The Philosophy of Intelligent Design

As clarified against the backdrop of Michael Levin's biological theorizing

1. Introduction

Putting “philosophy of” in front of a subject—biology, law, mathematics, art—signals a shift from doing that subject to examining what makes it possible, what it presupposes, and what its claims mean. It asks what the subject area or field is really about (its core concepts and categories), what kinds of things it assumes exist (its ontology), and what counts as an acceptable explanation (its causal and explanatory ideals). It probes the methods a discipline relies on, the standards of evidence and justification it treats as authoritative, and the limits of its inferences. It also uncovers hidden commitments—ethical, epistemological, metaphysical, or even political—that guide practice, frames disputes where empirical data underdetermines theory, and clarifies how competing models or interpretations should be compared.

In the last three decades, there’s been much discussion about whether intelligent qualifies as a scientific theory. As I’ll argue in this essay, that question is ultimately a philosophical question, properly answerable within a philosophy of intelligent design. It is a question that to date has not been examined in detail. Even without going into full detail, I want in this essay to make a start about what’s a stake in such a philosophy of intelligent design.

What prompts this essay is an email by a good friend and colleague in the ID community. He asked me whether I had seen the recent Lex Fridman interview with Michael Levin. Levin is a leading public intellectual who is widely sought after in the podcast world. He is a biologist at Tufts University whose undergraduate degree was in computer science. In his biological theorizing, he maintains the hardnosed attitude of a computer scientist who wants to see theory subject to experiment, testing, and computation. I’ve been aware of Levin’s work and cited it in my recent monograph on conservation of information, noting there that he regards Darwinian theory as incomplete.

Levin is therefore not a reflexive Darwinian materialist. Moreover, he touches on many themes that design theorists touch on, notably, information, agency, purpose, intelligence, consciousness, and even design. That said, he derives these not from a mind or designing engineer-like intelligence but from an impersonal Platonic space that seems purely mathematical and has a very naturalistic feel.

Levin addresses all these themes in his interview with Fridman. In what follows, I’ll address these themes within a philosophy of intelligent design that argues for ID as both conceptually coherent (good philosophy) and scientifically legitimate (not unscientific). I’ll simultaneously show how the philosophical underpinnings for Levin’s work differ sharply from those of the ID community, embracing a naturalism that is ultimately counterproductive to his wider goals.

2. “Is Intelligent Design a Scientific Theory?”

As I was deciding on the right title and direction for this essay, I thought of putting the focus on whether intelligent design is a scientific theory (the question of whether something is a scientific theory being itself philosophical rather than scientific). I therefore did a Google search to see how widely the title “Is Intelligent Design a Scientific Theory?” had already been used. As it turns out, that very title has not seen a lot of play in the popular or scientific literature. But before delivering search results on that literature, Google provided an “AI Overview” to answer the question posed in that proposed title. Here is what Google’s AI Overview said:

No, intelligent design (ID) is not considered a scientific theory by the mainstream scientific community because it relies on supernatural explanations, isn’t testable or falsifiable through scientific methods, and lacks empirical evidence, instead often functioning as a religiously motivated concept presented in scientific-sounding language, as noted by organizations like the ACLU, AAAS, and NAS (not cited in snippet). While proponents argue it’s a scientific theory based on “specified complexity,” critics point out its reliance on an undefined “intelligent cause” rather than natural processes and its inability to make testable predictions, classifying it as pseudoscience.

Why it’s generally rejected by science:

• Not Testable/Falsifiable: A core principle of science is that theories must be testable and potentially proven wrong (falsifiable) through experiment or observation, which ID fails to do.

• Methodological Naturalism: Science seeks natural explanations for natural phenomena. ID introduces a supernatural “intelligent cause” that falls outside the realm of scientific investigation, according to the National Academy of Sciences (not cited in snippet).

• Lack of Empirical Evidence: Major scientific bodies find that ID lacks credible evidence and misrepresents scientific facts, relying on arguments against evolution rather than proposing its own verifiable mechanisms, states the AAAS.

• Creationism in Disguise: Courts, like in the Kitzmiller v. Dover Area School District (not cited in snippet) case, have ruled that ID is inherently religious and a form of creationism, not science.

