Information dissemination, and how the pandemic has distanced us and brought us together

During Marko Korhonen’s talk, he warned that there are lots of abbreviations he uses in his work that many outside of it are unfamiliar with, and that he might slip up and use them without explaining them. This is what is called “knowledge bias” or “curse of expertise”. As someone lacking the relevant background, I was also thrown off by some terms during this week’s readings. For example, when Sawhney (2019) mentioned “CSS concepts”—which I only know as “Cascading Style Sheets” from developing websites—or when trying to understand the difference between computer-supported cooperative work (CSCW) and human-computer interaction (HCI)… if any.

Nevertheless, I found it to be a very compelling read, particularly in how social media is being increasingly used in crisis response. In artificial intelligence (AI) or more specifically my field of study, machine learning (ML), the priority is very often on performance in narrow fields, and rather detached from the real world. It is likely not a coincidence that for this year’s NeurIPS, one of the most prominent machine learning (ML) conferences, introduced a mandatory broader impact section for submissions, to discuss ethical aspects and future societal consequences. Similarly, the European Medicines Agency (EMA), which regulates new drug candidates, has “made it mandatory to submit a lay summary alongside each new clinical study“. It is of vital importance that researchers and domain experts know how to communicate their findings to the public, lest it gets misrepresented or even subverted by a less-informed media. With increased virtual presence, the virality of fake news is all the more potent.

I enjoyed reading about the requirements of AI systems in crisis contexts (Sawhney, 2019), and started looking up the references. Although, I was often thwarted by paywalls. This was also new to me, as in my field, almost all papers are freely available on arXiv. Of particular interest to me was “Social media for intelligent public information and warning in disasters: An interdisciplinary review“, a very thorough survey of existing research in computing, spatial analysis, medical research, etc. There are so many things that different fields could learn from one another. So often, the same core problem or idea comes under different names—CSCW and HCI, or “out-of-distribution detection”, “robustness to data shift”, and “open-set recognition”. However, naturally the survey raises orders of magnitude more questions than it answers. This world of social media is very much a brave new world that we are so very ill-equipped to navigate with foresight, only stumbling around in the dark.

My personal experience with the COVID-19 pandemic is a reshuffling of personal priorities, from global to local, from outward to inward. Like many others, I have lost opportunities, money, and time. At the same time, I have become more interested in the bigger picture of where humanity is headed, when previously I was caught up in my own field of study. The vicissitudes of the pandemic have driven many of us apart physically, but have also inspired feelings of solidarity and determination to do something, as evidenced by the living work-in-progress Google documents of this week’s readings. This global survey of 12,845 participants in 13 countries reported that over half of respondents in most countries said that they were watching more news coverage. Perhaps linked to the idea of disaster capitalism that Naomi Klein describes, when disaster strikes, it is when people are the most galvanized. As this Guardian piece puts it, we have collectively broken out of “autopilot”.