Rise of the Robot Teachers

In New Mexico, social media platforms like Facebook and Instagram are labeled a public nuisance, yet the policy now allows artificial intelligence systems to record and store the voices of young children and to assess their reading abilities.

Starting in the previous school year, the state mandated public schools to implement Amira—a personified AI system designed to evaluate and instruct young students—for compulsory literacy assessments, screenings for dyslexia, and weekly tutoring sessions for students in kindergarten through second grade.

The rest of the content that followed in the original article, including the form-related HTML and extraneous technical details, has been omitted here to focus on the core reporting and analysis.

As the 2026–27 academic year began, several parents and school districts voiced concerns about how Amira handles voice recordings from students and whether the tool actually provides reliable reading assessments.

In response, the state’s Public Education Department (PED) let districts opt out of Amira for this year, provided they secure PED approval.

Nevertheless, robo-teachers and automated evaluators were rolled out with surprisingly little oversight or notice last year and remain in use across most New Mexico districts this year, despite limited evidence supporting the effectiveness of AI tutoring.

New Mexico isn’t alone in allowing or requiring AI tutors and evaluators. “Amira is mandated for assessments in two states—New Mexico and Idaho—and is authorized for state use in California, Georgia, Massachusetts, Michigan, Oklahoma, and Texas, though some New Mexico policymakers have acknowledged the rollout was rushed,” reports the Albuquerque Journal.

I’m no alarmist about technology in education, nor am I opposed to screen time for kids in general. Yet I instinctively pause—perhaps rightly, perhaps wrongly—when AI is used to supervise or teach very young children. And from what I’ve learned about Amira and New Mexico’s hurried deployment, parents have legitimate reasons for concern.

Recording Children’s Voice

One of the most pressing worries raised by families and educators in New Mexico concerns the reading sessions Amira uses to listen to students.

“Our communities are asking questions we can’t fully answer: where are these voice recordings stored, and for how long? Who can access them? Are they used to train the vendor’s AI models? What safeguards exist to protect against breaches or misuse?” wrote Jennifer Guy, superintendent of Los Alamos Public Schools, in a July 31 letter to the New Mexico PED. “A child’s voice is, effectively, biometric data, and parents are rightly uneasy about it being collected and retained by a third-party system as a condition of a state-mandated assessment.”

“We do not believe the state should require children to submit their voices to the platform,” Guy added.

Some of these concerns have since been addressed by Amira or by New Mexico education officials. Amira “does not and will not ever sell student data, or use recordings of New Mexico students to build products for anyone else,” said Amira CEO Mark Angel in a August 28 letter to worried New Mexico lawmakers.

And “PED reached an agreement with Amira so recordings wouldn’t be stored long-term,” according to KOB 4. “Last year’s recordings were deleted, and going forward, the data will be deleted every month unless a district asks for otherwise.” Moreover, individual school districts can now impose stricter retention limits, and some have done so.

Still, these issues are being debated after schools had already rolled out Amira in K–2 classrooms for a full academic year. Under the new policy, longer data-retention periods remain a possibility, and schools do not fully control the data. “We aren’t in charge of their server,” Maxwell Municipal Schools Superintendent Amy Roble told a August 17 school board meeting, where she voiced concerns about student data security.

Hasty Rollout and Little Accountability

New Mexico’s Amira rollout has often appeared to be the pattern of “deploy AI teachers first, ask questions later,” a strategy that followed the absorption of its previous student-assessment partner, Istation, into the Amira Learning company.

Starting in the 2025–26 school year—and continuing this year unless districts choose to opt out—Amira became mandatory for literacy assessments of kindergarten through second grade, occurring three times per school year, with monthly assessments for students identified as reading-delayed. It is also used in some schools for first-grade dyslexia screening and for weekly, one-on-one literacy tutoring.

The results of these required assessments can determine which additional interventions teachers and schools implement. The state reportedly spends about $2.7 million annually on Amira for software access, professional development, data analysis, and related tutoring and support programs.

Yet, a May 2026 briefing from the state Legislative Education Study Committee notes a lack of clarity about the vetting processes and stakeholder input used to implement Amira. The report highlights that the state appears to lack a formal mechanism for PED to assess whether the tool achieves its intended outcomes, to standardize how student reading data is collected and shared across districts, or to provide the Legislature with timely, reliable information about Amira’s statewide impact. It also notes that New Mexico does not currently have statewide procurement standards for AI tools, student data privacy requirements specific to AI, or a formal evaluation process for AI-based educational tools.

