The Gates Foundation announced this week a $1 billion commitment to build and deliver equitable artificial intelligence tools for schools, frontline health care workers, farmers, and other communities. Forty percent of that pledge—roughly $400 million—is earmarked for education, including AI tutoring intended to individualize learning in classrooms across the United States and abroad.

The education allocation flows from a broader U.S. education strategy the foundation laid out in July, which aims to help up to 10 million more Americans earn “credentials of value” by 2045—the year the foundation plans to close. The sunset date puts the entire education strategy on a fixed clock.

The announcement lands at a moment when the value of classroom technology is being questioned. Test scores have dropped among some students over the same period in which computer-based assessments were widely implemented. Meanwhile, research shows that Gen Z has proved less cognitively capable than previous generations ever since the U.S. spent $30 billion to swap textbooks for laptops, according to written testimony by neuroscientist Jared Cooney Horvath presented to the U.S. Senate Committee on Commerce earlier this year.

“Evidence indicates that indiscriminate digital expansion has weakened learning environments rather than strengthened them,” Horvath wrote.

That skepticism is already shaping policy. On Sept. 2, New York City Mayor Zohran Mamdani and Schools Chancellor Kamar Samuels announced what the city called the nation’s broadest moratorium on student-facing generative AI—a one-year pause for students in 2-K through 8th grade, affecting nearly 600,000 students, or roughly two-thirds of the system.

“The tech industry wants us to believe that AI-powered early education is not only inevitable, but necessary,” Mamdani said in a statement. “We do not see it that way.” He added that the city would spend the year studying the technology’s impacts.

Yet major philanthropic organizations such as the Gates Foundation continue to pour money into introducing more technology in schools. And while the foundation’s stated intention is to close gaps in access to AI, not everyone is convinced of how effective the effort will be—or how necessary it is for student success. The stakes are concrete: one of the nation’s largest school systems is pressing pause on student-facing AI to study its effects even as a major philanthropy funds its deployment, setting up two parallel real-world tests of the technology’s place in the classroom.

A Gates Foundation spokesperson told Fortune its approach is built around those very concerns. The tools will be tested in schools serving high shares of low-income students, paired with teacher training, and evaluated by independent researchers before being scaled. The foundation’s own strategy warns that “too many tools have been adopted without evidence they work” and insists that teachers “remain central.”

Teachers’ concerns with widespread AI implementation

Siobhan Casey, who has worked in education for 21 years as both a teacher and in leadership roles, said that while AI has been useful as a starting point for lesson planning and administrative work, it is not an end-all, be-all for learning outcomes.

“I still need to check its accuracy, decide whether it supports the learning objective, and make sure it provides the right level of challenge for my students,” said Casey, principal of the Greenwich Campus of Crimson Global Academy. “AI can support teacher expertise, but it cannot replace it.”

The Gates Foundation spokesperson said a goal of the commitment is to accelerate learning and help students catch up, one example being the use of AI to give students more repetition and practice.

Other teachers are concerned about widespread implementation of AI in schools. Because AI is a relatively new tool, schools also need the resources to train staff on how best to use the technology.

“Handing a school an AI tool doesn’t mean teachers suddenly have the time, training or support to use it well,” said Amy McB, a teacher at Pepin Academies, a special education school in Pasco County, Fla.

McBride said she also worried about unintentionally widening the very access and use gaps that investments like this are intended to close.

“The students who need the most help may also need the most adult guidance in learning how to use AI effectively,” she said. “Technology can be a powerful equalizer, but only if we don’t mistake access to the technology for access to a great education.”

The Gates Foundation, however, argues that its investment is fixing a “basic market failure”: the people with the most need for technology have the least power to shape where investment goes.

“Left to the market alone, the most capable tools will be built first for the people and institutions most able to pay for them—not necessarily for those who could benefit most,” Bill Gates, Microsoft cofounder and chair of the Gates Foundation, wrote in a statement.

Doubts about absorption and learning gains

Others in the edtech industry worry about whether lower-resourced schools can actually absorb the gift. While more funding for better AI education is welcome, it fails to address what schools may need first, said Dora Palfi, CEO and cofounder of edtech platform Imagi.

“The schools this money is meant to reach are some of the least able to absorb it for its desired purpose,” she argued. “They often tell me they have to fix literacy and numeracy first, and they’re not necessarily wrong, but it does mean that the districts moving fastest are best placed to use the money.”

“The danger here is that the exact funding made to close a divide could end up widening it,” she added.

Even where schools can absorb the technology, some teachers question whether it delivers the learning it promises. Salena Davis, a history teacher at online Laurel Springs School, said the gains may not be what they appear.

“The improved outcomes may only be short-term in nature and are not genuinely demonstrating mastery-based learning,” she said.

The students the funding is meant to help may also be the least equipped to use AI well, Davis added. Those without “a strong foundational knowledge are not as well positioned to recognize when AI is inaccurate or incomplete, making them more willing to accept responses at face value,” she said.

A Gates Foundation spokesperson, however, said the pilots will be evaluated by outside researchers, including a Johns Hopkins study examining one such tool. The evaluations will test whether early results hold across different schools. Success will be measured by whether students who have fallen behind actually catch up—not by raising scores overall, the spokesperson said.

AI implementation in classrooms has had a positive impact for some. McBride said the technology gave her students another way to ask questions, receive an explanation delivered differently, or work through material at their own pace. For her as a teacher, it saved time creating materials and adapting lessons.

“The biggest change wasn’t that AI started doing the teaching,” she said. “It gave me more time to teach.”

Which view proves out will depend on evidence already in motion. The Johns Hopkins evaluation and other outside reviews will test whether students who have fallen behind catch up using the foundation’s tools, while New York City spends its one-year moratorium studying the technology’s impacts. Those findings will land at the center of the debate over whether classroom AI narrows the divides it aims to close—or widens them.

This story was originally featured on Fortune.com.