CDS student, Aditya Singhal, awarded Dean’s Undergraduate Research Fund grant
CDS students at all levels are constantly finding new ways to innovate and make an impact with their research. And CDS undergrad student, Aditya Singhal (Founder of the NYU Data Science Club), is one such student.. He was recently given a Dean’s Undergraduate Research Fund grant, a highly selective award for, as the name suggests, undergraduate researchers. Past receipts have conducted research in neural science, anthropology and even journalism. Colton Laferriere, from the CDS social media team, spoke to Aditya about the how the project came to be and how he hopes his research will make an impact in the world:
The following has been edited for clarity.
CL: Can you tell me more about the project you’re working on?
AS: I went to something called HCOMP2020. It was my first conference ever. At the conference I was listening to some professors talk to each other and ask questions, which I did not understand, but one of them was saying something that a lot of people had questions about and everyone was excited about. And that professor was from NYU and because I was from NYU I was like “You know what? I’m gonna make a connection with this person.” I texted him in the conference Slack and said, “I’m an NYU student, can we arrange a meeting?” He agreed and I told him about this thing that I was trying to do.
Now, the thing is, I have been journaling for the last five and a half years and I have more than 3600 journal entries. That’s a lot of natural language data. I thought, okay, what are some cool things that I can find out about myself about how I feel, what I feel, and when I feel it? I had done a project over the summer — some of basic natural language processing stuff.
So I told this to the professor from the conference, saying, “Okay, this is all the stuff that I’m interested in and now I’m trying to use GPT to try to get a machine to talk like me.” And then he said, “Oh, you know what, that’s very cool because I have done something exactly like this already.” Within five minutes we figured out we were trying to do the same thing. He told me “I already have some students and I’m doing this cool chatbot development. We recently got access to this data set that we can train a chatbot on. Let’s get you involved and we can apply for the DURF together.”
In the natural language processing/chatbot creation world you can finetune existing models that can generate human-like speech on certain data sets which have a predominant emotion. So basically what we’re trying to do is take one of these existing models that was already trained on being empathetic and fine tune it on this new data set to become more trustworthy to humans.
CL: How does grant funding help you reach your project goals?
AS: The grant itself gives me enough credibility to be able to use NYU’s HPC. Which you need to be able to do heavy model making such as the one that we’re doing. I’ll use the grant money for MTurk, meaning Mechanical Turk. For anyone unfamiliar, it’s an Amazon service that you can use to basically crowdsource your work. And with this chatbot when we want to test how much more empathetic it is.
CL: It sounds like also just having this sort of award helps you sort of get access to resources within the university as well, which is pretty cool.
AS: Exactly so. It’s great! I can freely talk about my project and people know sort of what I’m talking about. I can sound more legit when I need to access the library or when I want to access NYU HPC. And professors are more inclined to help you and think that the work you’re doing is worth their time if they know that you’ve gotten some sort of grant for it.
CL: The project sounds really interesting. I don’t know how much you are at liberty to talk about the project didn’t detail, but how would you like to see the results of this research working in the real world?
AS: Very good question. The aim is to try to see if it’s possible and to try to see if the chatbot is quantifiably more empathetic. The implications would be, for example, in Telehealth. It’s an emerging field that’s extremely important and would be much more efficient with more empathetic chatbots. If patients don’t trust chatbots, if they don’t feel comfortable enough to share private information with them, even though everything is confidential, then things don’t work. But being able to teach machines how to interact like human beings — there’s like infinite applications to it. We’re just a tiny drop in the ocean of NLP, trying to make our chatbot more empathetic.
We at CDS would like to give Aditya a huge congratulations on his award and wish him the best of luck on his research. If you would like to follow the project or to get in touch with Aditya, you can find him on his Linkedin page.