Free your inner Llama

LLAMAS

Llamas are of a special kind. At the Berkeley Campus of the University of California, during the exam period, llamas are brought in for a ‘Llamapalooza’, where students can pet and cuddle them (Cantor, 2019). The students seem to reboot while enjoying the physical proximity of their grass munching companions. Llamas are ‘charismatic’, in the sense of having a natural attractiveness. Some might claim this demonstrates their cleverness but it has also resulted in llamas as therapeutic pets, not merely with stressed students but also in nursing homes or engaging with people with cerebral palsy or autism. Llamas are described as patient, kind, capable of ‘reading’ their human counterparts, smart and curious, as well as strong-willed, insubordinate and reasonably sovereign when challenged to take on unwarranted tasks. In the meantime, there is much more to learn from llamas and other subspecies of the camelidae, for instance at the level of their immune system.

Llamas are part of the species of  camelidae  (e.g. camels, llamas and alpacas). On the Ablynx website we find that camelidae ‘possess fully functional antibodies’ that can be cloned and isolated with ‘full antigen binding capacity and are very stable’, and ‘form the basis of a new generation of therapeutic molecules which Ablynx has named “Nanobodies”’. This originates in the invention by the founders of this chair, who first isolated the relevant antibodies. Interestingly, in 2020 ‘[a]ntibodies (nanobodies) derived from llamas have been shown to neutralize the SARS-CoV-2 virus in laboratory tests by UK researchers’, capable of stopping the virus in patients with COVID-19 (Lynch 2020). This reminds us of the crucial role played by fundamental research in generating societal benefits, but also raises fascinating questions about the political economy that informs the extraction of added value from patented animal material, and thus brings us straight to the core of this chair: the history and the role of sciences in society.

SCIENCE AND SOCIETY

As Felix Stalder noted, the corona-virus pandemic has caused ‘a return to the real’, and as many have noted, it has highlighted the pivotal role of science in society, while also raising new questions about the increasing dependence on data- and code-driven infrastructure.  Key questions  must be addressed, such as:

1. How to steer free from fake news and post-truth relativism without reverting to naive understandings of what scientific knowledge means? 
2. How to ensure that emerging technologies contribute to a thriving economy that affords a fair distribution of human capabilities across dominant and marginalized groups, both globally and locally? 
3. How to resist investment in technologies that diminish human agency and how to reform business models whose added value is based on predatory extraction? 

These questions cannot be addressed by adding more computation or by investing in more data gathering. They concern the assumptions of data- and code-driven research, the assumptions of our information ecosystem, and those of prevailing economic science. Time is ripe to reconsider the relationship between the sciences and the humanities, developing new alliances, effective mutual respect and learning – way beyond the ‘two cultures’ approach. We should note that in the English language academic study is divided between the natural sciences, the social sciences and the humanities, suggesting the humanities are not scientific, whereas in French and German we have three types of science under one heading (Wissenschaft, Science): Sciences naturelles, Naturwissenschaften; Sciences sociales, Sozialwissenschaften; Sciences humaines, Geisteswissenschaften. This chair will take the perspective of continental Europe: though methodologies and objects may differ, not one method or framing should colonize the others (which is not to say that anything goes), all sciences should be recognized as such. We should also note that an open society aligning with democratic values sustained by the rule of law must ensure that those subject to political decisions understand the value of recognizing uncertainty without succumbing to either data-fetishism or post-truth bullshit (Bergstrom and West 2020; Popper, Ryan and Gombric, 2013; Cilliers Preiser, 2016).

THE LLAMA FOR POETS AND QUANTS

So, where does the llama come in? If it is true that llamas are ‘patient, kind, capable of "reading" their human counterparts, smart and curious, as well as strong-willed, insubordinate and reasonably sovereign when challenged to take on unwarranted tasks’ we have a lot to learn. Though we have a different type of agency, being ‘linguistic bodies’ in a way that llamas are not (Di Paolo, Cuffari and de Jaegher, 2018), we could do with some inner llamafication, both in the sense of enhancing our resilience with effective nanobodies and in the sense of becoming more trusting, curious and subversive, depending on what our environment demands. I would even dare to stretch the metaphor a bit further by suggesting that democracy and the rule of law require precisely this kind of sprezzatura, concern and resistance.

