Barnard professor tests custom A.I. tutor in first-year writing courses

Benjamin Breyer, a professor in Barnard College’s first-year writing program in Manhattan, has been testing a custom A.I. chatbot called Althea in his sections even though the program generally bans generative A.I. tools such as ChatGPT and Gemini. The program’s policy warns that A.I.

“is often factually wrong, and it is also deeply problematic, perpetuating misogyny and racial and cultural biases,” and the program director, Wendy Schor‑Haim, said she runs screen‑free classes and “has never tried ChatGPT.” Breyer and a former Barnard software developer, Marko Krkeljas, spent thousands of hours building Althea, applied for about $30,000 in grants and additional technical support, and trained the bot on his lectures, course readings and sample student work.

He set its persona as a “tutor at an elite liberal arts college,” had it refuse outright to write assignments, and described it as an interactive workbook that prompts students to improve their answers. Breyer said early testing produced mixed results: after one year students who did not use A.I.

performed better on his exercises, but after refining Althea he reintroduced it and this past semester students who used the bot did better on exercises than those who did not. Other English professors are likewise experimenting: Alexa Alice Joubin built an open‑source teaching bot, and Columbia’s vice dean for A.I.

Key Topics

Tech, Althea, Benjamin Breyer, Barnard College, Generative Ai, Marko Krkeljas