January 12, 2022
Dr. Jonathan Cheng, who graduated last year with a doctorate in literature, is using his degree in surprising and exciting ways in the Research and Development division of Apple.
At UNL, Dr. Cheng earned a Certificate in Nineteenth Century Studies and a Certificate in the Digital Humanities (DH), an interdisciplinary field in which computational tools are applied to studies in the humanities. For his dissertation, he used digital tools to search literary texts dating back to the nineteenth century, focusing on gender and characterization, and then investigated the gendered ways in which male and female characters’ actions have been described differently.
Although Dr. Cheng studied computer science as an undergraduate, and his DH certification—along with the Digital Scholarship Incubator and his work at the Center for Digital Research in the Humanities—helped to prepare him for his new role, he also sees connections to his literary studies, in general—to analysis, argumentation, and considering context, among other skills.
These experiences not only equipped him for an exciting career in Alt Ac, but also set him on a path of taking on personally fulfilling, challenging projects, even outside of work, as our conversation revealed.
What projects are you currently most excited about?
What they have us do (at Apple) is work on a bunch of different projects in different domains. One day, I might be working on video recommendations; another day, the apps team needs another pair of hands, so I’ll jump into that. I’m trying to talk about this in a roundabout way because I can’t talk about the things I’m actually working on (for confidentiality reasons).
The thing I’m always super excited about is that everyone on my team has a Ph.D., and I walk into the room, and I feel like … everyone here is way smarter than me, and I’m always super excited to talk to them about what we’re working on. Every Friday, we meet up, and (have conversations like) “I’m trying to implement this new approach, and this is what I’m looking for…” That’s what I’m most excited about (at work). All these smart people are in a room, and you get a ton of great feedback!
… Some of the research that I’m really getting into, I’m not doing for work.
Basically, my time at UNL helped me learn certain kinds of programming that help me do this new form of modeling and machine learning, where you (input phrases) and you learn what art style is best represented by (that text). So… (sharing the screen to show an example) someone put “King of the podcast summons his subjects,” and it auto-generates all these paintings! … It’s taking all the machine learning that we use for auto-translation—(translating from) English to Spanish, English to French, et cetera—but we use that to go from English to paintings. And I’ve been obsessed with this recently …
Now that I feel more comfortable with flexing the math, science, art, and literature parts of my brain as well, I feel like this kind of (project) allows me to fully indulge in all that—and do it not just for (work). That’s probably what I’m most excited about right now! People creating art with these very computer science sorts of approaches.
I do think that someone will try to monetize this eventually, so we’ll see what happens then. The thing I’m a little leery about is that the models are so good that instead of paying artists, you might just end up with a model that does it for free.
Is this illustration programming something that a lay person could do?
Yeah, what’s really cool about this community is that people have put up more and more demos that people can play around with. So for the group that I do anime face painting with (using illustration programming), there’s a tutorial where you just drop an image and you click submit, and it does a rendition. For the actual training, the model, and getting a computer to learn this, that, I think, takes, even for me and a group of people like me, a lot of work. But yeah, there’s a couple demos …
Basically, you’re saying (to the program), “Take all these combinations of texts that describe an image that exist all over the internet,” and you’re scraping whatever you can find from whatever source you can find. And you’re saying, “Take these billions of examples that we have of image-plus-text and try to learn the relationship between these two things.” And now by the end of this, hopefully you have a program you can develop where you can type in some input text, and it’ll generate something, and it’ll see if you like it or not …
It takes a lot of resources to do this kind of work, so what’s really cool is there are all these random online communities where people meet up and pool our computers together to do all this …
I’ve been working on some anime renditions of faces … What’s cool is that because your (computer) model has learned from what you’ve scraped from the internet, you can say, “Paint a dragon breathing fire (in the style of) a mixture of this painter and this painter,” and the model will do the best to approximate a style for this text prompt between those painters. I was very excited about this because it did a very good approximation of where these two painters are …
And people have learned to treat video as many images in a row, so you have an example of “Translate this scene from The Shining into a Gustave Dore painting,” and you can animate this … I’m definitely enjoying it!
