Harnessing AI with Behavioral Science

Richard Chataway, Director of Behavioral Science at Concentrix, and author of “The Behaviour Business,” will be presenting at CCW Berlin on February 28 about the potential of behavioral science for harnessing AI and emerging technology. In this article, Richard looks at some of the ways in which behavioral science can begin to prepare us for this brave new world.  

It turns out the hype is real. There’s no denying generative AI is the next great leap forward in artificial intelligence. But for any technology to be effective, it’s important to understand what people really want from it. And that means first understanding how they really think—and behave.

As customers, we constantly make choices about how to interact with a business, from choosing a service provider to completing a purchase journey. But none of us make those choices like we think we do.

The groundbreaking work of psychologist Daniel Kahneman and colleagues, categorize decision-making into what he calls System 1 and System 2, won him a Nobel Prize in 2002.

• System 1 is fast thinking: instinctive, emotionally-driven, unconscious decision-making.
• System 2 thinking is slow thinking: rational, deductive, controlled, reflective decision-making.

While we’d all like to think we make most of our decisions in reflective, rational System 2 mode, behavioral scientists have proven that System 1’s instinctive fast thinking accounts for the majority of our daily behavior. They’ve also identified over 200 different cognitive short-cuts and biases that affect everyday decision-making—usually without us realizing.

Customers Are Not Perfectly Rational Beings

Accordingly, the vast majority of decisions customers make are hugely influenced by non-conscious human responses like trust, fear, laziness, even disgust. Businesses can reduce risk by taking advantage of this understanding of how people actually make decisions—often driven by non-conscious choices or external factors that they’re not even aware of.

However, the algorithms that run our world these days are based on logical, rational System 2-like rules, which stand in contrast to the System 1 thinking that people are largely ruled by (complete with our own biases).

Automated, machine-powered System 2 customer experiences are perfect for a whole range of simple queries that people want handled with simplicity, certainty, and consistency, as fast and as easily as possible, such as notifying a business about a change of address.

But for complex customer queries, that kind of rapid, simple response has its limitations. Take financial services for example, where customers might need to have conversations about stressful, emotional tasks (such as struggling to make mortgage payments or the financial aftermath of losing a loved one) for which empathy and the human touch are essential.

Customers will always have uniquely human characteristics that need to be factored in, from our need to be heard or our innate sense of right and wrong.

The New Technological Gold Rush

The rise of generative AI promises to supercharge the wave of automation, similar to the market’s previous rush to introduce chat-based technology. It partially reflects consumer trends of moving from offline to online transactions. But some businesses are rushing to adopt the latest technologies, having been largely sold on the financial benefits, without considering the longer-term impacts on well-being, brand perception, and profitability.

Building a digital solution without first considering the behavioral and human challenges will lead to a solution that prioritizes efficiency over effectiveness. In CX, where meeting the (frequently emotional) needs of customers is key, these challenges are particularly acute. It’s imperative to understand what distinguishes humans from machines, to build businesses that work for both.

The Human vs. Algorithm Problem

As humans, we assess the risks of our actions and predict the outcome based on experience. However, because we don’t have the same data processing capacity as AI-based systems—and have fallible memories—we often make decisions using simple rules of thumb (known in behavioral science as heuristics). And these can frequently lead us to behave in irrational, unpredictable, and sub-optimal ways.

By contrast, an algorithm (and the machine learning or artificial intelligence that relies on it) can only lead to rational System 2 decisions, because of the rule-based, predictable nature of those decisions. An algorithm is simply a rule, and humans frequently break them. Algorithms struggle to cope when confronted with some of the irregularities and irrationalities of human behavior—the very things that make us human.

That’s how we end up with some of the AI public relations disasters of the last few years, where smart technologies “learn” negative or inaccurate information or develop biases from human interactions.

That said, how might generative AI be different?

How Is Generative AI Different?

When I wrote my book “The Behaviour Business,” on how to apply behavioral science to achieve business success, the chapters on how to understand (and harness) technological advances like AI and bots were some of the most challenging to write because of the complex and dynamic nature of the topic.

With the recent advent of generative AI, this complexity has only increased, as has the potential of this powerful new generation of artificial intelligence. Up until recently, AI researchers primarily focused on creating basic algorithms that used predefined rules to process information. The results were often limited capabilities that needed a high level of human input.

The introduction of generative models is promising a revolution by creating large language models (LLMs) that enable machines to learn from so that they can generate new data, such as images or text, on their own.

Generative AI algorithms are based on deep learning techniques and neural networks that can understand patterns and generate results that can look and feel like content created by humans. This shift promises to unlock almost limitless possibilities for sectors like the creative industries, healthcare, and more.

The same human challenges will continue to exist though. That’s because any innovation designed and created by humans will always have our own cognitive and behavioral biases baked in. Generative AI, for all its groundbreaking sophistication, is no different. That’s because all technology created by humans, no matter how advanced or sophisticated, inevitably bears traces of the input data’s own biases.

While the opportunities from LLMs and other AI-based tools are hugely exciting, humans remain human. Businesses need to be aware of the risks of implementing new technologies without an adequate understanding of customer behavior. As the discipline of understanding human decision-making, behavioral science can play a key role in harnessing AI to maximize the potential return on your tech investment.

Join Concentrix at CCW Berlin to learn more about how to boost your business performance with behavioral science and meet with our team of experts to explore tech-powered customer journeys, multilingual CX for new markets, scaling businesses, and generative AI.