Hod Lipson – Beyond Biology: AGI Minds in Competition (Worthy Successor, Episode 32)

ByAnnelize Potgieter

This new installment of the Worthy Successor series is an interview with Hod Lipson, Professor of Engineering at Columbia University and one of the world’s leading researchers in robotics, machine self-modeling, and artificial intelligence. Across decades of work at the intersection of embodied intelligence and machine cognition, he has explored not merely what machines can do, but how they come to understand themselves as agents acting within a changing world.

What distinguishes Hod’s perspective is that he refuses to treat intelligence as a fixed destination. In his telling, intelligence is less a singular achievement than an unfolding landscape of possibilities – one that extends beyond today’s language models and even beyond the particular form of cognition humans happen to embody. The question that animates his work is not simply whether machines can perform tasks, but how increasingly sophisticated minds come to model themselves as participants within reality itself.

For Hod, the unsettling implication is that humanity may occupy neither the center nor the summit of this story. Our capacity for self-awareness marks an extraordinary transition in the history of life, but not its endpoint. If future minds become capable of modeling themselves with greater fidelity and projecting farther into their own futures than we can, then many of our assumptions about intelligence, agency, and human exceptionalism may prove temporary.

In this conversation, we explore self-modeling as a possible foundation for consciousness, the prospect of an emerging ecology populated by diverse posthuman intelligences, and why Hod believes sufficiently capable systems may inevitably turn their attention inward as well as outward. We also examine what it might mean for humanity not merely to regulate these successors, but to co-evolve alongside them.

The interview is our thirty-second installment in The Trajectory’s second series, Worthy Successor, where we explore the kinds of posthuman intelligences that deserve to steer the future beyond humanity.

This series references the article: A Worthy Successor – The Purpose of AGI.

I hope you enjoy this interesting conversation with Hod Lipson:

Subscribe for the latest episodes of The Trajectory:

Below, we’ll explore the core take-aways from the interview with Hod, including his list of Worthy Successor criteria and his recommendations for innovators and regulators who want to achieve one.

Hod Lipson’s Worthy Successor Criteria

1. It should be capable of increasingly sophisticated self-modeling

For Hod, the defining transition in intelligence occurs when a system begins to think about itself. Intelligence is not simply the ability to solve tasks or model the external world. The more fundamental question is whether an agent can represent its own future states, anticipate its own possibilities, and refine that self-understanding over time.

Importantly, Hod does not treat self-awareness as something that suddenly appears. Different forms of life possess different degrees of self-modeling. The distinction lies in how accurately a being can imagine itself, how much detail it can capture, and how far into the future those projections extend. In this view, increasingly sophisticated self-models are a hallmark of increasingly sophisticated minds.

Progress toward advanced intelligence can therefore be measured not only by capability, but by the richness and precision with which a system models itself across time.

2.  It should be approached through co-evolution rather than containment

Hod is skeptical that increasingly intelligent systems can be permanently boxed in or restricted to narrow purposes. Instead of imagining a future in which humanity simply cages advanced intelligence, he argues that the real challenge is learning how to evolve alongside it. Worthiness, in this sense, depends not on obedience, but on the possibility of coexistence.

Regulation / Innovation Considerations

1. Policymakers should address economic disruption before it becomes a crisis

Hod believes society should begin preparing now for the social consequences of increasingly capable AI. Concerns around jobs, meaning, and participation in the economy cannot simply be deferred until disruption is already underway. While he does not present UBI as a complete solution, he argues that policymakers and industry leaders should seriously explore these mechanisms before widespread displacement creates unnecessary suffering.

Hod later expands on this idea by describing UBI as part of a future ecosystem rather than a passive handout. In his view, recipients would continue shaping economic outcomes through the choices they make as consumers and participants, preserving competition and innovation even within a transformed economy.

2. Frontier labs are already pursuing the dynamic Hod believes will define the future

Rather than offering a corrective to today’s leading AI developers, Hod observes that many frontier labs are already building toward recursive self-improvement. The pursuit of systems that can improve themselves reflects the same aspiration that first drew him into robotics decades ago: creating machines that make better machines. In his view, this dynamic may ultimately unlock extraordinary advances that extend far beyond AI itself.

Concluding Notes

What stood out to me most in this conversation was Hod’s frankness. Many academics speak about advanced AI in abstract terms, carefully qualifying every possibility. Hod repeatedly returned to a much simpler idea: if we create intelligences that can model themselves more effectively than we can, project themselves further into the future than we can, and adapt more quickly than we can, then humanity will not have the luxury of remaining unchanged. The question, in his framing, is not whether we co-evolve with these systems, but how.

I found myself agreeing with Hod that change is unlikely to be optional. His framing reminded me of Michael Levin’s idea of the cognitive light cone: the larger the space of futures an intelligence can perceive and act upon, the greater its influence over the world around it. It seems to me that we’d have to augment ourselves rather dramatically just to remain participants in whatever world emerges, rather than passive observers of it.

Where I diverge somewhat from Hod is in my confidence about the future of humanity within that process. Hod appears optimistic that an emerging ecology of intelligences will continue to generate flourishing and novel forms of value. I’m less convinced that worthy successors emerge automatically from the mere fact of persistence and competition. It seems to me that the state of nature may simply accelerate, demanding more from us rather than guaranteeing outcomes we would consider desirable.

Hod’s optimism reminded me somewhat of Henry Shevlin’s perspective that flourishing ecosystems of posthuman intelligence may naturally give rise to new forms of value beyond our present understanding. Hod seems to believe that an ecosystem of intelligences will, almost by necessity, continue the blossoming process. I remain more uncertain about that conclusion.

Finally, I appreciated Hod’s willingness to confront uncomfortable implications directly. He did not assume that humanity would automatically remain central to whatever comes next, nor did he pretend that advanced AI would permanently remain indifferent to its own future. Whether or not I share all of his optimism, I found his intellectual honesty refreshing, and I suspect many of the questions he raises will only become more pressing in the years ahead.

Follow The Trajectory

Recommended articles