The Two-Speed Staff Problem: When Half Your Teachers Use AI and Half Refuse

Walk through most schools today and you will find two groups of teachers. One uses AI every day and talks about it with enthusiasm. The other has avoided it, sometimes by principle and sometimes because nobody has shown them a reason to start. The two groups often work in the same corridor and barely speak about it.

This is the two-speed staff problem. Left alone, it widens, and the usual response, a mandate from above, tends to make it worse.

Why the split forms

1. Confidence compounds. Early users gain skills, share wins and get recognised, which encourages more use.
2. Hesitancy compounds too. A teacher who tries once, gets a poor result and hears no guidance concludes the tools are not worth the time.
3. The reasons for refusal differ. Some teachers have principled objections about learning, authorship or data. Others are anxious, short of time or worried about looking unskilled. Treating all refusal as one thing misses most of it.
4. There is no shared baseline. Without clear school guidance, each teacher sets their own rules, and the gap in practice grows unseen.

Why mandates deepen it

A mandate such as “all teachers will use AI in lessons” makes three predictable mistakes.

• It turns principled caution into defiance, because teachers are told to adopt before their concerns are heard.
• It produces box-ticking use that looks like adoption and changes nothing in learning.
• It pushes enthusiasts to take shortcuts, since they feel the school is behind them and the rules are unclear.

Neither extreme is what a school needs. Uncontrolled enthusiasm creates data and integrity risks, and blanket refusal leaves students with inconsistent and sometimes unsafe experiences.

The equity dimension

Students pay for the split. A child in one classroom may be taught to question AI carefully, while a child next door is told nothing, or told to avoid it entirely, or is allowed to use it for everything. Their experience of school, and of assessment, differs depending on whose class they land in.

It also affects staff fairness. Early adopters may be seen as innovators and promoted, while sceptical colleagues with valid pedagogical concerns are treated as obstacles. That risks losing the critical voices a school needs most when adoption moves quickly.

A shared agreement instead of a mandate

The practical alternative is a short, shared agreement that sets limits for everyone and leaves room for professional judgement within them. I call it red lines and green lights.

1. Red lines. Things that nobody does, regardless of enthusiasm: entering identifiable student data into unapproved tools, using AI to make final decisions about a student’s grade or welfare, hiding AI use from students or parents.
2. Green lights. Uses the school actively supports and will help staff to try: lesson planning support, generating practice questions for teacher review, accessibility adaptations.
3. The amber zone. Uses that teachers may try with care and a duty to report back, such as AI feedback on drafts.
4. The right to decline. Teachers may choose not to use AI in their own practice, provided they follow the red lines and the school’s expectations on student AI literacy.
5. A review date. The agreement is revisited each term, with input from both groups.

This links closely to the decision-rights gap in The AI Policy Vacuum.

How to narrow the gap

1. Listen to the sceptics first. Hold open sessions where concerns are recorded and answered. Some will improve your policy.
2. Pair across the divide. Match an enthusiast with a hesitant colleague for a single, specific task, with the enthusiast’s role to help, not to persuade.
3. Make the first step small and safe. One low-risk use, one lesson, one reflection.
4. Share failures as well as wins. It shows that learning to use AI includes getting it wrong.
5. Do not use adoption as a performance measure. Measure learning, not tool usage.
6. Support with coaching. The approach in Training Teachers Without Another PD Day works well for this.

The question for your next leadership meeting

Can you describe, in two sentences, what every teacher in your school is expected to do and not do with AI? If you cannot, each teacher is writing their own policy, and the two speeds are the result.

Related Reading

• The Last Generalist Teacher
• AI in Schools: Ban It or Teach It? The Question Is Wrong
• The Teacher Onboarding Gap
• The AI Paradox

Saif Ullah Khalid · saifullahkhalid.com