to be a better amateur

At the beginning of this chat I had with Noah Millman, you’ll note my caveat: I speak as a dedicated but decidedly amateur student of artificial intelligence. Noah makes a similar announcement. I was thrilled to be invited by him to discuss issues of the philosophy and theory of knowledge of AI, and I had a great time chatting with him. I announced my amateur status because I felt compelled to: whenever I write about more quantitatively oriented issues, people try to check my card — they make some sort of aggressive statement about my lack of expertise. Sometimes these statements are accurate, sometimes inaccurate, but the essential message is always the same: numbers-based ways to understand the world are meant to be discussed by a certain credentialed minority. I think that’s a terrible mistake, and in fact that’s why I was eager to have this discussion with Noah. I believe it’s essential for people with non-STEM backgrounds to be conversant in these topics, as they’re so important for the future. I think an informed conversation on AI between a guy with a background in writing and the humanities, and a guy with a background in history, finance, and the arts can be fruitful and useful.

To lay out my (beginner, amateur, but informed and passionate) understanding of AI, I’ve read Doug Hofstadter’s Godel, Escher, Bach twice; I’ve read a significant majority of Artificial Intelligence: A Modern Approach by Stuart Russell and Peter Norvig; assorted books in cognitive science, ranging from popular like Steven Pinker’s The Stuff of Thought to academic like Randy Gallistel’s Memory and the Computational Brain; a couple hundred articles, popular and academic; and a good deal of natural language processing that I utilize in my own academic research. Let me say straight up: there are large chunks of essentially all of this that I don’t understand, and the actual computer science that real understanding would require eludes me. There’s huge chunks that I just don’t grasp because I can’t follow the algorithms or the code or similar. Like I said, I’m an amateur. Just one who wants to learn as much as his amateur brain will allow.

There are some topics on which I am not an amateur. I’m not the type to act as though expertise is all a lie, which seems to me to create a tyranny of the ignorant. I will claim some expertise in the fields of writing assessment, particularly quantitative approaches; second language writing specifically and applied linguistics generally; writing program administration; and standardized tests of higher education, a topic which has occupied most of my attention for the past two years. There’s no bright line between things that I define as matters of my professional competence and those that I see as the interests of a beginner, but I maintain the distinction all the same.

I have, for the past six years of graduate education, gradually brought myself to a fitful and inconsistent understanding of statistics and research methods appropriate to my research interests. As I’m sure is common, this didn’t really come from some grand scheme to get quantitative. I just found that I had certain questions that I couldn’t answer without using numbers, and as time went on, I needed to know more and more. Which means that I know how to do a few things that are quite sophisticated and don’t know how to do some very basic things. When I want to do a simple confidence interval, for example, I often find myself reaching for a textbook. There’s something embarrassing about that, I guess, but I don’t mind too much; the point’s not to pass someone else’s test. It’s just to know how to ask certain questions, or to know how to find out, or how to ask. It’s like anything else: you study and you think you know something, and then you learn more and you look back on your old understanding and you say, boy, I didn’t get it back then, but now….

Which is not to say that I am not subject to the insecurity that comes with attempts to develop quantitative skills. There’s so much ingrained disrespect for the liberal arts, and such a schizophrenic set of attitudes about quantification within them. Oftentimes, it feels like you just can’t win: your work isn’t serious if it doesn’t involve numbers, but if you incorporate numbers into a subject they see as unworthy, or in a way they see as unworthy, that’s ridiculous, too. In such a context, it doesn’t surprise me that many humanities people simply wash their hands of the whole thing, and say “they aren’t going to respect me anyway, so why not just do my own thing?”

I want to stress that a majority of the STEM-oriented people I’ve worked with (and in the last couple years in particular, I’ve worked with many) have been friendly, approachable, and generous. For the past two years I’ve worked with a series of international graduate students, almost all of them in the STEM fields, and the collaboration has been among the most meaningful and satisfying elements of my recent life. My statistics professor and those I’ve worked with in the statistical consulting service here, as  well as my private R tutor, have been patient and kind with me. There have been a few people in the STEM disciplines here, and many more who claim to be online, who have been… less generous. That’s life, I guess. People have weird ownership issues over this stuff.

One thing I’ve gained: I am much less subject to mathematical intimidation than I once was. A lot of people (and it’s far from just in academia) will just try to wing claims by you by throwing in some numbers and statistical terminology. What I’ve developed is a lack of fear of really interpreting those claims, thinking them over, and performing a critical review. I won’t always know if they’re right or wrong, but I won’t fear looking deeper and saying if I’ve found problems. I’ve gained the confidence to inquire more deeply, and the framework for understanding how to l earn more.

Sorry this is so scattered. I’ve tried and failed to write a post here a thousand times about the humanities, numbers, and the future, but I am apparently incapable of writing coherently on the subject. I guess I will just say this: I write about statistics and research methods here not because I know everything but because I so certainly don’t. I am feeling my way through, thinking my way through, and day by day getting a little better. That has never meant that I have left the liberal arts behind, or that I have come to embrace a purely quantitative or positivist way of knowing. I’ve just had these questions, and have wanted to find  answers to them, and I think that I can do a service by talking through some of these issues from the standpoint of someone who has been growing and has more growing to do.

Neil Degrasse Tyson sometimes says that his shelves are filled with books from history and literature and the arts, but that professors in history and literature and the arts usually don’t have books on the sciences. I’m not sure that’s as true as he thinks; he’s obviously a brilliant man and a great science communicator, but he sometimes seems incurious about people. But either way: I am determined not to be one of those types, and would be moved by curiosity to learn more even without a philosophical commitment to doing so. I am convinced that there must be a way to pursue these interests without denigrating or sidelining the traditional values and methods of the humanities, and without suggesting that only numbers can tell us more about our world. My goal is simply to learn more, to gather more expertise where I can, and in those fields where I am sure to remain an  amateur, to be a better amateur.