Arvind Narayanan — Princeton

I'm a professor of computer science at Princeton University, affiliated with the Center for Information Technology Policy.

I study the societal impact of digital technologies, especially AI.

Office: 308 Sherrerd

609-258-9302

arvindn@cs.princeton.edu

Twitter: @random_walker

AI snake oil

I’m writing a book about AI Snake Oil — AI that does not and cannot work — with Sayash Kapoor. Here’s a sneak peek. We're sharing our developing ideas through substack; subscribe here. We are also tackling the machine learning reproducibility crisis.

I first spoke about this topic in 2019; the slides have been widely circulated. I co-taught a related course on limits to prediction; here are the course materials.

Algorithmic amplification on social media

For 2022-23 I’m a visiting senior researcher at Columbia's Knight First Amendment Institute, studying how social media algorithms amplify some speech and suppress others. My capstone essay: Understanding Social Media Recommendation Algorithms.

I'm co-organizing a symposium on algorithmic amplification that will take place April 28/29, 2023, at Columbia University and online. Register here.

Fairness and ethics in computing » see more

I coauthored a textbook on fairness and machine learning, available online. My work was among the first to rigorously show how machine learning reflects racial, gender, and other cultural biases. I've also worked on exposing dark patterns online.

Web privacy » see more

I led Princeton's Web Transparency and Accountability Project. Through large-scale, automated web measurement, we uncovered how companies collect and use our personal information. Our open-source tool OpenWPM has enabled over 100 studies of web privacy.


Cryptocurrencies and blockchains » see more

I led the creation of an online course / textbook on cryptocurrencies which has been used in over 150 courses worldwide.

My main interest these days is helping shape public policy to counter the harms of cryptocurrency.

De-anonymization » see more

I've shown how sensitive information can be inferred from seemingly innocuous "anonymized" data, ranging from browsing histories to genomes. See this primer of the research and this policy piece on what it means for privacy.