Xenobots and Anthrobots

What They Are

Xenobots: Synthetic living machines created by reassembling cells from African clawed frog (Xenopus) embryos into novel morphologies and configurations not found in nature. They are not robots in the traditional sense — they are living tissue organized in new ways.

Anthrobots: Similar synthetic organisms created from biopsied human cells. The same principle: human cells capable of building entirely novel structures and behaviors outside their normal developmental context.

Demonstrated Capabilities

Learning and Memory

Xenobots demonstrate learning — they form memories of events in their environment and retain them
These memories can be read out experimentally
This is demonstrated cognition at the level of individual xenobots

Self-Replication

Kinematic self-replication: Xenobots collect loose epithelial cells scattered in their environment
They assemble these cells into structured piles
Because the cells themselves have computational/bioelectric properties, these piles become the next generation of xenobots
The new generation continues this process (Vonnegut's self-replicating robot dream)
Note: Heredity not yet strongly demonstrated; form is reproducible but not yet clearly transgenerational

Novel Gene Expression

Xenobots upregulate several hundred genes not normally expressed in frog embryos
Notably: clusters of genes related to sound perception/hearing
When exposed to sound waves, xenobots respond to specific frequencies; embryos do not
This demonstrates that the new morphology enables entirely new sensory capabilities

Open Behavioral Questions

Levin explicitly states these are still being tested: - Specific goal states - Degree of intentional intelligence vs reactive behavior - Preferences and decision-making - Learning beyond memory formation - Any form of true agency or volition

Why This Matters

Xenobots/anthrobots demonstrate that:

Morphology is not deterministic from DNA — cells can self-organize into novel forms that aren't "programmed" in their genetic instructions
Living systems are programmable — cell collectives respond to bioelectric signals and can be "redirected" toward different goals
The substrate of intelligence might be generalizable — if frog cells can learn and organize in novel ways, what constraints limit possible forms of intelligence?
Ethics becomes urgent — if we can create synthetic organisms with unknown cognitive capacities, we need frameworks for understanding their moral status

Cross-Domain Connections

AI / Machine Learning: Xenobots are a proof that biological substrates can solve problems without explicit algorithms or neural networks. This challenges the assumption that intelligence requires neurons. Related to concepts of embodied cognition and morphological computation.

Materials Science: The ability to reprogram cell collectives to build specific structures suggests a future of "grown" materials and programmable matter — far beyond current 3D printing.

Philosophy of Mind: Xenobots are living counterexamples to the equation "mind = brain." They exhibit learning and goal-directed behavior with no central nervous system. This resurrects questions about the location and substrate of consciousness.

Medicine: If we can reprogram human cells (anthrobots) to solve problems, could we use this same principle to fix birth defects, regenerate organs, or reprogram cancer cells?

Open Questions

Can xenobots learn to reprogram themselves, or can only external bioelectric signals change their goals?
What is the upper limit of their cognitive lightcone — how complex can their goals become?
Do they have preferences? Desires? Any form of subjective experience?
How do we ethically treat organisms we create but don't fully understand?

Michael Levin | Cognitive Lightcone | Bioelectric Pattern Language | Basal Cognition

Data from Philosophy Everyday interview, June 2026. Levin emphasizes that cognitive claims about xenobots are still being tested and have not yet been demonstrated beyond learning/memory.