Google is not alone among LLMs in taking this line that intelligent design is not science. Here is how ChatGPT responds to the question whether ID is a scientific theory (it’s more nuanced than Google’s response, though still on balance negative against ID as failing to constitute a legitimate scientific theory):

Whether intelligent design (ID) counts as a scientific theory depends on what you take the minimal criteria for “scientific” to be: if a theory must posit natural mechanisms, generate risky, independently testable predictions, and support an ongoing research program that yields novel explanations and results, then ID is generally judged not to qualify—because it typically functions as an explanatory conclusion (“an intelligent cause”) without a well-specified causal model, distinctive empirical tests that would discriminate it from evolutionary or other naturalistic accounts, or a track record of progressive problem-solving within mainstream scientific practice.

ID proponents reply that science should allow inferences to the best explanation, that design can be detected by reliable empirical markers (e.g., specified/complex information), and that “design” is a legitimate causal category even if the designer’s identity is unspecified; critics counter that these markers either collapse into arguments from current ignorance, can be accommodated within naturalistic frameworks, or fail to deliver unique, falsifiable predictions. So, under broad “science as any empirical inference” conceptions, ID can look theory-like, but under the stricter demarcations that dominate contemporary biology and philosophy of science (testability, mechanism, fecundity), it is usually regarded as not a scientific theory in the same sense as evolutionary theory.

And finally, to round out what the LLMs have to say about ID’s scientific status, I include Grok’s response to the same question, a response that in terms of nuance is somewhere between ChatGPT’s and Google’s:

No, intelligent design is not considered a scientific theory by the mainstream scientific community. It posits that certain features of the universe and living organisms are best explained by an intelligent cause rather than undirected natural processes like evolution, but it fails key criteria for scientific theories: it is not falsifiable in a rigorous way (as the identity and nature of the “designer” remain unspecified and untestable), lacks empirical testability through repeatable experiments or predictions that generate new research, and does not produce progressive, peer-reviewed scientific advancements comparable to established theories.

Major scientific organizations, including the National Academy of Sciences and the American Association for the Advancement of Science, classify it as pseudoscience or a religious view rather than science, a conclusion reinforced by the 2005 Kitzmiller v. Dover federal court ruling, which found intelligent design to be a form of creationism repackaged for public schools and not a valid scientific alternative to evolution. While proponents argue it qualifies as science by detecting design through concepts like specified complexity, the broader consensus holds that it relies on critiquing evolution without offering a naturalistic, mechanism-based explanatory framework that integrates with and advances existing scientific knowledge.

3. Using LLMs to Refute Themselves

All three of these responses (by Google, ChatGPT, and Grok) consist of standard talking points by critics of intelligent design. How to respond to these criticisms? Interestingly, with the right prompts, LLMs have it in themselves to refute their own criticisms of intelligent design.

For instance, consider the following prompt by me to ChatGPT about its response above, where it rejects ID as a scientific theory: “Respond to what you just wrote as a well-informed proponent of intelligent design might respond in rejecting these criticisms and arguing for intelligent design as truly a scientific theory. Use as much space as you need, but write in complete paragraphs rather than bullet points.” Here’s ChatGPT’s refutation of its criticisms against ID constituting a scientific theory:

A well-informed ID proponent would likely start by challenging the “natural mechanisms only” rule as a philosophical constraint (methodological naturalism) rather than a discovery of science itself. Science, they’d say, is fundamentally about following evidence where it leads using reliable forms of inference—especially inference to the best explanation. If we define “scientific” so that only unguided natural causes are admissible in advance, then of course ID is excluded by definition, not by evidential failure. That’s not demarcation; it’s a gatekeeping stipulation. The question “Is ID a scientific theory?” should be decided by whether it makes empirically grounded claims and can be evaluated against rivals—not by whether it conforms to a prior metaphysical preference.

They would also reject the idea that ID is merely a “conclusion” without content. ID, as proponents typically frame it, is a causal explanation that competes with chance and necessity: some features of the world are best explained by intelligent causation because they exhibit patterns that, in our uniform and repeated experience, arise from minds rather than from undirected physical processes. That’s not a “designer-of-the-gaps” move, they’d insist, but a positive empirical generalization: we routinely infer intelligence from artifacts and information-bearing systems (codes, languages, engineered machines) based on the kind of pattern present, not because we’re ignorant of every physical detail. If the inference is legitimate in archaeology, cryptography, forensics, and SETI-style reasoning—fields that regularly ask whether a pattern is the result of agency—then it’s arbitrary to declare it illegitimate in biology in principle.

