Since our unit a few weeks ago on AI/Algorithmic Knowledge a couple of things have been stuck in my brain. Lauren Klein talked about using AI to take notes during doctors appointments that provide a summary at the end for the patients records. She talks about doctor burnout and how this would make their administrative tasks easier, giving them more time to actually practice medicine. This is clearly problem that needs to be addressed and AI could be the answer. I am already skeptical of AI chatbots/assistants, but I have serious concerns about who owns that AI and what kind of access they would have to medical records. What will happen when AI is able to have intimate knowledge of discussions between doctors and their patients? From a surveillance perspective this is a terrifying thought. Aside from the existential questions, Klein goes on to explain that the AI isn’t always accurate, it makes errors in word choice and even subs in different names for medicines. I guess more training in this environment could improve the AI, but what is the cost of that, both from a privacy perspective and the errors made in the training process?

This goes into the other thing I cannot stop thinking about, which is how AI is trained on large data sets. In The Atlas of AI Crawford gives a brief history about where IBM found training data sets in the 1970s for early AI – user manuals, children’s books, patents, and then ultimately an IBM lawsuit. She goes on to explain that the internet changed the scarcity problem with large data sets, suddenly there was plenty of freely available content in high quantities. Crawford describes the internet as a natural resource for AI (106). I now feel acutely aware of the ways my own data can be scraped from the internet and what it can be used for. The who, what, and why of this concerns me. I have come to accept that AI is inevitable. My wife consistently reminds me that it is here whether I like it or not. I think it can be used as a tool to make life easier, or I guess I think that’s what it should be used for. But, I do not trust the owners of these tech platforms and the way they can manipulate us using data they scrape from our online footprint. At this point it feels like the bad outweighs the good… is it possible for AI to exist only for “good” (and who decides what that means)? For it to only be a tool to make life easier (rather than the be all end all)?