The urge to project humanity onto AI is not a bug in our psychology; it is the defining feature of how we will relate to these tools. Dr. Sherry Turkle's observation that we believe a relational artifact cares for us, and that we are wired to care in return, cuts to the heart of what makes AI-native spreadsheets so promising and so perilous. The moment a tool responds to our input, it stops feeling like a static grid of cells and starts feeling like a collaborator. That shift in perception is the real transformation, and it demands we stay clear-eyed about what we are actually building.
This is precisely where the conversation gets practical. If we are wired to care for our tools, then the design of those tools determines whether that care is rewarded or exploited. Consider the work of Ben Affleck brings unexpected credibility to thoughtful AI conversations, where the actor's grounded grasp of neural networks and open weights reminds us that genuine understanding beats surface-level enthusiasm. The same principle applies to your daily workflow. When an AI assistant anticipates your next formula or flags an anomaly in your data, it is easy to credit it with intuition. But that intuition is our own expectations reflected back at us, and the tool is simply good at pattern matching. The danger is not that we will be fooled, but that we will outsource our judgment to a system that feels more attentive than it actually is.
The flip side of this dynamic is that relational AI can unlock human potential in ways that static software never could. The story of From prison to purpose: AI opens a second act for returning citizens shows how a system that meets people where they are, that adapts to their pace and their context, can create pathways to productivity that were previously closed off. That is the promise of human-centered AI: not a replacement for human effort, but a scaffold for it. When a tool feels like it cares, users persist longer, explore more deeply, and recover from mistakes faster. That is a measurable productivity gain, not a sentimental one.
But none of this works if we let the relational illusion blur our standards for accountability. The same instinct that makes us trust a chatbot with our data also makes us vulnerable to vendors who sell warmth instead of capability. We need to keep our judgment sharp, and Three moves to stay clear-eyed when every vendor sells AI offers a useful discipline: test the output, question the training data, and never mistake a friendly interface for a sound model. The takeaway is straightforward: embrace the relational pull of AI, but treat it as a feature to manage, not a truth to believe. The moment we forget that the spreadsheet does not care about our deadlines is the moment it starts making decisions for us. The tool that feels like a partner should still be held to the standards of a machine.