metamorworks - stock.adobe.com

Google Learn About AI tool could fail to hit the target

While the new tool is geared toward professors, students and researchers, it may still fail to stand up to popular chatbots like ChatGPT. The tool also remains prone to hallucinations.

While Google's new Learn About tool targets students and academics, the cloud provider could find it difficult to get those in the academic world to use it.

Introduced Monday, Learn About is built on the Google LearnLM AI model. LearnLM is part of a family of AI models that Google rolled out earlier this year. The model is grounded in educational research and tailored to help people learn.

Google partnered with Teachers College, Columbia University; Arizona State University; New York University Tisch School of the Arts; and Khan Academy to provide feedback and improve LearnLM.

Learn About differs from Google Gemini because its answers are based on academic forums, research papers, documents and textbooks, rather than internet resources.

The benefits

"If this works well, it can replace teachers, professors and other educators," said Constellation Research analyst Andy Thurai, regarding Learn About. "It can also become a source for online education where the current tools are very outdated and static."

This is an interesting area that no other LLM player is seriously contesting.
Andy ThuraiAnalyst, Constellation Research

He added that employers could use Learn About to train employees on the job and help with career advancement. This would pit the tool against other similar systems, such as Skillsoft.

"This is an interesting area that no other LLM player is seriously contesting," Thurai continued. He added that the only competition is ChatGPT Edu, which was trained on the GPT-4o model.

With Learn About, Google can claim "user experience and knowledge formation advances over existing popular AI chatbots such as ChatGPT and Gemini," said Futurum Group analyst Ron Westfall.

"It is part of the AI ecosystem exploring ways to popularize alternatives to OpenAI," Westfall said. "Google's motives are clear, as its Gemini and Learn About offerings are already direct competitors to the Microsoft-OpenAI alliance."

The targeted market

However, whether Google can capture its targeted audience with Learn About is yet to be seen.

"I don't see this as a game changer," said Chirag Shah, a professor at the University of Washington Information School. He added that he's been having trouble using it, and even if some of the problems with the tool are solved, students are still more likely to use ChatGPT or Gemini.

Sarah Kreps, a professor at Cornell University, agreed.

"What's helpful about Learn is that it obviates the need to comb through pages of Google search results to come up with "the one answer,'" Kreps said. "The problem is that it's still doing too much editorializing that users found problematic with Gemini."

She added that when she asked ChatGPT during a class on international law to make an argument that "international law is not just the strong doing what they want and the weak doing what they must," ChatGPT responded with six bullet point arguments.

When she later asked Google to Learn About the same prompt, part of the response was: "The quote, 'The strong do what they will, and the weak suffer what they must,' is a very cynical view."

"Why use language like 'a very cynical view,' versus the ChatGPT position that was more agnostic and responsive to my request to make a counterargument?" Kreps argued.

She continued that when she asked both AI chatbots for a different prompt about author Dante's journey in the classic work of literature The Divine Comedy, ChatGPT gave succinct responses. At the same time, Learn About provided sections that would enable her to learn and click to explore more.

"The problem at this point may be that users are pretty wedded to the AI engine they use if they use one," Kreps continued. "A competitor has to be demonstrably better to lure people away from the features that they know and like. This platform, in my initial tests, was either too judgy or too trite, and will have to figure out how to do something better than the incumbents to knock them off their perches."

Moreover, since Learn About is supposed to address AI model hallucinations, Google has a responsibility to ensure that the chatbot is as accurate as possible, Shah said.

"It can still connect the wrong dots," he said. "If their target is the education market -- students, teachers, tutors -- that has to come with a huge responsibility. For those of us who are educators and who have students, this is something that matters. I don't see much or anything regarding that."

Google is not the only cloud and AI technology provider targeting the academic community.

Amazon on Nov. 12 revealed a $110 million investment in university-led research in generative AI. Through the Build on Trainium program, researchers can develop new AI architectures, machine learning libraries and performance optimizations for AWS.

Esther Ajao is a TechTarget Editorial news writer and podcaster covering artificial intelligence software and systems.

Dig Deeper on AI technologies