What connections do you see between what you do now and what you did at UNL, with your dissertation and in literary studies?
All the good intuitions! Like, “OK, we think this model can perform these sorts of tasks, and we think it can do it this percent of the time. But what’s the larger significance of that? And how do we interpret the historical significance of that?”
And: “What does this evidence say or not say? How is it persuasive in these ways, but not in these ways?” I can say, “The model can do this, and I can elaborate (on) that more fully,” but I can also say, “I wouldn’t use it for these things.”
I would say that most English Ph.D.’s are really good at making us really humble about our claims. I feel like that’s generally one of the main instincts that I got out of the English Ph.D. Being able to say, “This is true; that might be true; this is unclear.” … (My educational background) definitely makes me more comfortable about what promises we give and don’t give to our various customers … English analysis and computer science analysis, on a high level, there are some similarities, (although) they’re also very different on a lot of levels.
What was your favorite class or, if you can think of one, your favorite reading? And as you think back to those classes, was there a professor who was especially influential on you?
My favorite class, I think, was probably Pete (Capuano)’s “History of Science” course on nineteenth-century science. That class because, historically, it’s just very fascinating! It was very exciting reading Natural Theology and Principles of Geology. … This hoodie (I’m wearing) has “Principles of Geology” on it, actually! … But reading people who were saying “I think the world might be older than 4,000 years,” and everyone is like, “Impossible!” … It’s just historically fun stuff to read.
Also, any of Laura (White)’s classes are my favorite … Laura, as an academic force of nature, (was influential) … She’s very good at managing a lot of complicated research ideas, but at the end of the day, she’s good at saying, “We don’t do that” or “We can do that.” And having a sense of fun about it all, while also being very rigorous about every detail. Who had the most impact? Probably Laura.
Pete’s “History of Science” course and Laura’s “Gothic and Mystery” course, I think, were probably my two favorite courses.
What are your favorite memories of extracurricular activities or places you spent time at in Lincoln?
Did we have time for extracurricular activities? (laughing) With EGSA, it was very good to commiserate with everyone at the end of the semester (at the “Tea and Sympathy” event). That was definitely a highlight!
In terms of favorite places on campus or in town, I do really get nostalgic for the various Mill (coffee shops). I’m just nostalgic for (doing my work) in coffee shops, in general. I think the pandemic has taken that away from me. I love sitting and working in public places, and also you can hear all the gossip when you’re working in those places.
The Mill-plus-Ivana Cone was kind of a package deal for me, because I would work all day and get coffee, and then pick up an ice cream. One day of writing, you get one scoop of ice cream!
For getting work done, the Mill is my favorite place. If you’re wanting to transition from work into drinks, the Coffee House. Wanting to see local Lincoln life? Meadowlark Coffee Shop. Very interesting snapshot into Lincoln life. I highly recommend it for every graduate student, especially on Tuesday nights. They used to have an open mic night there, and … there was this guy who wrote his own piano (music) and lyrics, but it was all just various complaints about the youth in Lincoln or about the government (laughing) … It was very fun to watch.
After Lincoln, I moved to Chicago for a year because we worked fully remotely … and I hadn’t seen my undergraduate friends for some time … and they’re all in Chicago. After that, in August, I moved to Berkeley, and from there, I was looking up apartments to live at … I’m very excited because my roommate and I got this place for a steal … I was telling my mom, “We got such a great deal for such a great place!” and my mom was like, “How much was it for?” … (laughing) Especially since my place in Lincoln was 500 bucks or something like that, the adjustment (of moving from Lincoln to California) was a little weird.
I am a little nostalgic about working in Lincoln … Just a little. California is great.
There is this thing about Lincoln: Sunsets are weirdly very pretty. The sky gets very pink, and they get very pinkish orange on some days, and it is very pretty. Here it is perpetually pretty grey, which I also really like, but I do miss the Plains sunsets.