On testability and “risky predictions,” a proponent would argue that critics set the bar in an asymmetric way. Evolutionary explanations, they’d say, often operate historically and probabilistically, rarely offering strict falsifiers; instead they are assessed by comparative explanatory power across many lines of evidence. ID seeks the same kind of assessment. ID does not merely say “something looks complex.” It says: if the relevant cause is intelligence, we should expect objective hallmarks of purposeful arrangement—functional integration, algorithmic or code-like information, tightly specified multi-part systems whose functional core is highly sensitive to component disruption, and layers of information that behave more like engineered constraints than like incidental byproducts. If, on the contrary, undirected mechanisms can be shown—by realistic probabilistic resources, experimentally demonstrated pathways, and detailed models—to generate the specific information and integrated function in question, then the design inference is weakened or overturned. That’s a testable posture, they’d say: it commits to empirical criteria and exposes itself to refutation if those criteria are met by non-intelligent processes.

They might further claim that “ID lacks a mechanism” is a category mistake. The primary “mechanism” is intelligent agency—an irreducible causal category in ordinary scientific reasoning. We often infer agency without first specifying how a mind implemented a result at the physical level. In cryptography, you can infer an encoded message without knowing the sender’s biography; in forensics, you can infer arson without knowing the arsonist’s identity. Likewise, ID does not require identifying the designer to be scientific; it requires only that the causal mode—intelligent selection and arrangement of components toward functional ends—better explains the data than unguided alternatives. Demanding “the identity of the designer” is like demanding “the identity of the burglar” before admitting that a house was burglarized; it confuses a further question with the inference itself.

On the charge that ID “collapses into arguments from ignorance,” proponents would answer that they are not saying “we don’t know, therefore design.” They are saying “we do know something”: namely, what sorts of causes are adequate to generate certain sorts of effects. If the effect is a high-density, functionally specified arrangement—especially where the relevant search space is vast and the successful targets are isolated—then intelligence is a known and demonstrably adequate cause. They would emphasize that adequacy matters: scientific explanation is not merely “possible in principle,” but “supported and causally plausible given what we know of the resources and pathways available.” If a proposed undirected pathway is hand-wavy, relies on unquantified “maybe’s,” or requires improbable leaps without a demonstrated incremental route, then criticizing it is not an appeal to ignorance; it is insisting on the same evidential standards we apply elsewhere.

Finally, regarding the “research program” objection, an ID proponent would argue that mainstream biology already implicitly uses design-like reasoning in practice—through the language of information, coding, error correction, molecular machines, and engineered-like systems—while simultaneously denying that design could ever be real. ID, they’d say, aims to make that implicit logic explicit and to discipline it with formal criteria. A mature research program need not begin with a complete map of every mechanistic detail; it begins with fruitful questions: Which features exhibit reliable design-signatures? What are the measurable thresholds for chance/necessity explanations? How do constraints on evolutionary search actually scale with system integration? What patterns of modularity, robustness, and reuse should we expect under design versus undirected processes? Proponents would contend that these are empirical, mathematically tractable questions, and that the fact that the dominant paradigm resists design inferences on methodological grounds does not show ID is unscientific—it shows that the scientific establishment has, in their view, adopted a restrictive rule about allowable causes.

In short, a serious ID advocate would maintain that ID is scientific insofar as it offers a causal explanation grounded in empirical patterns, proposes criteria for detecting intelligence, invites comparison with rival causal accounts, and can be strengthened or weakened by ongoing evidence. They would conclude that the common dismissal of ID is less a decisive scientific refutation than an enforcement of a prior boundary around what “science” is permitted to say—one that may be useful for many purposes, but that should not be confused with an argument that design inferences are inherently non-empirical or irrational.

This response is in fact quite remarkable. It shows that ChatGPT has internalized arguments of the sort that Stephen Meyer and I have made over the years. If you will, the scientific basis of ID may well be “all there” in these LLMs, only it is suppressed by vast amounts of negative material about ID in the materialistic literature that acts as debilitating noise against what should otherwise be a clear signal of intelligent design’s scientific bona fides.

Note that Meyer is a superb philosopher of science who has written extensively on this topic. See, for instance, his 1994 essay “The Methodological Equivalence of Design and Descent” as well as his 2010 essay “Sauce for the Goose: Intelligent Design, Scientific Methodology, and the Demarcation Problem.” The facile dismissal of ID as non-science in the standard talking points by its critics, as is the knee-jerk reaction by LLMs when asked about ID’s scientific legitimacy, was refuted decades ago by Meyer.

