We all know that AI already started to impact jobs. The next question is: how ? Here is an example with one core element of management : the decision making process.
A decision to be taken relies on 3 different elements: the logic and the statistics or more broadly the data. These first two can pair. Then comes the third one: heuristics. These are general rules, simple principles that work in certain situations without being systematically true and precise. Heuristics are alternative problem-solving tools to more complex methods. They simplify our lives, so we use them constantly. Logic, data and heuristics are used by the brain in two different ways. Enter the dual process theory “System 1 and System 2” (in Daniel Kahneman’s words). It refers to an overarching set of theories from psychology claiming that there are two distinct operating systems that govern our behaviour. This is structured around the idea that these two ‘systems’ (1 and 2) govern our perception, our beliefs, and our actions. Generally, these theories hold that:
• System 1 is fast and intuitive/experiential. It is an automatic process constantly running in the background (nonconsciously). System 1 processing is effortless (no strong demand on our thinking such as computational capacity and working memory). Because System 1 is designed to process information efficiently, it often uses heuristics, to combine information and activate behaviours. But in this digital era, information overload is a pain, so we aggressively filter. As a consequence, some of the information we filter out is actually useful and important. Too much data obscures the search for meaning. And lack of meaning is confusing, so we fill in the gap and our search for meaning can generate illusions. We, sometimes, imagine details that were filled in by our assumptions and we construct meaning and stories that aren’t really there.
• System 2 is slow and reflective: it engages controlled processes (conscious) relying on more reasoning (rule-based, logical, analytical). Because this reasoning is more demanding, the thinking process is serial, slower, and effortful. System 2 is designed to make accurate decisions, so when it is “switched on”, decisions more often match rational norms.
You already understood that System 1 could be brilliant but is full of glitches: cognitive biases that pose problems for us. It relies on a limited rationality which can lead us to maintain erroneous beliefs. The worst part is that if System 1 is not always working well, System 2 can rationalize its conclusion, therefore reinforcing the vicious circle. Even if judgment and justification are two separate processes. The judging is powered by System 1. It happens almost instantaneously. It’s then justified in a rational and plausible language supplied by System 2. Knowing that, what role, if any,;) should people play in making decisions? The biases and bugs of System 1 will drive us to rely on a digital System 2 in some decisions. Objective formulas are, indeed, more effective than human judgment in recurring decisions and prescriptions. For these “simple” decisions, let AI take the lead on making decisions or judgment, predictions, diagnosis and so on. Then let people take the lead if others need to be convinced or persuaded to go along with these decisions. Algorithm are better than most of our judgment because machines are able to ingest all of the data. AI can recognize patterns in data and then act or make suggestions based on the patterns it found. This patterns recognition is beyond human capabilities in some specific areas. But today, artificial intelligence is not broadly intelligent in the following sense : a very general mental capability that, among other things, involves the ability to reason, plan, solve problems, see abstractly, comprehend complex ideas, learn quickly and learn from experience. AI and most precisely machine learning can just tick the last two boxes on the list: learn quickly and learn from experience. For these “complex” decisions, AI will enhance analytical insights and prepare decisions to be taken by humans.
Be they simple or complex, AI and big data will automate a growing number of human decisions in all sectors of activity.