Bard is a new artificial intelligence (AI) large language model from Google that is likely to have important implications for education. The much-hyped tool was unrolled publicly on Tuesday, March 22, and is widely seen as a competitor to ChatGPT and BingGPT, which both utilize technology developed by the Microsoft-backed developer OpenAI.
Anyone with a Google account can join the waitlist to get access to Bard. This is the latest chapter in the AI wars between Google and Microsoft, as well as other tech players. It comes a week after OpenAI released GPT-4 for paid subscribers to ChatGPT and for some select education purposes, including Khan Academy’s new learning assistant, Khanmigo.
Like ChatGPT, Bard is not designed specifically for education but will almost certainly have applications for teachers and students. As a writer and educator, I see this as another sign – if one was necessary – that for better or worse, and probably a little bit of both, AI text and technology will be a part of teaching going forward.
Many educators and potential students now have access to another powerful AI tool. I’m still on the waitlist to use Bard, but Ron Samul, an author and Assistant Director of Thames at Mitchell College, shares his impressions from his initial time with the technology.
“It can be a tool that can help students and kind of give them preliminary thoughts before they start writing, but it also can be very easy to just decide to plagiarize,” says Samul, who mentors in the MFA program in Creative & Professional Writing at Western Connecticut State University, where I also am a mentor.
What is Google Bard?
Google Bard is an AI text generator or chatbot that will work with Google’s search engine and provide text-based answers to user questions. Though it was first announced in early February 2023, it will be powered by Google’s “Language Model for Dialogue Applications,” or (LaMDA) system, which Google announced in 2021. You might remember LaMDA making headlines in June 2022 when a Google engineer was placed on paid administrative leave after he came to believe that LaMDA was sentient. (These claims of ChatBot sentience have been widely refuted by the scientific community.)
Early reviews of Bard indicate it is less likely to say outlandish or creepy things as was the case initially with BingGPT, which could famously go far off the rails when prompted. However, the technology is still far from perfect. When Google introduced Bard, one of the examples featured in the demo had an error about the James Webb Space Telescope that many believe caused the price of Google’s stock to plummet.
“Things that have text-based responses, like questions about books or summaries of movies and things like that, all come up really pretty focused and on point,” Samul says. However, when it’s a question requiring more nuance or one for which the answer isn’t available on the first few websites you’d find on Google, Bard currently struggles. For instance, when Samul asked Bard who killed Marlyn Monroe, the AI couldn’t muster a response.
What Are Some Exciting Ways Bard Can Be Used for Education?
When used as a more efficient search tool, Samul sees potential for how Bard is able to gather sources and provide ways to get started on a topic. He also believes it can help students visualize and understand the material by creating outlines, particularly for those with learning disabilities, with which he works frequently.
For instance, he asked it to create an outline of Macbeth. “The outline for Macbeth was pretty general, but it did help me understand the order of events,” he says. “So I thought that was interesting, especially for students who struggle with their organization.”
He also envisions it potentially taking on the role of a tutor or advisor. “It helps with a student who maybe needs prompting for different ways they can think about something. They might need an example or they might need a starter sentence,” he says.
What Are Some Ways Bard Can Be Used to Cheat?
ChatGPT can already write basic student essays but if Bard ultimately becomes a standard feature of Google search, as Google has indicated may happen, the temptation to use the technology for unethical shortcuts could be stronger.
“It really does come down to how teachers are going to have to be more selective in the way they create their questions and the way they create prompts for people to write assignments,” Samul says. “It can’t be just, ‘How did slavery change in the Civil War?’ They'll have to really come up with very specific questions.”
Tools are already available, with more in development, to detect AI-generated content, however, these are not foolproof yet, much like existing plagiarism detectors. That’s why ethics education is so important. “It's always gonna be abused, just like plagiarism is always an access point for students without these tools,” Samul says.
How Can You Sign Up For The Bard Waitlist?
Just go to the waitlist and enter your Google email. Wait times can vary. Samul got access shortly after requesting to be on the waitlist that morning. I haven’t yet received access more than 24 hours after my request was made and I’m not holding my breath.
I did get an immediate email from Google saying I’d been added to the waitlist with a short poem composed for me by Bard. The poem: “May your day be bright/Your Mood be light/ and your heart filled with delight.”
Not exactly work worthy of a true bard but not terrible either.
Will Bard Destroy the World?
Samul posed this question to arguably the top authority on the topic, Bard. “I asked it, ‘Is Bard, the new AI, going to destroy the world?’ And it didn't say, ‘I can't answer this question.’ It said, ‘I assure you that's not going to happen.’”
I felt better until I remembered that these AI language models are sometimes known to lie.