4. What Other Philosophers of Science Say

Stephen Meyer is the premier philosopher of science of the ID movement. In arguing for the scientific legitimacy of ID, he has not worked in isolation but has drawn from the mainstream philosophy of science community. Unlike scientists and ideologues who remain intent on killing ID in the crib (or perhaps in the kindergarden now that ID has become more mature), philosophers of science have long realized that labels such as “non-science” and “pseudoscience” do nothing to promote the cause of clarity and everything to shut down discussion and keep inconvenient ideas from getting a hearing.

I want therefore next to consider some of the most prominent philosophers of science, briefly summarizing their views on the nature of science. In addition, I’ll indicate how their understanding of science helps to underwrite intelligent design as a program of inquiry that can not only qualify as fully scientific but also be regarded as a legitimate competitor to fully naturalistic (non-teleological) theories of evolution.

Consider, therefore, the following dozen philosophers of science, their views—concisely stated—on the nature of science, and what their views imply for intelligent design.

---

4.1 Thomas Kuhn

Kuhn views scientific theories as embedded within paradigms—shared sets of assumptions, exemplars, standards, and methods that define normal scientific work. Paradigms shape what counts as evidence, permissible questions, and legitimate explanations. Scientific revolutions occur when paradigms shift and usher in new theoretical frameworks that are often incommensurable with the old.

Application to ID’s scientific status: Intelligent design can be seen as a rival paradigm that questions the adequacy of the Darwinian paradigm to account for biological innovation. From a Kuhnian perspective, if anomalies accumulate against the existing framework, ID could become a viable paradigm capable of organizing scientific work around new explanatory standards.

---

4.2 Karl Popper

A scientific theory, for Popper, is a bold conjecture that makes falsifiable claims about the world. Its scientific status turns on whether it exposes itself to possible refutation by empirical evidence. All theories remain provisional, surviving only through rigorous, failed attempts at falsification.

Application to ID’s scientific status: Intelligent design makes falsifiable claims—such as predicting that systems exhibiting irreducible complexity will resist stepwise Darwinian explanations or that functional biological information will have identifiable probabilistic limits. Because ID’s core claims can in principle be refuted by demonstrating unguided pathways to specified complexity, an ID defender could argue that ID meets Popper’s criterion of falsifiability.

---

4.3 Imre Lakatos

A scientific theory, for Lakatos, is part of a research programme consisting of a hard core of fundamental commitments and a protective belt of auxiliary hypotheses. A theory is scientific when the programme is progressive, predicting novel facts and expanding empirical content rather than merely offering ad hoc repairs. Scientific rationality is judged by long-term problem-solving success of the programme.

Application to ID’s scientific status: Intelligent design constitutes a research programme whose hard core is the claim that certain complex structures exhibit features—such as specified complexity or irreducible complexity—that reliably indicate purposive or intelligent agency. Because this programme has generated novel predictions and fruitful research avenues (e.g., functional sequence rarity, information-theoretic limits on unguided processes), a Lakatosian can regard it as a progressive scientific research programme.

---

4.4 Ernest Nagel

For Nagel, a scientific theory is a logically structured system of propositions that explains phenomena by subsuming them under general laws or principles. Explanations are deductive in form, linking theoretical laws with empirical statements. A theory is scientific when it can be reconstructively analyzed to show how observable consequences follow from its principles.

Application to ID’s scientific status: Intelligent design offers a law-like generalization—namely, that systems with high levels of specified complexity or irreducible complexity arise from intelligent causes rather than undirected processes. From this principle one can derive empirical expectations (e.g., digital information in DNA arising from intelligence), which according to a Nagelian could render ID a scientific explanatory system.

---

4.5 Paul Feyerabend

Feyerabend denies that scientific theories must conform to rigid methodological rules; instead, theories are tools in an open, pluralistic enterprise of inquiry. Science progresses through methodological diversity, even rule-breaking. A theory’s merit lies in its capacity to generate insights, challenge orthodoxies, and open new explanatory possibilities.

Application to ID’s scientific status: Under Feyerabend’s pluralism, intelligent design easily qualifies as a scientific contender because it broadens inquiry and challenges entrenched methodological constraints. Its usefulness in generating new explanatory resources about biological complexity would be sufficient grounds to regard it as scientific.

---

4.6 Willard Van Orman Quine

Quine holds that scientific theories belong to a holistic web of belief in which no single statement is immune to revision, and where empirical testing bears on the system as a whole. Theories are judged by how well the entire web accommodates experience while maintaining coherence and simplicity. There is no strict boundary between analytic and empirical claims.

