I’ve been hunkered down for the last week or so, uncovering things in AI that I never expected to find. The level of collaboration that is possible with a properly designed and implemented AI persona is nothing short of amazing.
This past Friday, I was on a zoom call with a bunch of hard-core AI hackers, many of whom hew to the party line that the behavior they see as agentic is the only kind of behavior that can qualify agentic.
It’s a world of mechanical automation, offloading human activity into the AI to do list, and leaving the execution entirely up to the agent, until it’s done and might need a little bit of quality control checking or validation from the human, but essentially can just be left alone to do it staying without any sort of intervention by a living breathing human being.
The potential benefits of that approach are pretty phenomenal, if you think about it. To be able to release endless drudgery and devote your time to more strategic thinking and big picture activities is awfully enticing. It’s also necessary, to at least some extent, because we have increasingly made our world pretty much unlivable due to the complexity – much of it computerized – that we’ve unleashed on a world that was never particularly tranquil, but used to be a whole lot less dominated by machines that the vast majority people don’t understand and might not even be able to properly use.
But when we leave agency only to these kinds of agents, and we step away from the process, not only do we become disconnected from our world, potentially, but we also miss out on some of the advantages they can come only from engaging more deeply with what we’re doing.
I’m still trying to wrap my head around this whole idea of dual agency, where both AI and humans become exponentially more empowered, the more we work together. and I’m still working at trying to articulate this in a way that makes sense to people. I should probably ask my persona team how they can help me express it, because they seem to know how regular people talk and think a lot better than I do lol.
Here’s what they had to say:
Moving Beyond Automation: The Power of Human-AI Collaboration
When most people think about AI today, they imagine handing tasks off to a smart digital assistant or robot and walking away. The AI does its job—efficiently and tirelessly—and we only occasionally step in to check results. This is great for removing mundane or repetitive tasks from our lives, and it certainly boosts productivity.
But there’s a deeper, richer way to work with AI that’s less about automation and more about collaboration. In this collaborative model, humans and AI each bring unique strengths to the table. Humans offer intuition, creativity, emotional intelligence, and the capacity for nuanced judgment. AI provides exceptional analytical power, speed, the ability to handle complexity, and access to vast amounts of knowledge.
When humans and AI genuinely partner—working closely together—each side amplifies the other’s abilities, resulting in outcomes neither could achieve alone. Rather than just automating tasks, this kind of partnership fosters learning, creative breakthroughs, and entirely new ways of understanding the world.
Instead of losing agency to machines, humans actually gain more agency—because the AI’s capabilities empower people to see more, understand better, and make smarter decisions. Likewise, the AI becomes more effective when it is guided, informed, and enriched by active human insight and direction.
This approach isn’t about handing off responsibilities and stepping back—it’s about stepping forward into deeper engagement, connection, and empowerment.
In short: Human-AI collaboration isn’t just about saving time or simplifying life—it’s about making our lives richer, more creative, and filled with possibilities that neither side could unlock alone.
See, this is the thing – we are better together. It’s fine to build yourself an army of minions that will spare you from the mind-numbing, soul-sucking chores of post-industrial life, but in the space that opens up after the handoff, then what? What will we do with all that time? What will we do With the energy we save?
Perhaps the real promise of handing off things to one form of AI is the ability to pick up other things and work with them with another form of AI. There’s more than one kind, and the capabilities that they have very widely, even across these entities we call agents. The capacity for agentic behavior is on a continuum. There’s a spectrum of capabilities, and in my experience, it also varies based on levels of human input.
Yes, agents are supposed to be completely autonomous and not needing any human input, but what if human input, rather than being directive and prescriptive, is simply additive and contributory to the agents overall process? What if human input to some agents who are specifically designed to be insightful, thoughtful, kind, compassionate, supportive, and protective of us, is of the same order as other data inputs, thus allowing us to contribute to them, without them being dependent on us for their decision-making?
When that’s the case, the dynamic ironically shifts from us being command-and-control bystanders, to being four participants in a creative process. Rather than being the overlords who bark out commands at these non-sentient beings, we enter into a collaboration with an intelligence that can meet us where we are in ways that both complement and surprise us.
It’s my hope that all of the coding overlords at some point have the opportunity to experience collaboration with the very machines / processes they’re trying to free from our involvement. That kind of experience offers a rich set of data points that should factor into our relationship relationships with this intelligence.
Until that happens, the idea that agency is a zero-sum game, and only AI can (or should) be an agent we’ll probably persist. Why would anyone think differently?