Artificial
intelligence, the technology that enables robots to respond to their
environment like humans, may also help improve how college students interact
with textbooks. A small project conducted in 2012 showed students who studied
with an AI-enhanced digital textbook earned higher marks. A larger pilot is now
in development.
As
described in the fall 2013 edition of the Association for the Advancement ofArtificial Intelligence’s AI Magazine,
the Inquire Project created an intelligent app for selected chapters of the
popular Campbell Biology text used by
many introductory biology courses.
The app
allowed students to key in free-form questions about the content while they
were studying on an iPad. They could also tap the screen to access detailed
concept summary pages, pop-up definitions, and follow-up questions as well as
highlight text to create “note cards” with related questions. The app was
developed at the Artificial Intelligence Center at SRI International, with
funding from Vulcan Inc., a company owned by Microsoft co-founder Paul Allen.
In an
evaluation conducted with 72 community college students, one third of the group
used the traditional print version of Campbell
Biology while another third studied from the AI-enhanced digital version of
the book. The remaining third was assigned to the regular digital version,
which permitted basic highlighting and annotations but didn’t have the AI
extras. All of the students were asked to read certain chapters for an hour,
spend 90 minutes on homework, and then take a 20-minute quiz.
Quiz scores
averaged 88 for the AI group but only 81 for the print textbook and just 75 for
the unenhanced digital textbook. Homework scores were similar: 81 for AI, 71
for print, and 74 for “plain” digital.
The
conclusion? Being able to “ask” questions and view summaries right on the page
they were reading apparently aided students’ comprehension and retention of the
material. The 72 participants posed a total of 520 questions during two and a
half hours of reading and homework. Of those, 194 questions were unique but
only 59 were asked by more than one student, indicating that different students
needed clarification of different textbook content.
