A Babble of Voices (AI, Law, and Otter Things #11)
Hello, dear reader, and welcome to another issue of AI, Law, and Otter Things! This time, I write about a few things — academic writing in/as a foreign language, personal data, positive feedback — loosely connected by the idea of "voice". While this might be partially driven by how much I miss karaoke nights, I won't spend much time discussing music. I will, however, suggest a few songs also loosely connected with the topic: one for the metalheads, one for those into Brazilian music and one for the Bene Gesserit. Feel free to pick one (or none, of course) before moving forward to the rest of the newsletter.
Finding my voice in writing
As I was ending the last issue, I mentioned that there was not enough time for a rant about humour and writing in English. Since writing that, the topic has stayed with me for a while, not so much because of the humour paper I discussed there, but more because I finally stopped putting off some Portuguese-language projects I'm working on. Thanks to the contrast between my writing in both languages, I could not help but wonder why I sound so different writing in my native language.
My personal answer to this question, of course, relates to the more general challenges facing non-native English speakers in academia. Even though I feel way more comfortable in writing scholarly work in English than in Portuguese — especially since legal English is somewhat less baroque than legal Portuguese —, I still find myself adopting an excessively cautious approach when polishing my first drafts into something readable. As a result, my self-editing cuts often end up removing lots of jokes and making explicit things that could have been left implicit.
To a certain exchange, this artificial character of my academic English voice is positive, as things such as pop culture references can make a text less accessible (though, contra Conklin, I think this is not always the case). And I am not alone on that, since nobody is a native speaker of academic English: even those who spend their entire lives on English-speaking environments must still get acquainted not only with the technical jargon of their fields but with the language needed to navigate the administrative mazes associated with degrees and faculty positions.
That being said, it still can be quite annoying to realise you cannot get away with things that are forgiven for native English speakers. Especially because people often are assholes when pointing out what needs to improve, to say the least. But, fortunately, I have found myself in a great environment for finding my voice, as the EUI offers considerable support in terms of revision services and language courses for improving one's presentations and written work (thank you, Nicki!). It might be a while before I am comfortable in expressing myself in the jokes and similes that often appear in my communication in Portuguese and non-academic English (even though some would argue this is for the better), but at least my writing process no longer feels like trying to avoid pools while walking in the rain.
Protecting voice as personal data
Moving on to a more literal meaning of "voice", my first English-language journal article (with Juliano Maranhão) examined the legal and ethical implications of voice-based diagnosis tools for Covid-19. In the early days of the pandemic, AI scholars were trying to find tools that would allow people to know their health situation without going to the hospital (thus exposing themselves to an increased risk of infection). Among the various approaches proposed for that task, some research projects — such as Cambridge's Covid-19 Sounds or the Brazilian SPIRA — suggested diagnosing Covid-19 by analysing a recording of a user's voice and comparing it with the voices of healthy and infected people. Our paper examines the ethical trade-offs involved in remote diagnosis and whether those tools were compatible with Brazilian data protection law.
At the end of the day, it turns out that AI diagnosis tools — voice-based or otherwise — were unsuccessful against Covid-19. Nevertheless, I still think the article had a few valid contributions. First, it provided an English-language overview of some Brazilian data protection law provisions, such as the lawful bases for data processing and the right to review automated decisions, which differ from the GDPR in subtle but important ways. It also contributes to the literature on voice as personal data by examining the inferences drawn from voice recordings and their relevance from a data protection perspective. While I don't think I will come back to the healthcare-centric aspects of the paper any time soon, both of these points are things that still interest me, and I hope to revisit them in the future.
Giving voice to what I like
Given that I dedicated quite a few words above to plugging my work and that of my collaborators, it is only fair that I now spend some time speaking about the work of other people that I like. Once in a while, I will try to write a few paragraphs on someone whose works I like, why you should like them too, and where can you get started with their work.
This decision was inspired not just by my self-indulgence last week but also by Jotwell, a great website in which legal scholars write about recent pieces of scholarship that inspired them. Not only Jotwell is an excellent way to stay in touch with the current state of the literature, but it also provides the kind of positive feedback that is often missing whenever we discuss a paper.
