Sensemaking & AI

Sense-making is an innate, biological urge to make sense of our lives. Humans devote an extraordinary amount of time, money, and attention to sense-making. Our behavior is driven by the drive to both feel positive about our lives and to live our lives in a way that makes sense.

Nick Chater, professor of Behavioural Science and George Loewenstein, professor of Economics and Psychology, suggest that the drive for sensemaking is under-appreciated. They argue that our sensemaking drive is as potent as other drives such as hunger, thirst, and sex. However, it only exists in our minds so is constrained by what our brains can do.

The drive for sense-making depends on cognitive state and external information. It operates via a carrot-and-stick mechanism—it doesn’t feel good when we can’t make sense of something and it does feel good when we feel like we are figuring things out. It feels especially good when we sense that a piece of information is key and explains many things or reduces overall complexity.

The experience of making sense of something, characterized by its tangible feeling, indicates it is a deeply creative process that often involves discomfort. This discomfort is necessary for us to appreciate the relief that comes from successful sensemaking.

Sensemaking often involves crafting narratives that help us perceive our lives as coherent and meaningful. The joy we derive from making sense of our existence is directly related to the significance we place on it, with the ultimate satisfaction coming from feeling that our lives genuinely make sense.

Memory plays a vital role here: human memory is not just a repository of facts but a dynamic, interpretative process that shapes our understanding of the world and ourselves. Sensemaking, in this context, involves drawing upon our memories to construct narratives that explain, justify, or give meaning to our experiences.

The recent announcement by OpenAI about introducing "memory" capabilities to ChatGPT marks a significant evolution in the relationship between memory, sensemaking, and artificial intelligence. The introduction of memory capabilities in AI could enable us to recognize and reflect on the patterns and transformations that occur throughout our lives more keenly.

For example: consider for a moment the evolving versus constant aspects of your identity. A common aspect of human consciousness is the belief in maintaining a consistent identity over time, despite the inevitable changes in our thoughts, behaviors, and even the physical composition of our bodies.

An AI companion equipped with the ability to recall and remind us of paths taken could provide us with varied perspectives of our "selves" at different points in time, enriching our sense of continuity and self-awareness but also challenging us to reflect on the insidious impact of long term habits, for example. With the ability to recall past interactions, AI can offer better advice and support, taking into account our previous dilemmas, decisions, and outcomes.

If the urge to make sense of our world is as fundamental as we believe, then it's essential to acknowledge that humans stand to lose if AI's capabilities either overwhelm us with too much information or strip us of our autonomy through excessive automation.

This might shed light on why mediocre automation leads to modest productivity gains at best—it leaves just enough room for human intervention without actually enabling us to leverage our natural inclination for sensemaking. We find ourselves trapped in a situation where ambiguity or failure necessitates human involvement, yet there's no real opportunity for us to apply our instinctive drive to decipher the problem. We're barred from exploring alternative solutions, ensnared in a scenario that is the very antithesis of autonomy, with our ability to make sense of things effectively crippled.

On one hand, we can intellectually appreciate the sheer brilliance of AI's capabilities, such as the discovery of novel structures and algorithms, which can indeed be awe-inspiring, as exemplified by the advancements made by teams like DeepMind. Yet, I find this sense of admiration is often accompanied by an equivalent unease. The "alien" nature of AI's thought processes and perspectives challenges my deeply ingrained preference for understandable, causally linked narratives which leaves me in a state of discomfort despite my recognition of its profound achievements.

On the other hand, AI's excellence at operating within a high-dimensional space can seem completely abstract. So I find that when AI lacks the ability to simplify complex data, theorize, or provide causal explanations in ways that make sense to me, I'm left grasping for more. The human inclination towards causal reasoning, amplified by storytelling and diagnostic analysis, predisposes us to seek explanations. We are fundamentally drawn to narratives that offer reasons and insights.

I've begun to see sensemaking as a process similar to dimensional reduction—a technical term, yet it feels applicable. It involves filtering through the complexities, discarding what's irrelevant, distracting, and entropic, and focusing on what truly matters. This task is challenging because it requires us to condense multiple factors into a single, coherent understanding, a task made even trickier by AI and a world of machine knowledge.

Sensemaking is one of the reasons I am fascinated with complexity. The very nature of a complexity mindset alters the way one sees the world. Foundational notions and features in complex systems provide metaphors, which, if used judiciously, offer explanatory power. For instance, how traffic flows at constraint—aka a traffic jam—offers insights into why decisions get stuck in organizations. (Hint: what matters most is the variability in individuals' speed).

Sensemaking is going to change. AI will allow us to find story-less, a-narrative yet meaningful correlations. We won't necessarily feel that fundamental satisfaction that comes with a causal story. Instead our minds will have to be open to a new kind of awe: that which a machine can make sense of that we cannot.

This won't make our lives senseless; on the contrary. It creates a whole new theater of thinking and action, new sensemaking opportunities and ways to decipher our own experiences and lives amid machines.