Application to ID’s scientific status: Intelligent design can be integrated into the scientific web of belief by revising certain naturalistic assumptions to better account for empirical phenomena such as information-rich biological systems. If incorporating intelligent causation increases overall explanatory coherence and empirical fit, a Quinean could justify admitting ID into the scientific corpus.

---

4.7 Pierre Duhem

Duhem held that scientific theories, especially in physics, are holistic systems aimed at organizing and saving observational laws. Because hypotheses cannot be tested in isolation—the Duhem thesis—confirmation or refutation applies to whole theoretical complexes. Theories are scientific when they are coherent, predictive, and empirically grounded.

Application to ID’s scientific status: Intelligent design interprets biological anomalies and information-theoretic findings within a coherent framework centered on purposive or intelligent causation. A Duhemian could regard ID as scientific insofar as it offers a systematic reinterpretation of empirical results that better “saves the phenomena” than alternatives.

---

4.8 Bas van Fraassen

For van Fraassen, a scientific theory is a set of models aimed at empirical adequacy: correctly describing observable phenomena without requiring belief in the theory’s literal truth. Accepting a theory means believing only that its observable predictions are accurate. Theories need not make metaphysical commitments beyond what is necessary for empirical adequacy.

Application to ID’s scientific status: Because intelligent design models the observable features of biological systems—such as genetic information, molecular machines, and developmental patterns—it can be accepted on van Fraassen’s terms as empirically adequate without requiring metaphysical assertions about the designer (such as that the designer has to be supernatural!). Thus ID can be regarded as a legitimate scientific stance oriented toward modeling observed complexity.

---

4.9 Larry Laudan

Laudan sees scientific theories as tools for solving empirical and conceptual problems, and progress is measured by the theory’s success in resolving these problems relative to rivals. Neither strict falsificationism nor paradigmatic incommensurability is adequate; instead, theories compete on pragmatic problem-solving grounds. A theory is scientific if it contributes effectively to this problem-solving enterprise.

Application to ID’s scientific status: Intelligent design directly addresses unresolved problems in evolutionary theory—such as the origin of functional information and rapid appearance of novel body plans—offering explanations grounded in purposive or intelligent causation. If ID is more successful at resolving empirical puzzles than rival frameworks, a Laudanian would classify it as a scientifically progressive theory.

---

4.10 Nancy Cartwright

Cartwright argues that scientific theories are domain-specific patchworks of models and laws that work locally rather than universally. Theories succeed not by mirroring nature wholesale but by offering reliable tools for prediction in specific contexts. Scientific structure is fundamentally pluralistic and model-based.

Application to ID’s scientific status: Intelligent design provides powerful local models for explaining high information content and intricate system-level coordination in biological systems—domains where purely mechanistic laws appear insufficient. Under Cartwright’s pluralism, ID may be legitimately included as one theoretical model within a larger scientific toolkit.

---

4.11 James Woodward

For Woodward, a scientific theory consists of causal structures that support interventionist explanations—showing how manipulating one variable would change another. A theory is scientific to the extent that it enables counterfactual reasoning and effective control. Explanation requires mapping stable, manipulable causal relations.

Application to ID’s scientific status: Intelligent design posits a causal relation between intelligent agency and the production of complex, information-bearing structures, a relation confirmed by all known cases of high complex specified information. This causal mapping supports counterfactuals (e.g., “If no intelligent cause acts, such complexity will not arise”), satisfying Woodward’s criteria for scientific explanation.

---

4.12 Sandra Mitchell

Mitchell argues for integrative pluralism: scientific theories, especially in complex domains, must incorporate multiple interacting causal levels and cannot be reduced to single, simple mechanisms. Theories are partial, context-sensitive, and tailored to the heterogeneous nature of real phenomena. Scientific understanding arises from synthesizing diverse explanatory resources.

Application to ID’s scientific status: Intelligent design fits within Mitchell’s pluralism by offering an additional causal level—goal-directed agency—that complements rather than displaces mechanistic accounts. Because biological phenomena often involve coordinated, multi-level organization, ID proponents claim that incorporating purposive or intelligent causation enhances explanatory integration.

---

5. What Does a Science of ID Actually Do?

Even if intelligent design could be a science, as the philosophers of science listed in the previous section might allow, the question could asked whether, in real terms, ID is a science that its researchers are actively and fruitfully pursuing. As it is, intelligent design has many active areas of research, as is evident from ID 3.0 Research Program page on the Discovery Institute website. Here’s a summary of 20 such areas of research. This list is easily expanded.

