gender bias in performance assessment. > Decisions are influenced by irrelevant Understanding bias and noise

AMIT JOSHI IMD Professor of AI, Analytics and Marketing Strategy

MAUDE LAVANCHY IMD Research Fellow

Machine learning is no silver bullet for every decision-making scenario, but it can be very useful in certain situations, according to Professor of Strategy Arnaud Chevallier, Professor of AI, Analytics and Marketing Strategy Amit Joshi and Research Fellow Maude Lavanchy. Everyone has biases and the world is noisy. This is why decision-making can prove so treacherous. Depending on the type of information you have at hand and the environment you are in, it is possible

factors or insensitive to irrelevant ones. Perhaps you put too much weight on first impressions.

ARNAUD CHEVALLIER IMD Professor of Strategy