A new bold paper in the Elsevier journal Progress in Biophysics and Molecular Biology states, “Darwinian evolution has become dogma; AI can rescue what is salvageable.” Authors Olen Brown and David Hullender note, “The publication of scientific disagreements with elements of Darwinian evolution including its modern variants is increasing” and they cite various examples from the literature:

The publication of scientific disagreements with elements of Darwinian evolution including its modern variants are increasing. The view that “Evolution is both a fact and … the most important theory in biology. Evolution explains every situation” (Russo and André, 2019) is being challenged. Wray and Hoekstra in the Comment Does evolutionary theory need a rethink” published in Nature (Wray and Hoekstra, 2014) reported that Kevin Leland and seven colleagues responded “Yes, urgently”, while Gregory Wray and five colleagues responded: “No, all is well”. Typical of balanced questioning is Dennis Noble who wrote “Something has gone deeply wrong in biology” (Noble, 2021).

The paper thus argues “that the theory of biological evolution, including its modern variants, suffers from several logical deficits, is absurdly improbable mathematically, and also biologically mechanism-deficient.” However, as the title suggests, the authors believe that “Darwinian evolution has become dogma” and some new method is needed to move past non-objective adherence to evolutionary models. They believe this method is artificial intelligence (AI).

Use AI to Challenge Evolution?

The authors propose that “the new approach of AI … is required to move forward scientifically.” They note that AI provides “powerful analytical tools” that can be used for evaluating the merits of scientific theories and ask, “[C]ould a complex computer be programmed to evaluate the theory (many say the fact) of biological evolution? Or perhaps test particular postulates essential to the theory?” They believe AI is well-suited for this task, since it has already been used to “rediscover fundamental equations” in fields such as physics and chemistry, and has been highly successful at playing games and solving puzzles. They believe this makes AI applicable to studying evolution: 

Evolution, also, is a puzzle. It necessarily involves the absurdly improbable self-assembly of many complex biological machines using simpler parts (Brown and Hullender, 2023). Gartner et al. (2020) stated, “self-assembly of a large biological molecule from small building blocks is like finishing a puzzle of magnetic pieces by shaking the box.” AI works well for chess; we propose that it would work well for assessing ideas about biological evolution, especially the problem of self-assembly. Initially, it should be applied to testing the limits of the usefulness of ‘survival of the fittest’ for microevolution and the highly-improbable self-assembly required for macroevolution.

If AI were used to test and evaluate evolution, would people trust its results? 

Some May Not Like This Approach 

They believe that using AI to test evolution will lead to a problem: evolution will be challenged, and some may not want AI applied in this manner. They write that this should not matter because dogmas should never prevent scientific questions from being asked:

A perceived potential difficulty, which might, however, produce the first positive result of applying AI, is that focused discussion of the tenets, assumptions, and established facts of biological evolution would result. Consensus without criticism is not healthy for science. Biologists can learn from the field of Physics which is open to recognizing new ideas. It has shown itself receptive to change with concepts about gravity evolving from Newton to Einstein and the current interpretations, an example. The consequences of challenging the overall theory and subcomponents of biological evolution are monumentally significant for progress in this field and has ramifications for all of science including the freedom to challenge dogmas. “The scientist is free, and must be free to ask any question, to doubt any assertion, to seek for any evidence, to correct any errors” —J. Robert Oppenheimer.

[…]

Problems with the Darwinian theory of evolution, including its modern variants, and origin of life theories are significant and require new approaches and a willingness of scientists to look for bold solutions. The application of AI has great promise both for assessing the problems and weaknesses and for providing innovative and significant solutions of great significance for science and humankind. AI can be the pathway to correcting the problems in evolutionary theory, but the human brain must create that pathway. There must be no barriers to freedom of inquiry. There is no place for dogma in science.

Not Turtles All the Way Down?

These are nice statements of the importance of freedom of scientific inquiry. But what would AI find if it were let loose to critically investigate biological evolution? Their main argument is that we need to apply it to such a task. But they predict that if AI were applied to questions of biological origins, it would find serious flaws in evolutionary models: “We conclude that AI has this potential and encourage its application immediately for evaluating theories of biological evolution. It seems remote that AI would conclude that it is ‘turtles all the way down’.”

Surely any conclusion from AI about theories of biological evolution would be highly controversial — people would either accept and tout them or criticize the AI model that was used as flawed and inadequate. But it would be an interesting project to undertake nonetheless.