AI pioneers want bots to switch human teachers – here's why that's unlikely

OpenAI co-founder Andrej Karpathy imagines a world In this vision, bots could be made into material experts using artificial intelligence, with “great passion, great teaching skills, infinite patience, and fluent in all of the world's languages.” In this vision, the bots could “provide personalized tutoring to all 8 billion of us on demand.”

The embodiment of this concept is his latest project, EurekaLabsThis is just the newest high-profile example of how technology entrepreneurs are using AI to revolutionize education.

Karpathy believes that AI can solve a long-standing challenge: Lack of excellent teachers who’re also material experts.

And he just isn’t alone. CEO of OpenAI Sam AltmanKhan Academy CEO Sal KhanVenture capitalists Marc Andreessen and computer scientist on the University of California, Berkeley Stuart Russell also dream that bots Tutors on call, Career advisor and possibly even replacements for human teachers.

As a researcher specializing in AI and other recent writing technologiesI actually have seen many cases where high-tech “solutions” to teaching problems have fizzled out. AI can definitely improve elements of education, but experience shows that bots are unlikely to be effective replacements for humans. That's because students have long shown resistance to machines, regardless of how sophisticated, and have a natural preference for connecting with and being inspired by other humans.

The costly challenge of teaching the masses to write down

As director of the English Composition Program on the University of Pittsburgh, I serve about 7,000 students every year. Programs like mine have long struggled with how one can efficiently and effectively teach writing to so many individuals without delay.

The best solution is to keep up class size. for a maximum of 15 students. Studies show that Students learn to write down higher in smaller classes because they’re more engaged.

However, smaller classes require more teachers, and this may expensive for varsity districts and universities.

Reviving dead scientists

Enter AI. Imagine Karpathy setsthat the nice theoretical physicist Richard Feynmanwho has been dead for over 35 years, might be brought back to life as a bot to tutor students.

For Karpathy, an excellent learning experience can be working through physics material “with Feynman there with you each step of the way in which.” Feynman, known for his accessible way of presenting theoretical physics, could work with an infinite number of scholars without delay.

In this vision, human teachers still design teaching materials, but are assisted by an AI teaching assistant. This teacher-AI team “could offer a whole curriculum on a standard platform,” Karpathy wrote. “If we're successful, it is going to be easy for anyone to learn anything,” whether it's many individuals learning one subject or one person learning many subjects.

Other efforts to personalize learning fall short

But technologies for private learning are nothing recent. Exactly 100 years ago, on the American Psychological Association meeting in 1924, inventor Sidney Pressey introduced an “automatic teacher.” produced from typewriter parts which contained multiple-choice questions.

In the Nineteen Fifties, the psychologist BF Skinner “teaching machines.” If a student answered an issue accurately, the machine would move on and ask for the subsequent step of the issue. If not, the coed would stay at that step of the issue until they solved it.

In each cases, students received positive feedback for his or her correct answers. This gave them confidence and expertise. The problem was that students didn't learn much – additionally they found these non-human approaches boring, as educational creator Audrey Watters explains in “Teaching machines.”

In recent times, the world of education saw the rise and fall of “Massive Open Online Courses” or MOOC. These courses, which provided videos and tests, were recognized by the New York Times and others for his or her Promise of democratization of education. Again, the scholars lost interest and dropped out.

Other web-based initiatives have emerged, including course platforms corresponding to Coursera and Outlier. But the identical problem stays: there is no such thing as a real interactivity to maintain students engaged. One of probably the most recent victims of online learning was 2U, which acquired leading MOOC company edX in 2021 and filed for bankruptcy in July 2024 to restructure its 945 million US dollars debt burdenThe perpetrator: falling demand for services.

Now comes the proliferation of AI-powered platforms. Khanmigo uses AI tutors like Sal Khan writes in his latest book“to personalize and customize coaching and adapt to the person needs of the person while supporting our learners of their work.”

The educational publisher Pearson integrates AI into its teaching materialsMore than 1,000 universities adopt these materials for autumn 2024.

AI in education just isn’t just coming, it’s already here. The query is how effective it is going to be.

Disadvantages of AI learning

Some technology leaders are confident that bots can personalize instruction and replace human teachers and tutors, but they’re more likely to face the identical problem as those earlier attempts: Students may not prefer it.

There are vital reasons for this. Students are unlikely to be as inspired and enthusiastic as they’re with a live teacher. Students in crisis situations reach out to trusted adults corresponding to teachers and coaches for help. Would they do the identical with a bot? And what would the bot do in the event that they did? We don't know yet.

Lack of information protection and security may also be a deterrent. These platforms collect vast amounts of knowledge about students and their academic achievements, which could be misused or sold. Legislation can try to stop this, but some popular platforms are based in Chinaoutside the scope of U.S. law.

Finally, there are still concerns even when AI tutors and teachers change into popular. If a bot teaches tens of millions of scholars concurrently, diversity of thought might be lost. Where does originality come from if everyone gets the identical lessons, especially if “academic success” relies on repeating what the AI ​​teacher says?

The idea of ​​an AI tutor in everyone's pocket sounds exciting. I'd like to learn physics from Richard Feynman, or writing from Maya Angelou, or astronomy from Carl Sagan. But history warns us to be cautious and pay close attention as to if students are literally learning anything. The guarantees of personalized learning are not any guarantee of positive outcomes.