Did you know 67% of Americans think ChatGPT might be conscious? University of Waterloo survey.
What IS intelligence, anyway?
In 2024 it’s THE front-burner question.
Dr. Michael Levin at Tufts university is demolishing conventional notions of intelligence.
Re-framing AI, genetics, robotics, cancer, longevity. Identity and memory. I’ve interviewed him several times. He’s 20 years ahead of everyone.
Now the Institute of Art and Ideas (iai.tv) has thrown the two of us into the intellectual thunderdome.
Dr. Levin says our notions of intelligence are waaaay too narrow:
https://iai.tv/articles/patterns-are-alive-and-we-are-living-patterns-auid-2919?_auid=2020
I argued…
Yes, our ideas ARE too narrow…. but before we start looking for E.T. & other exotic life forms… first let’s notice the intelligences literally right under our nose:
https://iai.tv/articles/patterns-cant-explain-lifes-complexity-auid-2922?_auid=2020
Hold onto your neurons!
Download The First 3 Chapters of Evolution 2.0 For Free, Here – https://evo2.org/evolution/Where Did Life And The Genetic Code Come From? Can The Answer Build Superior AI? The #1 Mystery In Science Now Has A $10 Million Prize. Learn More About It, Here – https://www.herox.com/evolution2.0
If Levin’s work is indeed correct, what are its implications to the current Modern Synthesis paradigm?
There are several parts of his work we can discuss.
The first part is his work in general ie https://evo2.org/picasso-tadpoles/ or http://www.evo2.org/anthrobots which demolishes the Modern Synthesis. Xenobots and anthrobots have the same DNA as the organisms they originate from but have totally different behaviors, and respond in brand new ways to situations that no ancestor could have likely encountered, or pre-programmed them to do.
As for the issues we debate in this IAI.tv series, his sorting algorithm experiments suggest that there are behaviors in “very simple” systems that we are largely clueless about and that our categories for “what X is” are just ways of describing them, but may be a long way from what X REALLY is.
It would at minimum suggest that many more mechanisms may be driving evolution than the Modern Synthesis ever imagined.
Part B of the answer:
Michael was a contributor for the book “Evolution on Purpose” by MIT press.
https://direct.mit.edu/books/oa-edited-volume/5634/Evolution-On-Purpose-Teleonomy-in-Living-Systems
And Michael Levin has an AI GPT bot on his website.
https://adorosario.github.io/LevinBot/index.html
I asked it: what is Michael’s opinion of the Neo-Darwinian Modern Synthesis?
LevinBot answer:
and molecular biology (Sample-Cognition-Based Biology – Chapter 1 – Darwin Reappraised, Page 16).
Denis Noble’s Critique:
Denis Noble’s critique of the Modern Synthesis highlights four now-untenable illusions:
a) Natural selection is a metaphor, not an active evolutionary force.
b) The Weismann Barrier has been superseded by epigenetics.
c) Darwin’s debunked theory of gemmules is now confirmed by the discovery of ubiquitous extracellular vesicles.
d) Crick’s Central Dogma has been shown to be invalid (Sample-Cognition-Based Biology – Chapter 1 – Darwin Reappraised, Page 17).
Alternative Frameworks:
Michael Levin’s work suggests a need for alternative frameworks that generate novel discoveries and empirical progress. He extends mind to unconventional aspects of the biosphere and merges smoothly into “machines” on the left side of the spectrum, which some organicists and molecular biologists find contentious (Frequently Asked Questions – Latest Version).
Evolutionary Paradigms:
References to works like “Beyond Neo-Darwinism: An Introduction to the New Evolutionary Paradigm” by Ho and Saunders, and “Epigenetic Inheritance and Evolution: The Lamarckian Dimension” by Jablonka and Lamb, indicate a shift towards new evolutionary paradigms (23e359eec026498668ea9de28bd80f52-html.html).
In summary, Michael Levin’s opinion reflects a critical stance towards the Neo-Darwinian Modern Synthesis, advocating for a more integrated and updated approach to understanding evolution that incorporates recent advances in various biological disciplines.
All the research I’ve read indicates that inheritance will always resolve at the unit of the genes.
Also, could you bring invite Alfonso Martinez Arias for an interview? He’s also a critic of the neo-Darwinian framework?
Josh,
We know for example that body plans are not encoded in the genes. Body plans are a pretty big part of development and are controlled by bioelectric fields. So genes are not the whole story.
If you can provide a good way to learn about Martinez’ work we can take a look.
You can read his book The Master Builder: How the New Science of the Cell Is Rewriting the Story of Life
You can also see his interview on the Theories of Everything youtube channel https://youtu.be/wyZeli50nXg?si=t3H8PW6bkaWWtPCT
Watched some of the interview. Two thoughts:
1) He gives Dawkins too much credit. “Selfish Gene” is only literally true for about 5% of the characteristics that we think of as genetic. For the vast majority of evolution and biology it’s harmfully oversimplified and pretends the last 50 years of molecular biology never happened.
2) He’s not giving Michael Levin nearly enough credit. Xenobots are capable of a form of biological reproduction never seen before. They exhibit behaviors such as self-motility, target finding, and information retention without any genomic alterations.
Part C:
ME: tell me more about Michael Levin’s views of neo-Darwinism.
LEVINBOT: Michael Levin’s views on Neo-Darwinism can be inferred from the following points in the knowledge base:
Dominant Paradigm:
The dominant paradigm in biology today is Neo-Darwinism (23e359eec026498668ea9de28bd80f52-html.html).
