Science in the storm 4 – Agnotology, Artificial Intelligence, and Democracy
28 September 2021
Esther Dorothea Almqvist, Sammankomsten, 1929. Credit: wikipedia.https://commons.wikimedia.org/wiki/File:Themeeting.jpg
Private interests sometimes indulge in disrupting scientific knowledge. The study of these strategies with human sciences’ methods is called agnotology. This text and the following are edited transcriptions of an open conversation organized to discuss this matter from the perspective of scientists directly confronted with this kind of practice.
This fourth text proposes that artificial intelligence should also be analyzed in agnotological terms—moreover, the latter shares with the central dogma a strictly mechanical perspective. An alternative would require the careful articulation of scientific and activist attitudes.
Thank you both. Before trying to continue my presentation, I would like to add an element to agnotology, this science of ignorance. I mean by this the “science” grounded on big data instead of knowledge, that is, the idea that one can predict and act without knowledge, without prioritizing, and that regularities detected by machines, with no need of proposing a causal frame typically, may replace knowledge construction. The denial of theory building and understanding should be an object of agnotology. Fortunately, by mathematics, we can show that this does not work because there are correlations everywhere, even in randomly produced databases, when they are large enough (1). But of course, there is so much money and commitment that our mathematical or theoretical remarks are not observed or listened to.
These two areas, agnostic science and the object of agnotology, did contribute, we know, to affect the debate and the discussion in the current controversies, such as the war on cancer, climate change, and an area where we have some colleagues and friends who are also listening to us now, the analysis of endocrine disruptors. We know that we introduced about 80 000 molecules in the environment in the 20th century, that is, new, artificial molecules. There has been robust evidence, accumulated particularly in the last 30 - 40 years, on how this may seriously affect endocrine cascades (2). Of course, this goes against the so-called central dogma of molecular biology and one of its fundamental components, the exact macro-molecular interactions. As long as one believes that macro-molecules interact in a key-lock paradigm, there is no way that these small molecules, as most of these newly invented molecules are small molecules, would affect their dynamics. However, they do, first of all, because macro-molecules interact in a largely stochastic way, and the associated probabilities depend on the context. Second, these different molecules may affect, by modifying the context, thus the probabilities of interactions, the activity of the proper molecules of their organism as endocrine components, either by interfering with the molecular cascades in the organisms or by affecting the molecular receptors.
Now, how to fight these beliefs? Of course, for our colleagues who first discovered the role of endocrine disruptors, for example, in cancer, like Sonnenschein and Soto, it was not a minor problem to have to face the interests of the industry which produced several million tons of bisphenol A per year, one of those early endocrine disruptors with carcinogenic effects. These are, obviously, immense financial interests as you know there are in GMOs. I’ve been in touch thanks to Ignacio with countries where there are pervasive applications of GMOs, like Mexico and Brazil. There are not many polemic or controversies; people are simply and directly killed for fighting against the disaster that is happening as a consequence of GMOs. The disaster has many reasons; the main one is the effects on roots and their microbiome, as side effects of pesticides on other life forms; since roundup is a general poison – only the intended GMO survives it, everything else dies. So, we are getting to this level, where an application which was supposed to save humanity from hunger, is now contributing to ecological disasters and death, in many countries, from Latin America to India. Recently in a presentation where the number of suicides was shown (they are unfortunately very common in agriculture in India), this number is much larger in areas where GMO cotton is cultivated. There are many reasons for this; one of them is that Monsanto increased by four the price of seeds in ten or fifteen years, and every year the peasants have to buy new seeds. So we are moving from the controversy to the polemic and then to a literally deadly battlefield. The underlying ideology of GMOs is the myth of a “programmable” life: now that we know the coding of life, DNA, we can manipulate it at will and control Evolution – we may enter into the mechanics of life. However, it does not work because ecosystemic and organismal effects result from an immense network of interactions at all levels of organization (3).
As I mentioned, democracy in the richest sense, namely the formation of the dissent of a minority constructing a different perspective, is at the core of both science and our democratic societies. But differences must remain, and this can only work with the dialogue.
To face these problems, we know that tools are applied from another area where the idea of mechanics is very important, artificial intelligence. Artificial intelligence is largely used now to disseminate disinformation to a targeted audience. By collecting information from individual people by machines that can keep track of everything we do in the network, disinformation is aimed at people with certain attitudes that usually search for certain information, have certain interests to be more effective. So those two areas, the biology of “programming life” and AI, joined together to provide the worst example of disseminating scientific disinformation, destroying science.
In this sense, I like very much to work in the institution which is hosting us today, Institut des systèmes complexes in Paris, because they showed how, on one side, these tools flatten knowledge as it happens on networks. When we have too many neighbors, we tend to be identical unless there is a willing construction of common and constructive networks of knowledge in the sense Ignacio mentioned before. On the other side, the networks allow the formation of isolated clusters just on purely political and ideological grounds. Within these clusters, and our friends here in this institution have shown it technically, there is some sort of resonance effect of ideas. So, fake news rebounds several times and increases in size by staying within a given cluster of people that a priori are looking at the world from a certain perspective, as beautifully summarized by the work at ISC (4). This is, of course, the opposite of constructing critical knowledge in a wide public, particularly in critical science – provided that science is always critical thinking.
