Will chatbots teach your kids anything?

Sal Khan, the CEO of Khan Academy, gave a rousing TED Talk last spring in which he predicted that AI chatbots would soon revolutionize education.

“We are on the cusp of using AI for probably the greatest positive transformation education has ever seen,” stated Mr. Khan, whose nonprofit education group has delivered online classes to millions of students. “And the way we’re going to do that is by giving every student on the planet an artificially intelligent but amazing personal teacher.”

Videos of Mr Khan’s bot speech collected millions of views. Soon, prominent tech executives, including Google CEO Sundar Pichai, began making similar education predictions.

Mr. Khan’s vision for teaching bots tapped into a decades-old Silicon Valley dream: automated education platforms that instantly tailor lessons to each student. Proponents argue that developing such systems would help close achievement gaps in schools by giving children relevant, individualized instruction faster and more efficiently than human teachers ever could.

In pursuit of such ideals, tech companies and philanthropists over the years have pushed schools to buy a laptop for every child, championed video tutorial platforms and funded learning apps that tailor students’ lessons. Some online math And literacy interventions to have positive effects reported. But a lot educational technology efforts to have It has not been shown to significantly close gaps in academic achievement or improving student outcomes, such as high school graduation rates.

Now the spread of generative AI tools like ChatGPT, which can provide answers to biology questions and produce human-sounding book reports, is renewing enthusiasm for automated instruction — even as critics warn that there is not yet evidence to support the idea that tutoring to bots will. transform education for the better.

Online learning platforms such as Khan Academy and Duolingo have introduced AI chatbot teachers based on GPT-4. This is a large language model, developed by OpenAI, that is trained on huge databases of texts and can generate answers based on user questions.

And some tech executives foresee that over time, bot teachers will be able to respond to and inspire individual students just like beloved human teachers.

“Imagine if you could give such a teacher for free to any student 24/7, whenever he wants,” Greg Brockmanthe president of OpenAI said last summer on an episode of the Podcast ‘Possible’. (The podcast is co-hosted by Reid Hoffman, an early investor in OpenAI.) “It’s still a little bit science fiction,” Mr. Brockman added, “but it’s a lot less science fiction than it used to be.”

The White House seems sold. In a recent one executive order On artificial intelligence, President Biden directed the administration to “shape AI’s potential to transform education by creating resources to support teachers in deploying AI-enabled education tools, such as personalized tutoring in schools,” according to a White House fact sheet.

Still, some education researchers say schools should be wary of the hype surrounding AI-enabled education.

First, they point out that AI chatbots liberally make things up and can give students false information. By making AI tools a mainstay of education, unreliable sources can be elevated to authority in the classroom. Critics also say AI systems can be biased and are often opaque, leaving teachers and students unable to understand exactly how chatbots come up with their responses.

In fact, generative AI tools may prove to have harmful or “degenerative” effects on student learning Ben Williamsona Chancellor’s Fellow at the University of Edinburgh’s Center for Research in Digital Education.

“There is a rush to proclaim the authority and usefulness of these types of chatbot interfaces and the underlying language models that power them,” said Dr. Williamson. “But the evidence that AI chatbots can achieve these effects does not yet exist.”

Another concern: the hype over unproven AI chatbot teachers could detract from more traditional, human-centered interventions – such as universal access to kindergarten – that has been proven to increase student graduation rates and college attendance.

There are also privacy and intellectual property issues. Many large language models are trained on enormous databases of texts taken from the Internet, without compensating the creators. That could be a problem for unionized teachers concerned about fair labor compensation. (The New York Times recently sued OpenAI and Microsoft over this issue.)

There are also concerns that some AI companies are using the material teachers input, or the comments students make, for their own business purposes, such as improving their chatbots.

Randi Weingarten, president of the American Federation of Teachers, which has more than 1.7 million members, said her union is working with Congress on regulations to ensure AI tools are fair and safe.

“Educators use educational technology every day, and they want more control over how this technology is deployed in classrooms,” said Ms. Weingarten. “The goal here is to promote the potential of AI and guard against its serious risks.”

This is hardly the first time that education reformers have championed computerized learning tools. In the 1960s, proponents predicted that mechanical and electronic devices called ‘teaching machines” – which were programmed to ask students questions about topics such as spelling or math – would revolutionize education.

Popular Mechanics captured the spirit of the times in an October 1961 article headlined: “Will Robots Teach Your Children Anything?” It described a “wave of experimental machine learning” sweeping schools across the United States, where students worked independently and entered answers into the machines at their own pace.

The article also warned that the newfangled machines raised some “profound” questions for educators and children. Would the teacher, the article asked, become “just a glorified babysitter”? And: “What does machine learning do to students’ critical thinking?”

The teaching machines were cumbersome and didactic and, in the short term, proved to be a classroom sensation, both overhyped and overfeared. The rollout of new AI education bots followed a similar story of potential education transformation and harm.

However, unlike the old teaching machines of the 20th century, AI chatbots seem improvisational. They generate immediate responses to individual students in conversational language. That means they can be fun, immersive and engaging.

Some enthusiasts envision AI tutoring bots becoming study buddies that students can consult quietly and without shame. If schools were to adopt such tools widely, they could profoundly change the way children learn.

That has inspired a number of former Big Tech executives to make the transition to education. Jerome Pesenti, former vice president of AI at Meta, recently founded a tutoring service called Hissing AI The app’s AI chatbot uses a multiple-choice format to help students solve math and science questions.

And Jared Grusd, former Chief Strategy Officer at social media company Snap, co-founded a writing startup called Ethical. The app’s AI chatbot can help students organize and structure essays and give them feedback on their writing.

Mr. Khan is one of the most visible advocates of teaching bots. Khan Academy last year introduced an AI chatbot called Khanmigo, specifically for school use. It is designed to help students think about problems in math and other subjects, not to do their school work for them.

The system also stores conversations students have with Khanmigo so teachers can review them. And the site clearly warns users: “Khanmigo sometimes makes mistakes.” Schools inside Indiana, New Jersey and other states are now testing the chatbot teacher.

Mr. Khan’s vision for teaching bots can be traced back in part to popular science fiction books like “The Diamond Age”, a cyberpunk novel by Neal Stephenson. In that novel, an imaginary tablet-like device can teach a young orphan exactly what she needs to know at exactly the right time – in part because it can instantly analyze her voice, facial expression, and environment.

Mr Khan predicted that teaching bots like Khanmigo could do something similar in about five years, with privacy and security braces.

“The AI ​​can just look at the student’s facial expression and say, ‘Hey, I think you’re a little distracted right now.’ Let us focus on this,” Mr Khan said.