Evolutionary Theory and Intelligence:
Michael Levin has co-authored essays on how evolution ‘hacked’ its way to intelligence from the bottom up, indicating a nuanced view of evolutionary processes beyond traditional Neo-Darwinism (Michael’s essay with Rafael Yuste).
Integration of Developmental Biology and Evolution:
Levin emphasizes the importance of taking developmental biology and evolution seriously, suggesting a continuity of various beings over time scales and the ability to replace biologically evolved components with engineered materials (title: “Diverse Intelligence” – a talk by Michael Levin – YouTube).
Critique of Understanding:
Levin’s research suggests that our understanding of biology is closer to 1% understanding and 99% is “Dark Matter,” indicating that much of the biological processes and evolutionary mechanisms are still not fully understood (title: Picasso Tadpoles: Michael Levin on the ‘Dark Matter’ of Biology – Evolution 2.0 | Acast).
Cross-Disciplinary Approach:
Levin’s work spans multiple disciplines, including computer science, embryo development, cancer and tumor research, limb regeneration, evolutionary theory, and neural networks, which suggests a broader and more integrated approach to understanding evolution (metadata: Picasso Tadpoles: Michael Levin on the ‘Dark Matter’ of Biology – Evolution 2.0 | Acast).
In summary, Michael Levin’s views on Neo-Darwinism reflect a critical stance and an advocacy for a more integrated and updated approach to understanding evolution, incorporating recent advances in various biological disciplines and recognizing the complexity and continuity of life forms.
My reading of Michael Levin’s work is different from him “demolishing conventional notions of intelligence.”
First, it is useful to separate his peer-reviewed papers from essays and talks for the general audience. When he presents complex ideas to the wide audience it’s not the same thing as his primary scientific literature. Second, what he does in his peer-reviewed work is (a) define a concept (b) make a claim based on these definitions (c) support his claim experimentally. (Without this approach, the discussions have a risk of falling into semantics where the two debaters have entirely different definitions, or they have pliable definitions and no experimental data in support of them.)
Example.
In some of his work he choses the definition of intelligence by William James – “An ability to reach the same goal by different means.” Then he claims that xenobots have a degree of intelligence specifically in that sense. Then he supports that claim experimentally.
My reading of his primary literature is that he is careful with extending this argument onto (a) consciousness (b) life. One of the important pillars of his TAME framework is commitment to graduality – the absence of the blight lines separating “truly intelligent” from “truly not intelligent.” He shows a continuum of intelligence (while using a specific definition), e.g. the size of the light cone in TAME, or a degree of persuadability.
A notion that he “demolished conventional notions of intelligence.” suggests that there is unity in “conventional notions of intelligence,” and that “conventional intelligence” is a category defined by bright lines.
I don’t think we have universal agreement in the literature – for example the FSIQ suggests that intelligence can be represented by one number, while Gardner challenges that notion taking about multiple intelligences – hunter’s and dancer’s intelligence are not measured by FSIQ. The state of the literature is diverse, not universally agreed upon.
One of the principles of Levin’s TAME framework is a commitment to empirical research – experiments. This is what leads to making more accurate predictions and control. Second one is commitment to graduality, and there are others. He follows these principles in his scientific work.
Alexey,
Mike’s approach to the continuum from non-intelligence to intelligence is decidedly “analog”. As I show in the article, I see points where large jumps occur. I don’t think those jumps are actually digital, but the definitions of cognition / code / chemicals (in my terminology) can be made so the distinctions are quite sharp.
One of my current conversations with Mike is whether the unexpected attractors exhibited by his sorting algorithms are basically “theorems” that deterministically arise from the logic, or if there is something more subtle going on here.
If the “consensus view” (or at least the view most often found in textbooks) is that evolution is blind, and only humans or maybe higher animals are evolving consciously… then Mike’s work is demolishing conventional views of intelligence. It’s only been in the last 10 years that James Shapiro was (to my knowledge) apply the term cognition in the literature to all life, all the way down to bacteria. I’m very happy that a new movement is afoot in biology exemplified by the Third Way of Evolution group and others who espouse cognition based evolution views, and that movement is gaining traction.
It could be that we understand Michael differently, or interpret differently. Because of that, I’ll be specific. Here is one of his foundational papers, outlining his principles
Levin, M. (2022). Technological approach to mind everywhere: an experimentally-grounded framework for understanding diverse bodies and minds. Frontiers in systems neuroscience, 16, 768201.
Citation from that paper, section The Core Tenets of TAME, Table 1, first bullet:
“Continuum of cognitive capacities—no binary categories, no bright line separating true cognition from “just physics,” as is clear from evolutionary process and ability to bioengineer chimeras between any two “natural kinds.”
He confirmed that very recently here – specifically as applied to cognition, intelligence, and agency:
https://www.youtube.com/watch?v=O0wO8FIdxzQ&t=170s
Quote: “I don’t like the binary framing of is it or isn’t it an agent. I think there is a spectrum and what we’re really talking about is where on that spectrum is the right bag of tools to use…”
__
Mathematically, both discrete and continuous variables exist. On intelligence specifically, I agree with the citations above – it’s a spectrum. Even in the traditional FSIQ testing with it’s limiting single-factor intelligence model it is still a spectrum and not a bright line.
My approach to the question is more “digital” than his. I do agree there is a continuum. I don’t think it’s a smooth continuum. I think it has sharp jumps, as I describe in the recent agential topics video I referenced in an earlier comment in this thread. I’m a huge fan of Mike’s work.
Here is new blog post with a presentation I did to Charles Munford’s “Agential Topics” group: https://evo2.org/codes-and-agency-in-biology/