OBJECTIVES AND RESEARCH QUESTIONS

The questions raised above highlight the need to investigate the relationship between science and technology, to examine the political economy of data- and code-driven architectures and the interaction between science, democracy and the rule of law. This involves bringing together ‘poets’ (those versed in the theoretical reflection and qualitative investigation) and ‘quants’ (those versed in calculation and quantitative inquiry) (Tunarosa and Glynn, 2016). In the context of the broader goal of the chair (see above), the chairholder will focus on three objectives:

1. An in-depth inquiry into the nature of counting and speech, calculation and qualification, computation and argumentation, across the domains of the sciences and the humanities, unearthing what distinguishes computational from narrative truth claims, how they relate to each other and how they interact. 

2. An in-depth inquiry into the performative effect of computational systems in the context of scientific research, including the role of theory, experimentation, data and code. 

3. An in-depth inquiry into the status of law as a scientific discipline, in particular in relation to political theory and computer science, to address the rise of computational ‘law’. 

Relevant research questions will be selected and developed in close collaboration with in-residence researchers. The following research questions indicate the kind of research we target:

1. If calculation and quantification presume qualification (to decide what counts as the same), how can we invite scientists to make their qualifications visible?

2. What can those mainly working with quantification learn from those working on theoretical reflection and/or qualitative research (Tunarosa and Glynn. 2016)?

3. How does formalization relate to quantification? 

4. What risks are involved in using observable proxies for target concepts that are not directly observable (Polack, 2020; Yarkoni, 2019) ? 

5. If machine learning generally lacks theory, what does this mean for the reliability of its output (Garip, 2020; Gigerenzer, 2018;  Lipton and Steinhardt, 2018 ; Sculley, Snoek,Wiltschko and Rahimi, 2018)? 

6. To what extent are predictions based on behavioral data liable to become self-fulfilling prophecies (Esposito, 2011;  Manheim and Garrabrant, 2019 ; Strathern, 1997) and what does this mean for policy research? 

7. To what extent do predictions based on behavioral data reinforce existing unfairness due to different false positive rates across marginalized groups (Barocas and Selbst, 2016)? 

8. In what sense is the study of law a scientific undertaking and what does methodological integrity mean in the context of law and the Rule of Law (Dworkin, 1991; Hildebrandt, 2020;  Kestemont, 2018;  Waldron, 2011) ? 

The chairholder will actively generate, enable, support and foster poets and quants across the borders of natural, social and human sciences, transversal and university-wide, but also invite transgressions across the borders of scientific research – seeking inspiration, confrontation, subversion and integration with the arts. This implies loyalty to the methodological integrity of the sciences, respect for tacit knowledge and unmethods of artistic endeavor and learning how ‘not everything that can be scientifically proven matters, and not everything that matters can be scientifically proven’.

WANT TO DELVE IN THE THEME?

Have a look at the bibliography!

This fellowship is part of the Casterman-Hamers chair for the history and philosophy of the sciences. Belgian immunologists Cecile Casterman and Raymond Hamers established this chair in 2013 in response to a concern for the role of the sciences in society. The current chairholder prof. Hildebrandt suggests that this includes not just the societal benefits of research but also the broader political economy in which (predatory) value extraction from new knowledge takes places. This is fitting as the discovery and patenting of camelid ‘nanobodies’ by Casterman and Hamers made them part of the biotech revolution that expanded epistemic and economic horizons.
 

THE BENEFITS OF SCIENCE IN THE ALGORITHMIC CONDITION

What technological shifts are today changing research and innovation ecosystems? How might the existing political economy of science be shaken up? What horizons for societal change can knowledge production produce? For this 2-year long Free your inner Llama fellowship I want to rethink and reimagine the role, effects and value of the sciences in the algorithmic condition (Colman et al. 2018). Similar to the promises and problems associated with biotech in the 1980s, many argue today that artificial intelligence (AI) might radically change the benefits and business models of science.

The locus of scientific research is particularly poignant for the effects of AI because they share certain assumptions and approaches, like modelling and prediction. And, advocates of both science and AI have claimed that the quantitative and calculative nature of these practices brings forth objective knowledge. It is no surprise, then, that many see great potential for the application of AI in scientific research: if the counting practices of the sciences were always already objective, and if code can recreate these ways of counting and calculating, then these practices can be automated, accelerated and even ameliorated through the use of algorithms. In this narrative, the reduction of the ‘human’ factor in research is applauded as it could decrease bias and mistakes.

