Multiverse Instructors Say AI Review of Every Class Is Causing Stress

A teacher answering a student’s question may now have to wonder whether an AI will call it a digression. Multiverse instructors told the Guardian that a new system reviewing their online classes has left them anxious and feeling continuously watched. The company says the tool spots possible problems for managers to check, not verdicts on teachers’ performance.

A wider view, or a longer watch?

Multiverse calls the tool Compass. It reviews sessions against the company’s teaching framework and combines its findings with learner feedback. Managers receive a review pointing to possible quality risks; instructors are meant to receive suggested areas for development. Multiverse argues that a monthly manager observation offered too narrow a view to catch emerging problems or tailor support. Its new approach greatly increases the amount of teaching examined.

What the system notices

The system analyzes transcripts of online classes, assigns instructors a risk status and confidence score, and writes comments for managers. According to the Guardian, its checks can flag unclear task instructions, repeated filler phrases and a slow response to a disrupted connection. Guidance described by the paper tells the AI not to fail a teacher for one or two hedging phrases; it should look for a repeated pattern across the transcript.

That distinction may matter less to an instructor who has to explain a flag. Teachers told the Guardian that managers had questioned suspected digressions that were actually answers to students’ questions. One said the need to avoid being marked down pulled attention away from listening to learners. The accounts show how even a flag meant for human review can affect a class before any formal performance decision is made.

The human decision does not end the dispute

Multiverse says delivery managers check Compass’s findings against what they know and decide whether to act. Performance management rests on human observation, it says, and the tool makes no decisions about anyone’s job. In its response to the Guardian, the company said the most consequential action the system can take is to recommend that a person review a session. That is a limit on the AI’s authority, not on how often it assesses instructors’ work.

Several instructors described sleep disruption, anxiety and a sense of continuous scrutiny. One told the Guardian they were waking in the night; another said they felt they had to watch their words rather than focus on the lesson. These are staff accounts of the tool’s effects, not a finding that every instructor has had the same experience. They challenge the company’s claim that broader review will produce better, more personalized feedback.

Who gets to see the assessment

Access is another point of friction. Multiverse says managers tested Compass reports in August, then introduced the tool to instructors in September. In September, only managers could see Compass reports; Multiverse plans to show instructors the same information in October. The company says it communicated the introduction in advance. An instructor writing for the Guardian said staff had been kept in the dark while the system was developed and could not yet inspect its analysis.

Multiverse trains workers in AI and digital skills, making teaching quality a substantial concern for the company. In February, UK education inspector Ofsted found shortcomings in several areas of its provision but praised its instructors as skilled and effective. Compass is intended to catch quality problems sooner than occasional observations could. Whether access to the reports and manager checks can make that process useful to the people being reviewed remains unsettled.