Who Is the Endosymbiont?
In a fun and short speculative PNAS article titled “Could humans and AI become a new evolutionary individual?” (PNAS 122 (2025) No. 37 e2509122122), Paul Rainey (Max Planck Institute for Evolutionary Biology in Plön, Germany) and Michael Hochberg (University of Montpellier, France) ask whether we could be witnessing the birth of a new form of life. The authors take an evolutionary biology perspective, using major evolutionary transitions (METs) as a core framework.
One of these METs is the formation of eukaryotic cells, which likely involved endosymbiosis: a host archaeal cell engulfed a bacterium, which instead of being digested, became a permanent resident. The relationship was mutualistic, meaning both prokaryotic organisms benefited (it may not have started that way). The endosymbiont (the bacterium) evolved into the mitochondrion, the organelle powerhouse of the eukaryotic cell that produces the energy-carrying molecule ATP. I will now diverge from the article to explore this particular metaphor.
I’ll skip over how the nucleus appeared over time, but the key point here is that the acquisition of mitochondria provided the host with a massive new source of energy. This metabolic boost was a key driver in the evolution of eukaryotic complexity, enabling the development of larger cells, membrane-bound organelles, and the ability to consume other cells. This extra power benefited the spread of the combined organism’s genetic material.
Bacteria and Archea represent two distinct domains of life. Similarly, AI may be a human artifact, but AI and humans represent two distinct domains of “stuff” (not life, perhaps intelligence, or something else entirely). If we were to merge with AI, my question is: who is the endosymbiont in this scenario?
Will AI become a powerful source of power to its host, or will the host evolve into a meat-based embodiment for AI?
Humans may evolve, in a biological sense, circuits and chemical networks to form a complete dependence on AI because it will be required for survival. For example, two individuals may not have the same ability to interface with neural implants, leaving one unable to perform tasks at the same level as the more “integrable” human. As for AI, it can’t evolve biologically, but it can evolve by means of technological improvement, either human-driven or self-driven. So both humans and AIs can “propagate” versions of themselves using one another as enablers. But who is in control? Or is the notion of "who"/self getting blurred?
2 additional references of interest
By Paul Rainey link
By @david Krakauer link