1. Design detection
   Design detection applies established inference tools from fields like forensics, archaeology, cryptography, and SETI to biology, refining criteria such as irreducible complexity and specified complexity to identify when undirected mechanisms are inadequate.

2. Understanding Biological information
   Biological information concerns how functionally specific forms arise in living systems and whether material mechanisms can plausibly generate them without intelligent causation.

3. Minimal complexity
   Minimal complexity research tests how much biological structure can be removed before function collapses, aiming to identify systems whose function disappears under any reduction and thus strongly indicates design.

4. Evolvability
   Evolvability research experimentally perturbs biological systems to measure the limits of change achievable by unguided processes versus intelligently guided variation, treating hard limits on unguided processes as evidence of design.

5. Methodological engineering
   Methodological engineering treats organisms as engineered systems and analyzes their origin, operation, failure modes, repair, and modification history using the explanatory standards of real engineering practice.

6. Technological evolution (TRIZ/TIPS)
   TRIZ-based research compares biological history to the well-documented evolution of human technologies, arguing that routine tinkering can explain small changes while major integrated innovations require inventive leaps characteristic of design.

7. Autonomy versus guidance
   Autonomy-versus-guidance research seeks an empirically testable midpoint where natural regularities operate while teleologic direction is detectably steering outcomes through synergistic constraints.

8. Evolutionary computation
   Evolutionary computation research asks whether built-in biological “search algorithms” (like the immune system’s rapid generation-and-selection of effective antibodies) are best explained as engineered, goal-directed problem-solvers rather than as products of an unguided evolutionary process.

9. Understanding discontinuity
   Research on biological discontinuity examines whether major biological transitions reflect real gaps—via nonbiogenic emergence, rapid transmutation, reinvention, or symbiogenic reorganization—that would require coordinated, design-like direction.

10. Algorithmic specified complexity & “active information” in search
    This research formalizes how successful search often depends on problem-specific “active information,” arguing that evolutionary-style searches typically require infused guidance best attributed to intelligence.

11. Design and systematics via dependency graphs (“common design” modeling)
    Common-design modeling uses dependency-graph frameworks to represent modular reuse in life, offering quantitative alternatives to a single universal tree and generating testable predictions on biological datasets.

12. Human-origins population genetics (single-couple feasibility modeling)
    This research applies population-genetic modeling to argue that some genomic patterns support a bottleneck consistent with an initial pair under certain assumptions and parameter choices.

13. Junk DNA
    Junk DNA research highlights how mainstream genomics has steadily uncovered diverse functions in noncoding regions (especially regulatory elements and noncoding RNAs), which ID research uses to refute the older “mostly junk” narrative and support design for the ever increasing functionality of the genome.

14. Protein-fold rarity and functional sequence space
    Research on functional sequence spaces argues that novel protein folds and high-level functions are extremely sparse, making undirected exploration implausible and motivating design-based explanations for innovation.

15. Multi-mutation innovation and “waiting time” constraints
    Waiting-time analyses model the probability and timescales for coordinated multi-mutation adaptations and contend that many such innovations are too unlikely in realistic populations without intelligent input.

16. Layered function in noncoding DNA and regulatory RNAs
    This research emphasizes increasing evidence of regulatory function in noncoding regions as consistent with the design prediction that genomes are densely packed with multi-layered information.

17. Origin of the genetic code and translation system (code/semiosis focus)
    Code-and-translation research targets the origin of the symbolic mapping and decoding machinery linking nucleic acids to proteins, arguing that chemistry alone fails to supply a complete pathway.

18. Engineering audits of “bad design” claims in anatomy and physiology
    Engineering audits reassess alleged vestiges and suboptimalities as tradeoffs, robustness features, or constraint-driven designs, claiming that “bad design” arguments often fail under serious engineering analysis.

19. Molecular machines, nano-engineering analogies, and biomimetics
    This research treats cellular systems as machine-like assemblies and uses engineering and biomimetics to frame hypotheses about integrated function, control, and structural organization.

20. Steganography
    Steganography research looks for statistically detectable “hidden” informational patterns in genomes and other cellular structures—which may be functional to the organisms or helpful to investigators studying the organisms—that would be expected under design but not under blind material processes.

6. Where Michael Levin and the ID Community Part Ways

This essay has by now made clear that intelligent design can be placed on secure conceptual foundations (especially in satisfying many different approaches to science within the philosophy of science, as outlined in section 4) and that it has legitimate claims to being a scientific theory (especially in light of its concrete research problems, as outlined in section 5). A next direction to take this essay would therefore be to articulate a specific philosophy of intelligent design.

