Screenshot of a woman sleeping with pillows printed with the labels of medicines popular in China for treating COVID after the country suddenly dropped its “Zero Covid” policy in December.

Well, things have started off with a bang. Just around the time classes were wrapping up last term, China dropped its “zero-covid” policy and leaped into a massive experiment of “let’s see how fast everyone can get covid.” Man, that was fast. After a few weeks, though, where I lived seems to have calmed down quite a bit, and people are more chill. I’m still trying to get used to the idea that now I really can just buy a train ticket, get on a train, and go somewhere. I don’t need to organize tests or track outbreaks to calculate my odds of being stuck in quarantine somewhere or blocked from staying somewhere or returning home. After about three years of prepping for all possibilities, this new laissez-faire world is… strange…

Term Start

But fortunately, there’s a new term to contend with, and many things are piling up in that regard (along with emails). Some random bits and pieces to the start of the term:

Ethical Data Futures — I am thrilled with this class so far. The reading list is spectacular, and a lot of the authors I’m already familiar with from past obsessions.

Inclusive Society Intensive — This was fascinating, and I’m enamored with my group’s topic, Data Activism. It turns out there’s been more than I would have expected about Data Activism carried out by environmental NGOs in China, and I happen to have a contact at Zhejiang University, where a lot of this research is coming from. I don’t know yet what to do with that piece of trivia.

Data Science for Society — This intensive isn’t coming up for a few more weeks, but I’m excited about it. Lots of Jupyter notebooks to complete for the intensive, but instructions all seem very straightforward.

Project Thoughts

The arc of my project research, which was chasing migration a few weeks ago, has fallen down the rabbit hole of how AI creates and, conversely, how it might combat disinformation/misinformation. Maybe create isn’t quite the right word; maybe amplifies is more accurate? A couple years ago, I did a bunch of digging into the human side of the “misinformation for clicks/money” industry. What I was intrigued by were these communities where people discovered they could earn money from social media platforms by producing articles that voiced extreme viewpoints (the viewpoints were often those found in disinformation campaigns, but they could be anything that got a strong reaction from people) backed up by additional made up facts. They even came up with formulas for the best mix of truth to fiction to spread articles faster. Generally, 70% truth mixed with 30% fiction was a popular mix that was pushed extremely quickly by algorithms. A lot of these people didn’t speak the languages they were publishing articles in, so they used existing articles remixed with additional machine-translated “improvements” grabbed from articles in other languages or from their own imagination (“Fun” fact: if you want to quickly remix the language in an article so it doesn’t look plagiarized, software translate it into another language and then back into the language it came from–or at least this was one method people found to do this before the rise of ChatGPT and the other AI writing software) When people talk about social media amplifying disinformation, a percentage of what’s getting amplified isn’t the disinformation produced by government actors but rather this second layer of misinformation that is monetizing disinformation narratives by remixing it.

Any AI trained via stuff picked up from scraping the web, by the way, is potentially feasting on that soup. Which is kind of terrifying to think about, no?

So one of the current ideas is to deal with opening Pandora’s Box by creating more AI to wrestle with the demons we’ve let out. MIT researchers have built a program that follows disinformation narrative sharing to hunt the accounts most responsible for a particular narrative’s amplification on Twitter.

I don’t yet know where I’m headed with this line of questioning. Eventually, I hope to loop it back to migration, but I was hoping to find a data approach (with data I have hope of actually acquiring) to look at migration and disinformation when I initially headed down this path. Might have taken the wrong trail. I guess at this point, I’m happy this project doesn’t need to get finished until next year.

New Term, Inclusive Society, etc – Weeks 1 and 2 / Knowledge Integration and Project Planning: Data, Inequality and Society (2022-2023) by blogadmin is licensed under a Creative Commons Attribution CC BY 3.0