Opinion: How to Define Life

Read The Scientist’s special issue on artificial intelligence.

The endeavor to define biological life is more than an academic exercise. One could argue that definitions only tell us about the meanings of words in our language, as opposed to telling us about the nature of the world. With respect to defining living personhood, there are of course additional legal as well as moral implications that must be considered and are beyond the scope of this article. Nevertheless, our moral imperatives in large measure depend on how we define life.

Historically, characterizing a precise definition of life is generally based on unique characteristics of all known living organisms. For example, according to current notions of life, living organisms must: a) have a biological genetic set of instructions (ours are found in DNA and RNA) that encodes and regulates its functional properties; b) be composed of individual units or cells surrounded by a plasma membrane and that contain and metabolize biological entities, such as nucleic acids, proteins, carbohydrates, and lipids; c) be capable of adaptation or mutation to alter their phenotypes and respond to environmental factors that can alter their genotypes or phenotypes; d) undergo metabolic homeostasis—regulated growth that responds to both internal and external environments in response to external environmental conditions; and e) reproduce to create new organisms and have finite lifetimes. Organisms created from synthetic nucleotides, and AI-based robots, may not fit all these criteria. 

In our definition, organisms that utilize synthetic DNA nucleotides may meet our criteria as living.

For natural selection to have generated such a diversity of living things on earth, time and the mortality of every individual organism to assure the future survival of species are both required. We propose a simple but challenging definition of life as the property of an organism that possesses any genetic code that allows for reproduction, natural selection, and individual mortality. 

This definition underscores the need to protect the unknowability of future life forms. The randomness of pre-adaptive mutation, the surviving genomes, and the phenotypes of our species in the future cannot be known with certainty, nor can we know what species will replace us, if any. 

Our definition is more expansive than NASA’s, which describes life as “a self-sustaining chemical system capable of Darwinian evolution.” AI robots would not fit into our definition because human beings can control all aspects of computer functions. There is no uncertainty, nor unknowability, with AI robots. AI-based human robots can be programed to replicate themselves and even can be programed to terminate. However, robots do not sense “mutations” or engage in any natural selection process and, therefore, would not meet our criteria as “living.” 

In our definition, organisms that utilize synthetic DNA nucleotides may meet our criteria as living. However, it is important to recognize that while developing synthetic “life forms” constitutes technologically exciting endeavors, the danger that they may destroy all existing life forms on Earth through the unpredictability of natural selection may push such projects across an ethical boundary. 

We argue that as living organisms, and, in particular, as mortal creatures who are aware of our own mortality and of our capacity and obligation to distinguish right from wrong, we must recognize this boundary between living and inanimate. We believe the definition we have presented creates a clean boundary around all living things that allows us to assess the living status of synthetic organisms and AI robots. 

John D. Loike, a professor of biology at Touro College and University Systems, writes a regular column on bioethics for The Scientist. Robert Pollack is a professor of biological sciences at Columbia University.