Don't get me wrong: I am not above complaining about other people's work, sometimes loudly, as my personal Twitter feed shows. I also think that negative feedback plays a vital role, as some relevant changes to my work have been prompted by commentaries that were neither kind nor fair. But — and this might be Ted Lasso influencing me — sometimes it is helpful to be reminded that you are doing something right, or at least getting it wrong in an exciting way.
Because of that, I will try to emphasise living scholars that do not receive as much attention as I think they should. This will usually mean early-career researchers, but I also hope to highlight some established authors underrated because of structural factors, such as the core of their work being in languages other than English or their geographical origin. Of course, the selection of authors will be shaped by my linguistic competencies, personal interests, ideological biases, and so on. And there is always the chance I will mischaracterise these authors by reading too much of Marco Almada into them. Still, I hope this helps to get some interesting people to read other people I think are interesting too.
Listening to the chorus
Finally, I would like to take the time to suggest a few things that caught my attention over the past week, either because I interacted with them over the past few days or because I ate a madeleine was otherwise reminded of them.
“Buffy the Vampire Slayer”/”Angel”: The Unified Skip It/Watch It Guide
In the last few days, I've decided to rewatch Buffy: The Vampire Slayer and its companion series Angel, as both are available on Disney+. However, I no longer have the patience or the time needed to sit through twelve full-length seasons of shows that, let's face it, could be pretty bad at their worst. This is why I am using this guide, which is pretty much in line with my impressions from the episodes I actually remember. Any such "abridged" run through a show means that you will end up losing something, naturally, but this one seems a good starting point for visiting two series that did not age well at some points.
Thomas S Mullaney and others (eds), Your Computer Is on Fire (The MIT Press 2021).
A collection of essays that deal with the materiality and the social embeddedness of computer systems. The book covers a broad range of topics and social issues, but I will highlight some essays that resonate with my interests. In Platforms Are Infrastructures on Fire, Paul N. Edwards explores the brittleness of the informational infrastructure that sustains (and is developed by) platforms. Ben Allen's Source Code Isn't drives home an important point: even full access to a program's code is not sufficient to understand a software program since that code only operates and generates effects through a series of intermediates (e.g. compilers). Finally, in a straightforward connection to this newsletter's topic, Halcyon M Lawrence's Siri Disciplines explores how voice recognition is not a playing field, as considerably more attention is dedicated to modelling languages, dialects, and accents associated with certain countries and social strata.
Jacob Leon Kröger, Otto Hans-Martin Lutz and Philip Raschke, ‘Privacy Implications of Voice and Speech Analysis – Information Disclosure by Inference’ in Michael Friedewald and others (eds), Privacy and Identity Management. (Springer International Publishing 2020).
The authors provide an extensive coverage of the various types of information that can be inferred from a person's voice. Sometimes those forms of inference are frightening because they manage to get good results. Still, even inaccurate inferences can end up having practical effects if they are seen as socially acceptable. Fortunately, this article provides a good overview of not just the inferences but of potential approaches to mitigating risks.
Bret Devereaux, ‘The Fremen Mirage, Part I: War at the Dawn of Civilization’ (A Collection of Unmitigated Pedantry, 17 January 2020).
The author coins "The Fremen mirage" as a term to describe the idea that "barbarians" are better fighters than the "decadent" complex societies that they occasionally overthrow. In this first part of a series, the author shows that the idea of decadence is connected to agricultural societies, but these have had a stellar record in their clashes against pastoral-nomadic societies. Other posts in the series show that victorious "barbarians" such as the Mongols are the exception rather than the rule, but discourses of civilisational decay are nevertheless invoked by authors in service of their own political goals.
Benjamin K Sovacool and David J Hess, ‘Ordering Theories: Typologies and Conceptual Frameworks for Sociotechnical Change’ (2017) 47 Social Studies of Science 703.
Through interviews with various Anglophone STS scholars, this article identifies 14 main theoretical approaches in the field. These approaches, as well as their variants and other "minor" approaches, fall into five main analytical strategies and areas of focal attention (agency, structure, meaning, relations, and norms) and four main underlying groups of underlying assumptions and goals (functionalist-institutional, culturalist-interpretivist, critical-humanist, and conflict). With this classification work, the authors provide a handy map to understanding the central tensions and issues in current STS literature.
Voices off
Thank you for taking the time to read this week's issue! I hope it had something of interest for you, and I would love to hear from you if that is the case. In any case, see you next week, hopefully with a theme that entertains me at least as much as this one.