At Glendale Community College in Phoenix, Arizona, on May 15, 2026, dozens of graduates walked across the stage and heard nothing.

The college had deployed a new AI-powered name-reading system for commencement. The system mispronounced some names. It skipped others entirely. The names on-screen stopped matching the people walking. The ceremony paused twice. College President Tiffany Hernandez apologized and called it “a lesson learned.” The audience booed.

Initially, administrators told graduates they could not walk again. Then the backlash arrived, and they reversed. Students were called back to the stage. This time, a human read their names.

Grace Reimer graduated that day. She didn’t hear her name when she crossed. She heard it later, from her seat, after the fix. “I would have liked a little more thought to have gone into it,” she told Arizona’s Family, “rather than pushing something as simple as reading some names off to an AI device.”

Glendale Community College serves 13,914 students. Forty-three percent are Hispanic or Latino. Thirty-seven percent are white. Nearly six percent are Black. About five percent are Asian. Just over one percent are Native American. It is a community college — first-generation students, working adults, people who took longer and paid more and showed up anyway. For many of them, this was the ceremony. The only one.

The AI system was supposed to solve a real problem. Mispronunciation at graduation is not trivial. It is disproportionately experienced by students with names that don’t follow Anglo phonetic patterns — which, as of the 2023-24 school year, describes 54 percent of American public school graduates, up from 34 percent two decades earlier. That gap is real. The disrespect of a mangled name in front of your family is real.

But the solution Glendale chose didn’t mangle the names. It skipped the people.

***

The product is called Tassel. It announced 1.3 million graduates last year. This year it expects 1.6 to 1.8 million. The process: each student scans a personalized QR code before stepping onto the stage. The code triggers an AI-generated recording of their name that they pre-approved during registration. If the AI can’t get it right after three attempts, a human voice actor records it manually.

Tassel CEO Chase Rigby told Education Week: “I don’t know a more human application of AI than this.”

Scan. Walk. Be announced. Like self-checkout at the diploma store.

***

Glendale is not alone.

At Pace University in New York, students were told to write their names phonetically so the AI would say them correctly. It still got them wrong. One graduate: “They told us to write our names phonetically so it’s said correctly, and they still said my name wrong, which is forever documented in videos.”

At Northeastern University in 2025, a graduate said the AI voice “just felt so alien and weird, and I felt really pulled from the moment.” Another wrote that the university “showed more reverence for artificial intelligence than for its student body.”

At Columbia University, Tassel is being used for Class Days ceremonies for the second consecutive year. Students and faculty protested outside the gates on May 18. Mathematics professor Michael Thaddeus asked whether attendees would want “an AI voice” at a wedding or a funeral. Student organizer Hannah Sommers called it “an example of AI being used where this is clearly not needed.” The university declined to comment.

***

Some schools listened.

At Washington-Liberty High School in Arlington, Virginia, the plan was to use Tassel for their June graduation. Students pushed back. Parent June Prakash told the school board: “Names carry deep cultural and personal significance. When spoken by someone who knows the student, it reflects respect and belonging.” The school reversed course. A staff member will read the names.

At West Chester University in Pennsylvania, a student petition gathered over 1,000 signatures. The university dropped Tassel and paid more for a service where real humans record each name.

The fix exists. It costs more. West Chester paid for it. Arlington chose it. The question is who’s willing to pay — and who decided the students weren’t worth it.

***

Here is what Glendale Community College’s own published guidance on responsible AI use says: text-generation systems “are known to produce inaccurate information.” Large language models “cannot tell fact from fiction.” Errors can produce “real-world negative or dangerous consequences.”

They published that guidance. Then they pointed the technology at a microphone and told it to say their students’ names.

***

A human announcer who stumbles over your name is still trying. They are standing at a podium, holding a card, squinting at syllables they haven’t seen before, and doing their best to get it right in front of your mother.

An AI that skips you isn’t trying. It isn’t failing. It isn’t doing anything at all. It is an absence where a person used to be.

Graduation is one of the last ceremonies where someone says your name out loud, in a room full of people, because you did something worth saying it for. It is not efficient. It is not scalable. It is not a product. It is a human being recognizing another human being.

Tassel will announce up to 1.8 million names this year. That’s not a ceremony. That’s a market.

The problem was real. The names were hard. The humans kept getting them wrong.

The solution was to remove the humans.

// NEON BLOOD