E012: Doing the Research
Kate,
I’ve been thinking a lot about the statement AI supporters make regarding the future of “prompt engineering”. They say that in the future, many positions will be available for prompt engineers, people who know how to ask the right questions of AI tools.
I mostly laugh at this, as if we haven’t attempted to train children and adults for years on proper search terms or keywords. Having worked in a library and spent multiple years of my life during research, people don’t know how to do that already. Actually, “prompt engineering” is already a skill people should have. And they largely don’t. Do you remember the commercials we saw as children about using the right keywords? Or the lessons we spent in the library learning how to use Google and not use Wikipedia to do our research? I feel like most people forgot about these things.
This came to mind last week when I someone livetweeted their experience of using some AI search to research the list of proposed statues from the NEH announcement. Their prompt, while decent, failed - after a half-hour, the tool was unable to provide details on five people from the list, and added research from those not on the list at all. This person was a good researcher - or at least, qualified to do research given their role in academia. They could’ve spent the time to do the research themselves - onerous for the purposes of self-exploration for this question.
I thought about this again over the weekend, when I was thinking about the quote from the 2003 The Cat in the Hat film where the titular Cat notes the uniqueness of a Rhode Island license plate. Of course, having grown up near the Rhode Island border, this joke didn’t really make sense to me as a kid - I saw those all the time. And then I wondered if the joke made sense in-universe - does Cat In The Hat take place far away from Rhode Island, meaning that location is the reason you never see those? Or was the Cat erroneously assuming that because Rhode Island is a small state, there would be less license plates from the state. And in that case, if we don’t account for location, which state issues the least amount of license plates?
I could do this research - it would be a lot of work for a passing thought about a throwaway line from a crude film, but I could probably do it. I would need to find the number of plates issued by state (accounting for duplicate plates, active plates, and all forms of vehicles with plates). That would give me a number for which state issues the least amount of plates, and presumably the least amount of registered vehicles. If I cared about location, I could probably do some math as well - I’m not sure off the top of my head how I would calculate those odds, but it’s possible.
I say this to mean that I know how to do the research. People often note how smart I am, but I attribute most of that to knowing how to search. If a question like this comes to mind, I do the brief research of looking through various search engines, databases, and libraries to find something. This often takes me on weird detours, where I pick up useful but odd information about the world. Again, it’s not that I’m smart - it’s that I’ve taken the time do the research. I know how to ask the questions, and I also know that most people don’t ask the questions. I know to account for all these things - to make a research plan, to consider the challenges of finding the data or the information, assessing the sources, compiling them, and making an assessment. Prompt engineering is already a thing - and it’s extremely undervalued by most people who aren’t librarians.
Also, did you know they are remaking The Cat In the Hat with Bill Hader? And Bowen Yang will be in it? I sense an SNL film in the making…
Off to do some research,
Emily