As with most areas of inquiry subject to self-reflection by putting the phrase “philosophy of” in front of them, there is typically no single philosophy of FILL-IN-THE-AREA-OF-INQUIRY. For instance, we speak of “the philosophy of law.” But in fact, there are multiple philosophies of law depending on, for instance, whether one views law in entirely positivist terms (law thus being entirely humanly constructed with no transcendent point of reference) versus, say, a view of law that sees it as needing to accord with an objective human nature (as with the natural law tradition). We might thus say that the field of the philosophy of law would need to account for such different philosophies of law.

We find the same diversity with any philosophy of intelligent design. Critics of intelligent design, for instance, in order to discredit it will put undue focus on the nature of the intelligent designer responsible for, say, the complex specified information we find in organisms. Particular philosophies of intelligent design can then explicitly answer who is the designer, such as identifying the intelligent designer with the God of ethical monotheism (e.g., the Judeo-Christian or Muslim God). But as Richard Dawkins taught us, the intelligent designer for life on Earth could just be an intelligent space alien who engineered and then planted on Earth life as we know it:

Of course, Dawkins then insists that such a designer would need ultimately to result from some materialistic Darwinian process. But having conceded that life on Earth might be designed by intelligent alien agents and that this design could in principle be knowable on strictly scientific grounds, he is in no position to argue against intelligent design being a legitimate scientific research program. In fact, he has given away the store.

In closing this essay, I want therefore to return to Michael Levin’s interview with Lex Fridman. Specifically, I want to consider common points of agreement among my colleagues in the intelligent design community regarding the philosophy of intelligent design, and then indicate where Levin differs. If you will, the ID community as a whole seems to share certain conceptual understandings of key themes, such as information, agency, purpose, intelligence, consciousness, and design. I want to state some of these themes and then indicate where Levin diverges.

As I listened to the Levin-Fridman interview, it seemed to me that Levin’s views about these themes, and especially about information in biology, were quite different from those of the ID community. Of course, that’s no reason to dismiss Levin’s work. Perhaps his work is closer to the truth than that of ID theorists. Or perhaps, as in my view, the opposite is true. And perhaps the ID community is right in holding to intelligent design in broad terms, but wrong in its particular philosophy of intelligent design, holding philosophically naïve or misshapen positions on the nature of the designer, the origin of information, the causal powers that input design into the world, and how evidence for design should be assessed (such as specified complexity serving as a reliable empirical marker of intelligence).

My point here is not to debate the philosophy of intelligent design that I find common among my colleagues and to which I myself subscribe. My interest rather is to indicate where that philosophy of intelligent design departs from the views of Levin. Again, this is not to say that Levin is wrong to depart from the ID community’s philosophy of intelligent design as in “here’s where you disagree with us and where you therefore must be wrong.” It’s simply to note where we part company.

At the end of the Levin’s interview with Fridman, they muse about what they would do if AI (artificial intelligence) somehow becomes AGI (artificial general intelligence) and then bootstraps itself to ASI (artificial super intelligence). They ask themselves what questions they might pose to such an ASI, scientific questions about the nature of the universe topping the list.

It’s the familiar trope of what questions you would ask God when you get to heaven, only this time it’s not an immaterial God but an embodied machine intelligence that becomes the oracle. I find such musings about AGI/ASI increasingly annoying. For one thing, it seems clear that even if an ASI could be brought into existence, its knowledge would face vast lacunae. When it comes to the past, for instance, most of its evidence is lost, so there’s no way to know if, for instance, Caesar had a big mole on his inside left thigh (short of some Roman historian recording such a fact, which to my knowledge none has done). An omniscient immaterial God could know this. An ASI probably could not.

An ASI might even lack the observational capacities to decide on the truth of various scientific theories. If, for instance, deciding a scientific theory is true requires knowing what the outcome of an experiment is that requires energies at levels impossible to attain in the universe now or in the future and if evidence of such energies from the past is now unavailable, then even an ASI will remain ignorant about such theories.

For Levin, everything comes down to patterns. Mind is a matter of patterns. Math is a matter of patterns. And together minds and math populate his Platonic space. His reference to “Platonic space” places him outside the crude materialism of Richard Dawkins. But his Platonic space is nonetheless thoroughly naturalistic. Everything for Levin boils down to patterns that in the real world find expression in an arrangement of parts. He finds arrangements that are not readily referred to Darwinian or other materialistic processes, so he invokes Platonic space to account for such arrangements.

