New AI could help write your next textbook, using Penn State technology
3D Render of a Robot writing with a pen
Penn State has announced the launch of BBookX—new technology they developed that works with faculty to use artificial intelligence to build textbooks from open resources. The software, which was created in August, works to create personalized textbooks by extracting open source information from the Web, based on user input.
“It’s an overlap between AI machine learning and education resources,” said C. Lee Giles, David Reese Professor in the College of Information Sciences and Technology. “These are two fields that don’t overlap often.”
Giles, who has a background in AI research, worked with staff from Teaching and Learning with Technology (TLT) to create BBookX. He first came up with the idea when he was teaching Information Retrieval and Search Engines and wanted to build course material for the class. “I bet we could automate some of this,” Giles thought. So he began to test it out.
BBookX: An Automatic Book Creation Framework
Image: C. Lee Giles
The software, still in prototype mode, works by allowing users to enter information into the program and then uses matching algorithms to retrieve articles based on the search terms. Right now, it is a search engine built on top of Wikipedia, although any open-source database could work. “It’s more assembling the textbook than doing the writing,” said Giles. “It’s done with your guidance—it’s a collaborative process.”
There are many advantages to using this kind of software:
Cost savings: “In some community colleges,” Giles said, “the textbook costs more than the course.”
Personalization: Since this is user-generated material, the textbooks can be tailored to the interests of the user.
Learning experience: The act of creating a personalized textbook is also an important educational experience for students, who are forced to map out an outline and evaluate the results.
Supplement material: This tool can be used to add new information into an already-existing curriculum.
Discovery: By using matching terms, BBookX can offer material that even experts in a field may not have already discovered, shedding new insight into a topic.
Staying up-to-date: By drawing on open-source material, this software is great for ensuring that material remains relevant and up-to-date. “In computer science,” Giles said, “Things change quickly. A lot of textbooks quickly go out of date.”
The system isn’t perfect, and it’s not necessarily suited equally for all fields of study. “It’ll be more relevant for fields with digital information available,” said Giles. “You can’t search for information that’s not digitized.” And he’s still tweaking the algorithm “to build in temporal order and continuity.”
Giles wants as many professors and students to use it as possible, to provide feedback that he can use to improve the tool. “This is a project where you learn a lot from your users,” he said. “The problem with using this is evaluation. We would have a hard time evaluating it for chemistry, since we’re not chemists. The best evaluators need a domain expertise to do the evaluation.”
While there’s still room to improve, BBookX is an important step in innovating education. “To our knowledge,” Giles said, “this is the first semi-automated process available.” When asked if he sees fully-automated textbook writing happening at some point in the future, Giles responded that “it’s probably pretty far off.” Writing by automation, however, is not all fantasy-already, there are sports articles being written by robots. And it seems more than likely that this will continue to spread to other types of writing, as well.
BBookX is still in testing mode, and Penn State students and professors are just starting to use it, offering suggestions for improving the system. Giles will continue to update the algorithm, and will use the software in his upcoming course this spring. Look for more coverage on TechRepublic on BBookX—what is and isn’t working when it comes to AI-assisted writing—later this year.