As mentioned before by this beautiful definition by Ignacio, science is this collective enterprise, a critical and sustained one; it is also a construction of objectivity. I try to avoid the word truth but instead use this perspective of objectivity construction, which is the result of the philosophy of best areas of physics. We learned from physics that there is a passage from the subjective absolute to the relative objective, which is at the core of the theoretical and practical successes of Relativity Theory. Relative objective means that the relativizing subject proposes a frame, a reference system, and is able to change it. So, what is stable, the invariants with respect to the transformations of a reference system or generally of the perspective proposed, are what is considered objective. Relativizing is not stating that everything goes; on the opposite, it means looking at the conceptual stability or invariance which respect to transformations of perspectives. We can, in part, transfer this technique which is very specific even in mathematics and physics, to areas like biology and historical science.
In this, I would like to propose an answer to Ignacio, is Evolution a theory? Yes, I think so. First of all, because of its first principle. Heredity is “descent with modification”. In any reproduction, there is a variation: that is the first principle introduced by Darwin in about four chapters out of the first six, a fundamental principle. Darwin states that even breeders who try to stabilize a species fail in doing so because there will be changes in reproduction. I think that this principle is very robust, widely confirmed from microbiology to ecology. Then there is the second principle, selection. Due to an ambiguous presentation by Darwin, who is referring to breeders, it is more delicate. Many keep interpreting selection as a “selection of the best”, which makes no sense. Of course, in the case of breeders, it does make sense: there is a human who selects a cow producing the largest amount of milk, fine. But in changing environments, in the dynamics of the changing phase space of Evolution, the notion of optimality makes no sense, as there is no pre-given space of possibilities where one could give a partial order and define an optimum. So, we should replace selection by enablement: the context makes it possible, history canalizes the variation, the historical traces channel changes, including this amazing chemical trace of history, DNA, the trace of the entire Evolution. DNA and other constraints canalize variation, beginning with the Brownian motions of molecules in the cytosol ... then constraints apply at all levels of organization.
I would like to hint at a final point. We discuss very frequently in our organization, AAGT, the relation of all these questions to politics and the relation between researchers and activists. There must be, and there is a difference between these two roles. Many young people on strike or fighting against climate change are activists, and they must remain so. But there is a space of thought for science that has a different nature. As I said before, making principles explicit, criticizing them, and possibly changing them is at the core of science; instead, the fact of modifying, while working in political activity, the very principles of action is much harder. Science is different from activism, it should not be confused with it, but the dialogue between the two is absolutely fundamental. This dialog is really what we have to develop in discussion with organizations stemming from civil society. In science, one must be able to say at any moment, “I was wrong”. This is not so easy, let’s be honest, in politics, even in the best of politics. Science aims to construct objectivity, as I said before, which is different from claiming political aims, from proposing a path for change in human society. But the two must go together, construct a dialogue. As I mentioned, democracy in the richest sense, namely the formation of the dissent of a minority constructing a different perspective, is at the core of both science and our democratic societies. But differences must remain, and this can only work with the dialogue. I think we need it strongly now because we have difficulties in Academia to develop science in the sense we intend it, and the three of us gave many reasons for the failures to do science only “by projects”. This is why I believe organizations like the ones we are in, ENSSER, AAGT, are essential to establish the original links that we are constructing, not so obvious, among science, activities, and political commitment, never to be confused. Because today’s ecosystemic challenges are scientific, but, no doubt, they are also intensely political. We are all aware that they have been transformed almost entirely and by many in a political issue, particularly concerning how knowledge construction is organized and divulgated. Proposing solutions is a new open scientific issue, it requires debates and lots of research, but as for knowledge of the ongoing ecosystemic change, climate change, there is no doubt that a lot of science is already there. Opposing it is not a scientific attitude, it is purely political, while politics blends to science to propose solutions.
By collecting information from individual people by machines that can keep track of everything we do in the network, disinformation is aimed at people with certain attitudes that usually search for certain information, have certain interests to be more effective. So those two areas, the biology of “programming life” and AI, joined together to provide the worst example of disseminating scientific disinformation, destroying science.
This is why the nature of this debate, the reasons why we are discussing here, are very important. And I think we need to invent the right tools in order to have political perspectives and action and science all connected in a non-subordinate way. So that is, I think, a question I would like to leave open. Thank you.
1. Calude C, Longo G. (2017). "The deluge of Spurious Correlations in Big Data". Foundations of Science 22: 595–612.
2. Sweeney M. F., Hasan N., Soto A. M., and Sonnenschein C. (2015) Environmental endocrine disruptors: Effects on the human male reproductive system. Rev Endocr Metab Disord 16:341-357
3. This is the title and content of 2017 book by J. Doudna, Nobel Award winner of 2020, see Longo G. (2021) Programming Evolution: a Crack in Science. A Review of the book by Jennifer A. Doudna and Samuel H.Sternberg “A Crack in Creation: Gene Editing and the Unthinkable Power to Control Evolution”, in print in Organisms. Journal of Biological Sciences.