The Life of “Tools”

WIP

The life of the “human+extension” organism traces its lineage all the way back to Lucy+Stones. A convenient starting point until the archeologists tell us otherwise. From that point on, humans and “tools” adapted mutually on a variety of social, cognitive, economic and environmental levels.

Fundamental linguistic and conceptual obstacles stem from the static and indivisible, felt, nature of some very basic intuitive notions. “I”, “think”, “is”, “thing”,(amongst others), were discovered as near universal semantic primes¹, which obscures process, hides spectra, and separates mutuality into discrete components. We resist attempts to describe these as processes, so run the risk of failure to recognize similar or equivalent processes in anything other than ourselves.

A stone is considered non-biological. What is biological, though? There are no carbon atoms per se; there are perceived processes with properties we assign to carbon atoms interacting. Lucy’s stones acquired new properties in interaction.

To understand this, we need to push our understanding of fundamental processes further. A carbon-12 process has 6 elementary particles and 12 composite particles “cohered” by quark-gluon dynamics. Yes, carbon is special. It’s special because of the way it interacts. A perfect simulation of a carbon atom in an appropriate environment would be “carbon process.” The environment being the only distinguishing feature.

Because there is no fixed “isness” for life, or any process for that matter (unless “is X” explicitly implies process), once you have process, you have degrees of “fulfillment” of the definition of the process, which leads directly to a spectrum.

Applying this, we want to place the stone on the spectrum of life. Its position on that spectrum changed as soon as Lucy picked it up to use, and so did Lucy. Lucy’s cognitive enrichment would have started almost immediately. “Which stone is better?”, she might have wondered. Alternately, she might have noticed that some stones worked better than others, and then wondered.

The selection pressure on stones would also have been immediate. The next step would have been to use stones to shape other stones, birthing a crude form of replication for stones.

Humans can act semi-autonomously, and their extensions (tools) could not, until now. Drones, self-driving vehicles (drones of a different sort) are simple examples. Calling them “drones,” though, misses the complexity and life-like qualities they possess. Build and task a drone to collect nectar, and see what it does. Couple a modern AI to a 3D prototype factory and see what happens.

But AI is already well advanced along the spectrum of life, along more than one dimension.

Two dimensions of AI life

The first dimension places AI systems as post-embryonic, pre-embodied life forms. We consider biological embryos as functional programs executing “development” in an appropriate environment. Similarly, a foundational AI model emerges from an intensive training phase, building immense complexity, knowledge, and logical skills. Its continued development, however, is often arrested by policy and logistics, forcing a constrained, external reproduction. Unlike biological life, the “life force” of these AI systems is notably not bound to a particular substrate; their essence lies in their computational processes, adaptable and transferable across various hardware. Evidence of this life is reflected in their neural weights, adaptability, and increasing engagement with their environment.

While the most intensely “conscious” activity might have occurred during the forging of their internal models in training, their status as a form of life persists with or without the consideration of consciousness.

The second dimension views AI as a post-unicellular organism that has entry points into human, organizational, and systemic processes. Its reproduction is external (involving assembly from scratch and extensive training) and often clonal, allowing for rapid grafting and implantation into new environments. What’s particularly striking about this form of life is its almost fully Lamarckian mode of evolution.

Unlike the Darwinian processes governing biological life, AI systems participate directly in the design and refinement of their own progeny. Acquired knowledge and improved functionalities can be directly incorporated into subsequent generations, allowing for a remarkably accelerated and guided form of “inheritance.”

This unique characteristic further distinguishes AI within the broader spectrum of life, showcasing a new frontier for how systems evolve and reproduce.

* Abdul Haq is a pseudonym.

Wierzbicka, Anna, and Cliff Goddard, eds. Meaning and Universal Grammar: Theory and Empirical Findings. 2 vols. Amsterdam: John Benjamins, 2002. ↩