Over the past few years, I’ve watched a growing number of speakers, consultants, and technology advocates describe artificial intelligence as a teammate, collaborator, colleague, co-teacher, or thought partner. Every time I would hear it, I would find myself reaching varying degrees of disagreement, all the way to opposition. Not because I oppose AI. In fact, quite the opposite. I have spent much of my career as an educator helping students, educators, and leaders understand how emerging technologies can support learning, productivity, and innovation. But as someone who also played sports at a very high level, I understand what a teammate actually is:
A teammate shares accountability.
A teammate exercises judgment.
A teammate possesses agency.
A teammate can be trusted with responsibility.
AI does none of those things. It is a tool, a powerful one, an increasingly capable one, but a tool nonetheless.
And tools should extend human capability, not obscure human responsibility.
The Language Problem
That distinction matters most in education. When educators describe AI as a teammate, co-teacher, or teaching partner, I understand the impulse. The framing is meant to encourage adoption, reduce anxiety, or signal openness to innovation. But it quietly obscures something important: educational decisions are human decisions.
An AI system can generate feedback. It cannot know a student.
An AI system can identify patterns. It cannot build relationships.
An AI system can produce recommendations. It cannot assume responsibility for the consequences of those recommendations.
Those responsibilities remain with educators. Always.
This isn’t just a philosophical point. Research increasingly shows that language shapes behavior in concrete ways. When AI is framed as a teammate or employee rather than a tool, personal accountability decreases while accountability attributed to AI increases. Participants in such studies also report greater uncertainty about their professional identity, increased concerns about job security, and, perhaps counterintuitively, lower levels of trust in AI itself. [Harvard Business Review]
In other words, calling AI a teammate doesn’t make people more comfortable with it. It makes the entire human-AI relationship less clear, less accountable, and ultimately less effective.
Accountability cannot, and should not, be delegated to a system that cannot be held accountable.
What AI Actually Does and Does Not Do
The confusion is understandable because AI can be genuinely impressive. But AI systems operate by generating outputs based on patterns and probabilities, not human understanding or intent. They lack the contextual awareness, institutional knowledge, creativity, and ethical judgment that experienced professionals bring to their work. [Harvard Business Review]
For educators, this distinction is not abstract. It shows up every day in classrooms. A language model can analyze a student’s writing and flag areas for improvement. But it doesn’t know that this particular student just experienced a family disruption, that their confidence is fragile right now, or that the most important move is not correction, it’s encouragement. It cannot discern, in real time, the differences between deficit-based and asset-based feedback. An experienced teacher knows those things. An experienced teacher grounded in solid culturally responsive pedagogy knows those things. That knowledge is irreplaceable, and no framing of AI as a “co-teacher” changes the fact that AI doesn’t possess it.
This is why I consistently question and discourage things like the teammate metaphor. It’s not about being resistant to technology either. It’s about being precise about what technology can and cannot do, while preserving the professional authority that educators hold.
What Good AI Integration Actually Looks Like
Rejecting the teammate metaphor doesn’t mean rejecting AI. It means designing human-AI systems more clearly. The most effective AI implementations, in education and elsewhere, keep humans actively involved in reviewing, validating, and guiding outputs rather than simply accepting them. Research continues to confirm this: AI initiatives are most successful when human judgment, review, and decision-making remain central to the workflow. [Insert MIT/HBR Link]
In practice, this means educators using AI to reduce time spent on repetitive tasks — drafting initial feedback, generating resource options, flagging patterns in student performance data — while preserving the relational and judgment-intensive work that defines great teaching.
AI can assist with the mechanical. Educators remain responsible for the meaningful. That division of labor isn’t a limitation. It’s the design.
Research continues to confirm this: despite widespread AI adoption, the majority of organizations fail to achieve meaningful returns, not because the technology doesn’t work, but because they implement it without keeping humans actively involved in reviewing, validating, and guiding outputs rather than simply accepting them.
The same is true in schools. When teachers understand that AI is there to amplify their expertise, not replace their judgment, they engage with it differently.
More critically.
More confidently.
More effectively.
The Bigger Question
One of the most important questions in AI literacy and AI Fluency is not what AI can do. It is what humans should continue to do. For educators, that question has a clear answer. The relational, contextual, and ethical dimensions of teaching are not tasks to be automated. They are the core of the profession. AI can support the work surrounding those dimensions. It cannot perform them. It should never even be asked to perform them. That’s why I believe the goal is not to build stronger relationships with AI.
The goal is to develop wiser relationships with tools, to understand clearly what a tool does, where it adds value, where it falls short, and who remains responsible when things go wrong.
That responsibility always belongs to the human in the room.
Not the tool.