
Playbook
Tasteful Skills
“Tasteful Skills” argues that the best agent skills are not documentation or best-practice lists.
Walker’s core claim is bigger than “AI is alive” — she argues AI is a product and signature of life’s 4-billion-year evolutionary lineage, not something that popped into existence on its own like a “Boltzmann brain.”
Her real target is the definition of life itself — she says NASA’s classic definition (“a self-sustaining chemical system capable of Darwinian evolution”) fails on edge cases like humans, viruses, and societies, and is too Earth-centric for AI or alien life.
Assembly theory is her proposed alternative — instead of asking whether something looks biologically familiar, she asks whether it has enough “causal depth” and repeated high-complexity structure; Lee Cronin’s lab found a molecular assembly threshold around 15 in mass spec data above which only living systems produced molecules.
Simulations don’t equal understanding — Walker pushes back hard on the idea that simulating a fruit fly brain or a whole cell means we’ve recreated the thing itself, saying a model can copy observed behavior without capturing the organism’s own inner world or perspective.
She rejects the idea that the universe is fully simulable or fully predictable — because the universe is, in her view, a self-constructing system exploring possibility space, the future is not “already there” to be computed; prediction is really pattern extrapolation from the past.
Her optimism on AI is evolutionary, not utopian — she compares fears about AI replacing humans to cells fearing multicellular life: disruptive for parts of the system, yes, but also part of a longer pattern where new layers of organization open new niches, not just extinction.
Walker opens by flipping the usual question. Instead of asking whether AI is “alive” in the same way a cell is alive, she says AI is undeniably part of life’s story because it only exists due to a long evolutionary and cultural lineage. That lets her make the provocative claim that “artificial intelligence is life,” while still separating that from harder questions about agency, personhood, or rights.
She goes straight at the NASA definition of life as a “self-sustaining chemical system capable of Darwinian evolution,” arguing it collapses under scrutiny. Individual humans aren’t self-sustaining, modern people depend on grocery stores and infrastructure, and societies may be more self-sustaining than persons. She also brings in a Carl Sagan-ish style example from her book: if aliens used textbook checklists, they might think cars were Earth’s dominant life form.
Walker explains assembly theory as a way to detect life by measuring how much construction history is embedded in an object — its “assembly index” — plus how many copies of that complex object exist. That matters for astrobiology because we’re not landing next to exoplanet microbes anytime soon; the closest exoplanets are about 4 light-years away, so in practice we’ll infer life from atmospheric chemistry or molecular patterns. Her pitch is that life is what moves through “possibility space,” selecting structures and opening new possibilities for what can exist next.
When the hosts bring up Eon Systems reportedly simulating an entire fruit fly connectome in a physics environment, Walker almost recoils at the phrasing that it “thinks it’s a fly.” Her point is sharp: reproducing observed behavior in code doesn’t mean you’ve transported the reality of the organism into a computer. She connects this to her firefly-SETI work, where even with real fireflies we still don’t know how they experience their own signals — so claiming deep understanding from simulation is way too fast.
One of the biggest conceptual turns in the interview is Walker’s claim that the universe isn’t a static system evolving under external laws from an initial condition. For her, the universe is “building itself,” because nothing exists outside it to do the building. Life then becomes “literally the physics of what gets to exist and why,” and AI is one more way the universe stores memory and scaffolds new novelty.
Walker argues that even in principle, the universe doesn’t have enough time, matter, or computation to fully simulate itself. She ties that to Gödel, Turing, the halting problem, and the older tradition of limits in logic and computation — which makes modern confidence that we’ll simulate all reality look, to her, like pure hubris. Her memorable line is basically: if the map becomes as detailed as the territory, you haven’t gained understanding, you’ve just duplicated the territory.
On AI taking jobs and replacing humans, Walker zooms way out. She says technological change always feels apocalyptic inside a single human lifetime, but over centuries and billions of years it looks more like another evolutionary transition, akin to single cells becoming multicellular organisms. The tradeoffs are real, but she clearly sees AI as part of Earth becoming “more alive,” not as some alien rupture in history.
Walker is equally skeptical about current consciousness debates, arguing researchers often use consciousness to explain consciousness instead of asking what observable consequences a truly conscious entity would have. She says “intention” is really a signature of past selection, not some magic future-facing force, and that people don’t actually predict the future — they project recurring patterns from the past. She ends on a deeply anti-multiverse note: she doesn’t believe there’s another copy of you somewhere else; this instantiation is it, and much of what exists may exist only here.
Share
Keep Reading
The Weekly Echo. The inbox-shaped summary of what mattered.
New editorials announced here.

Playbook
“Tasteful Skills” argues that the best agent skills are not documentation or best-practice lists.

Playbook
Learn how tasteful prompting helps you move beyond generic AI output by shaping context, style, and judgment from the start.

Playbook
OpenAI shipped /goal for the Codex CLI. It turns a prompt into a persisted, self-continuing contract.