In a state that once pursued court actions against major tech platforms over youth screen time, some schools are removing time with human teachers in favor of screen interactions with an AI avatar, and this shift occurred without robust rules or accountability measures in place.

But Does It Work?

Given reports from teachers and parents about incorrect scoring and heightened anxiety among young learners, the question of effectiveness is pressing.

“Speech-recognition technology remains imperfect with developing voices and articulation in young children,” noted Guy in her correspondence. This has led Amira to score some students at levels that diverge from what parents and teachers observe. “Because teachers aren’t administering or scoring Amira sessions, and because the ability to correct scoring mistakes has been removed at the state level, our staff and families may face results that are wrong, with no means to fix them.”

There is also concern that Amira assessments may induce anxiety, Guy adds. “Our early-grade teachers and reading intervention specialists report that the testing experience itself becomes a barrier for beginning readers. Children become anxious or unsettled when the software fails to register responses or asks them to repeat themselves, and for students already struggling with reading, the technology layer compounds the challenge… often, the results reflect a child’s comfort with a device and a speech engine or the software’s failures more than the student’s true reading ability.”

Nevertheless, Janelle Taylor García of the New Mexico Public Education Department stated in early August that data show Amira is a positive tool that supports improved academic outcomes for students.

But Guy and other teachers aren’t the only ones who have observed problems.

“The reason families are outraged is that we were never told this was happening,” wrote Sally DeRadke, a mother of three Albuquerque Public Schools students, in a Santa Fe New Mexican op-ed. “We learned about it from our 5-, 6-, and 7-year-olds who said, ‘Amira is like a new teacher, but they don’t really understand me,’ and ‘I don’t like reading anymore.'”

In Maxwell, Roble noted staff suspected some student scores reflected speech patterns rather than actual reading ability, according to the World Journal. “Students with speech delays or other speech difficulties sometimes received scores that teachers did not believe accurately represented their reading skills.”

There are numerous Reddit posts from parents and teachers expressing dismay about Amira’s use in classrooms.

According to a state education department poll, only 8 percent of teachers and administrators surveyed at a training last year said they had “no major concerns” about using Amira.

Fools Rush In

Most efficacy data on Amira come from the company itself, though I did encounter one doctoral dissertation suggesting Amira could improve oral reading fluency scores for first graders.

Yet the overarching issue isn’t whether Amira or similar AI tools can be effective teachers. It’s the sense that schools are deploying these tools with young children as test subjects, without solid data on privacy, child development considerations, evidence of real benefit, meaningful parental input, or sound plans to measure long-term impact.

AI certainly has a place in education. Over time, we may find that tools like Amira can bolster individualized learning—tailoring lessons to a child’s abilities in ways that outpace traditional tutoring, or at least providing support in crowded schools where one-on-one time with every student isn’t feasible.

Right now, however, institutions appear to be leaping before looking. There remains a lot we don’t know about AI’s effectiveness with preschoolers and early elementary students, or about the broader impact of children engaging with AI systems. This seems like a domain where public schools should proceed with extreme caution rather than rushing in.

We might end up with a superior method for student learning. Alternatively, we could face another scenario akin to “iPad for every child”—an expensive experiment with limited payoff. Or something potentially worse. Until more data emerge, public schools would do well to slow the pace of robo-teachers.

---

More Sex & Tech News

• The Safer GAMING Act, approved last month by the House as part of the KIDS Act, would effectively bar children from naming their Pokémon—and much more. TechFreedom attorney Andy Jung explains how this bill risks First Amendment rights by “attempting to silence minors in online video games.”

• There are significant First Amendment questions surrounding the Meta settlement, argues First Amendment lawyer Jeff Kosseff. “Proponents claim the content restrictions and age-verification requirements are not government mandates but voluntary terms accepted by Meta. Yet the Supreme Court has repeatedly scrutinized the voluntariness of such constraints when government pressure is involved.”

• The Anthropic victory against the Trump administration underscores the idea that private firms can condition their dealings with the government, writes J.D. Tuccille.

• “Many sites are now blocking AI crawlers,” observe Australian scholars Dana McKay and Damiano Spina. This makes AI-generated summaries less likely to cite those sources, which could distort information. “Misinformation sites are far less likely to ban AI crawlers, so AI answers won’t be grounded in high-quality sources.”

• “While states’ growth-oriented impulses have lately undermined data centers, national security concerns about leading in the AI race still justify some federal support for data infrastructure,” notes Reason’s Christian Britschgi.