Counting, calculating and measuring are indeed, epistemologically and historically, at the core of many of the sciences. But, these quantification practices do not simply re-present the world. Philosophers, historians and sociologists of science have shown how, actually, counting (re)constructs a world. No meaningful calculation can actually take place without a prior circumscription of what matters. Quantification and qualification – or qualculation (Law & Callon, 2005) – are strongly intertwined, also in the practice of scientific research.

A similar criticism is often launched against artificial intelligence: there is still a great deal of human knowledge and values going into the algorithmic systems that deal, arguably,  ‘automatically’ with massive bodies of data (Bechmann & Bowker, 2019). Again, it’s critics from the humanities and the arts that show that also for code- or data-driven applications there is no such a thing as effortless objectivity.

In a salto mortale AI is not only changing various research practices but also its legitimation in terms of its societal benefit or impact. Various private enterprises, like Elsevier, Vertigo Ventures and Clarivate, are selling data-driven tools to institutes of higher education with which they can map, track and visualize contributions of their research to the UN’s sustainable development goals (SDG). These goals, however, are themselves the result of arduous political bargaining (Fukuda-Parr & McNeill, 2019) and represent a particular political-economic perspective that refuses to challenge the ‘hegemony of growth’, which many today argue is actually at the root of our climate crisis (Schmelzer, Vetter & Vasintjan, 2022).

This poses a problem beyond the well-known rule for data-driven technologies of ‘garbage in, garbage out’ (GIGO). When discussing the role of AI in knowledge production, we should deal with the PIPO principle: ‘politics in, politics out’. But not just as a raised fist in anger. PIPO could be the battle cry for our desire to explore the potential for resistance to the status quo and subterranean experimentation with alternatives. Strategies of rarefaction, proliferation and exploration could allow the imagination and creation of ‘alternative ways of life’, slow science, or commons for ‘other’ knowledges (Santos, 2008; Stengers 2017; Wark, 2015).

Coming back to the focal point of this fellowship, I want to explore creative means by which to grasp what is happening in and to the sciences as the use of AI spreads and diversifies. Similarly to the agenda of the chairholder this project will take issue with specific qualculative technologies, the assumptions that inform their functioning, as well as the potential to resist and find alternatives. Instead of focusing on legaltech, this project is concerned with epistech, digital technology that intervenes in the processes of knowledge production, circulation and evaluation. As algorithms infuse more and more parts of scientific practices, the question is raised what remains distinctively human about them, and what ‘epistemic effects’ automation might have.

The goal is to appraise the affordances of algorithmic interventions without disregarding critical perspectives on the ‘strong objectivity’ of the sciences, the political economy of knowledge production and the ‘ethics of coding’ in our algorithmic condition. Then we can address questions like: What futures await us when the production of so-called objective knowledge in the sciences is increasingly automated? Does this indeed produce objectivity2, or rather accelerate bias2 and obscure ‘PIPO’? What parts and processes of human knowledge production can and can’t we translate into code? What assumptions, fields of action and interests do algorithmic infrastructures inscribe into scientific research?

Bechmann A. & Bowker G.C. (2019). Unsupervised by any other name: Hidden layers of knowledge production in artificial intelligence on social media. Big Data & Society. 2019. DOI: 10.1177/2053951718819569.

Callon M. & Law J. (2005). On qualculation, agency, and otherness. Environment and Planning D: Society and Space, 23 (5), 717-733.

Colman, Felicity, Vera Bühlmann, Aislinn O’Donnell, and Iris van der Tuin. 2018. “Ethics of Coding: A Report on the Algorithmic Condition.” European Commission.Santos,

Boaventura de Sousa, ed. 2008. Another Knowledge Is Possible: Beyond Northern Epistemologies. Verso Books.

Fukuda-Parr, S. & McNeill, D. (2019). Knowledge and Politics in Setting and Measuring the SDGs: Introduction to Special Issue. Global Policy 10 (S1): 5–15.

Schmelzer, Matthias, Andrea Vetter, and Aaron Vansintjan. 2022. The Future Is Degrowth: A Guide to a World beyond Capitalism. Verso Books.

Stengers, I. (2017). Another science is possible: Manifesto for a slow science. Cambridge: Polity Press.

Wark, McKenzie. 2021. Capital Is Dead: Is This Something Worse? Verso Books.