What does social intelligence mean to you?
In the context of my work, social intelligence is largely about modeling the public sphere. I’m originally coming from a background in history and digital humanities. I did my thesis on new methodological tools to study the history of newspapers since the early 19th century and later on wrote a book about the long history of fake news before the Internet. Social media is not that different! They have become a huge part of our political and cultural landscape. Yet, they can be easily manipulated, which also happened quite often in the traditional media in the past.
Now what’s fascinating with social media is that you can really “see” new ideas, communities and cultural representations emerging almost live. And thanks to the progress in artificial intelligence, we are having more and more tools at our disposal to explore this.
What motivates you in your work? What makes you want to keep working in social intelligence?
At the core, I have the conviction that social media is amazing at canalizing and gathering “collective” intelligence. But it’s not always easily visible and frequently derided. That’s something that has been made very tangible when Twitter had all these issues lately: people care about what they post here. And yet, there is surprisingly little way to archive or highlight all this, as preservation of social content is really poor and the filtering algorithms of the platforms.
Beyond our occasional projects at OpSci, we have started to create more general collections to ensure it could be used in the future even if the infrastructure of social networks comes to fail. We are also developing “alt-algorithms” and indicators, as a way to highlight aspects of the social media that may be lost in the feed.
It’s also a research motivation: social network studies is a very dynamic field with plenty of innovations going on. I really feel part of a collective journey and I also see how all the new methods and tools which create could be beneficial as well for plenty of other areas in social science and beyond.
What do you think makes you successful in your work?
I can roughly define my current expertise as “panoramic reading”: I basically create automated categorization systems that make it possible to browse a very large collection of texts, images or other “things”. What’s crucial is not only to have a rough idea of what are the most important topics or patterns, but to be able to zoom in on interesting things: new concepts emerging, influential debates or significant anomalies that may suggest that there has been some manipulation.
Besides that I have been passionate about digital media for a long time. I have been a contributor and admin on the French Wikipedia for a long time and care a lot about the “digital commons”, which also lives on private networks.
What are the key skills that have contributed to your success?
A combination of soft and “hard” skills. As I started my research career, it became obvious that the dramatic expansion of text and data available online created new opportunities. So I started working with all kinds of text mining tools and experienced the AI revolution from the inside, starting with “word embeddings” (which look now “dated” although they are less than 10 years old!) to BERT and GPT.
As of today, I am used to training and applying deep learning models. Thanks to the huge development in open source programs, it’s actually not that hard. Most of the work resides nowadays in the training and the documentation of the model inputs or outputs. In a weird twist, as the techniques have become more sophisticated, the focus turns more and more on qualitative expertise.
So aside from the technical aspect, being aware of the culture of social media and the way they work in practice is extremely important.
What makes social data special compared to other data sources?
I’ve worked with other very large digitized corpus (especially newspapers, but also literary fiction). Where social media stands out is not so much that they are big and contain a very diverse snapshot of what people think at a given time. It’s rather that they are so “alive” and all changes happen so quickly. It also brings immediate connections between so many different people and networks, which is not something that has happened before, with discussions that would be held in private professional circles which suddenly find themselves superimposed on the trending memes of the day.
There is some downside to it. For some project at OpSci we had started to have a longer time perspective on social media output, such as looking at the evolution of the representation and perception of nuclear energy on Twitter and Facebook for ten years. And this is hard! The infrastructure has been constantly evolving and most metrics are not very usable on a long term basis. I think in the coming years, there will be a real reflection over social networks as an “archive” but we are not there yet.
What does being a social intelligence pioneer mean in the context of your work?
Mostly a lot of freedom to analyze and to invent new methods. I would say that I spend a lot of time trying to “translate” a qualitative approach into models and algorithms.
This can be also a bit disturbing as the tools we have continue to evolve as we confront them to the actual data, trends or insights and get new ideas. This is really not about simply pushing a few buttons and getting the final output but more about constantly recreating the framework so that it matches what we want to do.
I think I have had some exposure to ways of working that may become much more widespread in the future, with the increasing progress and availability of AI.
Get Social with SILab