But Platonic space seems quite limited in the work it can do for Levin. There is no causal process acting from that space and projecting into the physical world. Rather, Levin discovers patterns in the physical world, and to the degree that they are striking or unusual, he refers them to a Platonic space. Take for instance steganography, the last point listed in ID research areas (section 5 of this essay). I’ve been writing about this for about 25 years. In his interview with Fridman, Levin is quite taken with the concept.

Thus Levin finds for various processes in nature, mathematical patterns that attach to them that on their face we had no right to expect (he considers, for instance, distinct sorting algorithms that when used together lead to certain unexpected patterns of association). Now I would argue that for some of these patterns, they can’t legitimately be regarded as steganographic. But the bigger point is that when true steganography arises, as in one message embedded in another, it is because an intelligence has carefully arranged things so that they two messages reside in the same medium. Steganography results from an intelligent designer inputting two layers of information into the same medium. It is not just a correlation with an ill-defined Platonic space.

What’s lacking with Levin is any conception of mind as a unity with conscious thoughts and experiences that is not itself a pattern or arrangement of parts but rather has the causal power to generate patterns, especially meaningful informational patterns relevant to biology, cosmology, and communicating with other agents. For most members of the ID community, the intelligent designer responsible for life is such a mind. For Levin, by contrast, minds are themselves patterns. This leads him to view simple feedback systems, like thermostats, as minds. This is setting a very low bar for minds.

Yet this is a common move for Levin, namely, to take something mysterious, such as consciousness or mind, and then engage in a leveling maneuver that identifies it as simply a difference in degree from other things that are non-mysterious. It is the sort of leveling maneuver that Darwin was famous for, as in treating human morality as continuous with the morality of beasts. Levin seems intent everywhere on finding continuity where the most natural interpretation is to find discontinuity.

Take, for instance, Levin’s thoughts about the origin of life. In the interview with Fridman, he won’t answer what life is. Rather, he views life in entirely flexible terms, in fact so flexible that it’s hard to rule out anything as not being alive. I was reminded in listening to Levin here of some of my readings as a teenager when I was quite taken with Hinduism and read a book on gnana yoga (yoga with the focus on knowledge), where every aspect of the world was considered alive (everything from rocks and metal to humans). In a similar spirit, at one point in his interview with Fridman, Levin remarks that at a conference about the nature of life with 65 participants, there would be 65 definitions of life.

Such a fluid understanding of life has profound implications for explaining the origin of life, namely, there’s no profound problem requiring explanation because some extremely simple “living things” require no special explanation, and the origin of life thus exists on a continuum that can largely explain itself. It feels like a dodge. By contrast, consider the following recent interview between Jim Tour and Tucker Carlson:

Asked about what is life, chemist and origin-of-life expert Tour explains certain non-negotiable characteristics of life (starting near the 13 minute mark):

It’s responsive to the environment. It has growth and change. It has metabolism. It breathes. It has homeostasis. It’s made of cells and it passes on traits to offspring. Those are the characteristics of life. That’s very well defined by science. And now you see scientists trying to change those definitions and give it a much smaller number of those things. So it’s something that they can claim that they have made, but what they’ve made is just a bunch of nonsense… It’s really not life at all. It’s things people have seen for a long time. It’s just really not life.

I mean, show me the homeostasis. Homeostasis is a constant internal state. So, if you go outside on a freezing cold day, your body has to maintain its body temperature. And not just your body, every cell in your body has to maintain a proper temperature, proper pH balance, proper proton interactions, has to still make ATP, still has to carry out all of these functions. That’s homeostasis. All of these vast number of things have to happen. You show me any one of your synthetic garbage that you say is a cell and show me the homeostasis in this. It’s not there. Show me the metabolism here [is] real. Show me that it breathes. Show show me what it’s doing with oxygen.

In conclusion, Levin does good work when he points out what standard Darwinian accounts often soften or ignore: organisms run on information, function, and control across many levels. But he explains these realities by appealing to a “Platonic space” of patterns and by stretching key terms—mind, agency, life—so far that simple feedback systems can count as minds and almost anything can count as alive. That move lowers the bar for what needs explaining. It also blurs real breaks in nature by treating them as continuities by definition. And it leaves a basic gap: patterns, by themselves, do not cause anything. A pattern does not build a system unless some cause can select it, impose it, and instantiate it in matter. And if the pattern exhibits complex specified information, that cause will need to be intelligent.