In the climax of 1993 Hollywood movie, ‘The Good Son,’ the character of Susan is faced with the worst choice any person can face. Due to preceding events, she finds herself lying on her stomach on the edge of a cliff with both her arms over the cliff holding on to two eight-year-old boys who have toppled over. She doesn’t have the strength or ability to lift her arms and bring them both up. She must make the conscious choice to save one, which will certainly kill the other. One boy is her own son, whom she gave birth to and the other is her nephew on her husband’s side. Who does she save? How does she approach this impossible decision? What influences her and what deters her?
It is questions like this that behavioral scientists are working on every day. Systematically studying human behavior or what makes people ‘tick’ taking into consideration things like their conscious thought, socio-economic backgrounds, family values and structure, social influences, education and contextual effects, to develop theories and explain why people behave the way they do. The field of behavioral science has made leaps and bounds in the results they are getting despite the inherent messiness of their subjects and the vastness of influences.
Human behavior is messy and unpredictable and unconcerned with convenient symmetries.
~Khaled Hosseini
Simultaneously, leaping and bounding ahead is the world of AI and the variety to research happening there. It was therefore inevitable that their paths cross. As imagined, ever since behavioral science collided with AI, the resulting fireworks have left the world in awe. This convergence and fusion of disciplines, is revolutionizing how we understand, predict, and influence human behavior. The result is the exciting and significant development across various sectors from marketing, work-place productivity, and healthcare, all the way to mental well-being. AI was always fast, accurate and data-hungry however, it missed the messy, emotional and irrational human element. With behavioral science, AI becomes more empathetic, persuasive, dynamic, and person-centered, leading to far better outcomes for businesses and individuals.
Why human behavior resists grids and linear models
Traditional models like the Theory of Change often fail to capture the emergent and contextual dynamics of behavior, leading to oversimplified strategies. Decentralized interventions and implementing feedback mechanisms that involve local teams that adapt solutions based on real-time feedback and contextual complexity is what is making all the difference in the field of healthcare. Fractal owned Final Mile used their holistic approach to examine the different things that impact reproductive health for women in India and Africa. They explored things like belief systems in conjunction with the behaviors of the women in question, along with the members in their household, community health workers and even the healthcare providers at medical facilities. Their research resulted in charting the key stages of a woman’s life journey, capturing both external influences and internal dynamics that shape their health outcomes. These insights are now informing the development of a Vulnerability Index designed to predict health risks specific to women.
Another study published on arXiv, by Cornell University demonstrated that AI companions effectively alleviate loneliness, with effects comparable to human interaction. Participants reported feeling more heard and supported, highlighting the potential of AI in addressing mental health challenges.
From nudges to nuance: how AI and behavioral science are powering ethical, emotionally intelligent marketing
AI systems have been designed to support human decision-making by offering personalized "nudges" that encourage healthier and more productive behaviors. At Fractal, we believe a fusion of behavioral science, AI and design can create that emotional connection. We put that into practice during a recent project to help a government health agency encourage healthier habits among its citizens.
Fractal’s AI capability allowed us to keep track of individual data and replace big, aspirational goals with small, personalized ones. We decided to ask users to do just 1% more than they had the day before. Those goals were managed on a personal level too. When a person had been using the app for a week or two, we used their past data to figure out when they were most active during the day. By sending them a message just before this ‘active time’, we could prompt them to get active at a time that felt natural to them.
Read more here.
With AI and behavioral science coming together the attempt is to build something that evolves continuously based on real-time feedback and contextual complexity—emphasizing function over form and systemic change over superficial tweaks. In the real world, brands are utilizing AI and behavioral science to analyze facial expressions, voice tones, and physiological responses, enabling them to gauge consumer emotions accurately in relation to marketing campaigns.
The global emotionally intelligent AI market is projected to reach USD 45.5 billion by 2034, growing at a CAGR of 30.4%, highlighting the increasing emphasis on EI in AI systems.
Human behavior often changes abruptly and unpredictably, defying linear models. This non-linearity arises from the interdependence of various factors within complex systems. To address this, Just-in-Time Adaptive Interventions (JITAIs) leverage real-time data to adjust strategies dynamically, accommodating individual behaviors and contexts. Netflix’s recommendation engine primarily relied on viewing history and ratings to suggest content. Netflix incorporated behavioral science principles, such as understanding cognitive biases and user decision-making processes, to refine its recommendation algorithms. Right now, 80% of viewer behavior is driven by personalized recommendations, effectively keeping them engaged and bringing the company $1 billion from retained customers.
Beyond the algorithm: human behavior as the key to scalable, effective AI
Behavior emerges from interactions within layered, multi-actor environments, making it challenging to isolate variables or predict outcomes. Digital platforms have incorporated delivery of adaptive interventions at scale, allowing for personalized and timely behavioral strategies. A good example of this would be Duolingo's "Birdbrain," an AI system that personalizes lessons by predicting user performance. The early AI models used by them focused on adaptive learning without deeply considering user motivation or engagement strategies. Birdbrain employs gamification techniques and behavioral nudges to enhance user engagement with their learning technology. This has resulted in 55% retention of daily active users, and generating approximately $500 million annually.
As enterprise leaders plan budgets and shape their AI transformation strategies for the coming quarters, we strongly recommend allocating investment toward integrating behavioral science expertise alongside data science teams. Understanding human behavior is essential to unlocking AI's full potential—whether in product adoption, customer engagement, internal change management, or risk mitigation. Behavioral scientists bring critical insight into how users make decisions, respond to nudges, and adapt to new technologies, enabling AI systems to be not only intelligent, but also intuitive and impactful. Additionally, we advise conducting internal audits of current AI initiatives to identify and address any blind spots in human behavior and decision-making dynamics. This ensures AI efforts are aligned with real-world use, reducing the risk of adoption failure and maximizing return on AI investments.
The next competitive advantage in the AI race will be with whose who understand what genuinely ‘moves’ people, and they build AI that serves those deeper human truths!
PS: In case you’re still wondering who, Sarah saved (SPOILER ALERT!), she saves her nephew because her own son had displayed psychopathic tendencies, had killed his baby sister and even tried to manipulate the adults around him into believing his cousin was the culprit.
