“My interest here is really in the interaction between AI and humans and the extraordinary importance in human decision-making with context plays,” said Rory Sutherland, opening a fascinating session the event organized by Shoosmiths.

His remarks delved into the fascinating interplay between AI and human decision-making, highlighting the crucial role of context and the dangers of blind algorithmic reliance. He argued that while AI can offer valuable insights, it often lacks the nuanced understanding of human behavior and situational factors.

Real-world conditions

Sutherland used the example of sat navs (GPS) to illustrate this point. While they can suggest ingenious routes, they often fail to account for real-world conditions like weather or personal preferences. He also raised the ethical limitations of AI, citing the example of Google Maps allegedly refusing to offer scenic routes because programming variables classed as scenic would involve introducing a set of biases that could lead travellers away from low-income areas on the presumption that they were more likely not to be scenic.

He further discussed the limitations of crude AI, such as speed cameras, which operate without considering context. While acknowledging their general usefulness, he pointed out instances where they fail to account for mitigating circumstances, like avoiding an accident, or systemic problems like poor sign placement.

Sutherland argued that businesses often suffer from “quantification bias,” prioritizing easily measurable numerical metrics over intangible human factors such as emotional wellbeing. He cited the “SCARF” model (Status, Certainty, Autonomy, Relatedness, Fairness) to emphasize the importance of these emotional metrics in decision-making.  

Drawing parallels with criminal investigations, where initial stages rely on anecdotal evidence and observation of outliers, rather than immediate, conclusive proof, he challenged the prevailing business culture that demands “irrefutable proof” before action, arguing that this “stifles innovation” and “serendipitous discovery.”   

Black Box thinking

A key concern Sutherland raised is the potential for AI to stifle human judgment. He noted that while we readily accept “black box” thinking from computers, we demand rigorous justification for human decisions.

Sutherland explained the lack of “Black Box thinking” within organizations, “every single person before they can do anything in the modern corporation has to win an argument first.” Furthermore, that argument is usually with finance and it’s often impossible to win because of the of level of proof that is needed. However, people “are fairly content with Black Box thinking when it emerges from a computer.”

This creates a situation where individuals are incentivized to follow AI recommendations, even when they know they are flawed, to avoid legal repercussions.

Explore v exploit

Sutherland introduced the concept of “explore versus exploit,” drawing inspiration from bee behavior. He argued that businesses, like beehives, need a balance between optimizing existing processes (“exploit”) and exploring new possibilities (“explore”). He criticized the current focus on short-term gains and predictable outcomes, arguing that it hinders long-term growth and adaptability.  

He believes that AI could potentially shift the balance towards exploration by generating novel ideas that humans might not consider. However, he cautioned against blindly trusting AI, emphasizing the importance of human judgment and the ability to recognize and discard flawed suggestions.

Power of framing

In addition, Sutherland highlighted the power of framing and context in influencing human behavior. He cited the Italian restaurant that turned a disliked basement into a coveted “Capri Lounge” by creating an illusion of scarcity. He also praised Ronald Reagan’s masterful handling of age concerns during his re-election campaign, demonstrating how a well-timed joke can be more effective than a logical defense (Reagan rebranded his age as experience, suggesting his opponent – Walter Mondale – was at a disadvantage because of his youth).

Embracing “the weird” and “the slow” can lead to unexpected and valuable outcomes.

Sutherland argued that often, the most effective solutions lie not in objective improvements, but in reframing perceptions and understanding human psychology.  

Embrace experimentation

Sutherland advocated for a more nuanced approach to decision-making, one that integrates the strengths of AI with the irreplaceable value of human judgment, intuition, and contextual awareness. He urged businesses to embrace experimentation, challenge conventional wisdom, and recognize the importance of both exploration and exploitation for long-term success. Sutherland envisions a future where AI plays a more active role in facilitating better decision-making.

Furthermore, he proposed a shift towards “slow decision-making”, allowing individuals to refine their preferences based on what they discover.

Sutherland concluded by emphasizing the need to balance logical optimization with psychological considerations. He advocated for a more holistic approach to problem-solving, one that recognizes the power of reframing, understanding human behavior, and creating experiences that resonate on an emotional level.

He challenged the notion that all decisions should be fast and optimized, and suggested that embracing “the weird” and “the slow” can lead to unexpected and valuable outcomes.