My Why of AI
An open badge that asked educators to sit alone with the same four questions and say what they want AI to do in their classrooms. This is what 75 of them said, read across every transcript the badge produced.
How the cohort engaged
My Why of AI ran as an open community badge. Educators enrolled individually rather than as school teams, and the Mark Cuban Foundation Teacher Fellows took part as a participating cohort. Everything below is reported in aggregate. Location was recorded for only a handful of enrollees, so no geographic breakdown is available.
Each dot is one educator. Every session in this badge was a solo reflection of roughly 10 to 15 minutes, guided by the platform. Once someone started, they almost always finished: attrition across four sessions was 75 to 72 to 69 to 68.
When educators recorded. An open badge with no cohort deadline still produced a rhythm: a July opening, an August peak as the school year approached, and a September tail as classrooms reopened.
This badge used solo reflection only. No paired sessions appear in the record, so peer pairing data is not available. Ten educators completed all four sessions in a single day.
Every one of the 343 Ideas was generated by the platform and written back to the educator in the second person. None were authored by educators themselves. Twenty-one were system messages noting that a session captured no audio rather than reflections. The 27 Small Wins are the badge's only hand-written artifact.
Success Signs are observable evidence indicators from the AI for Deeper Learning Framework. These counts come from the signs educators chose when writing their own Small Wins, so they show what people reached for when naming progress.
Learning Skills and Mindsets carries more selections than the other five strands combined. Transcript evidence is distributed differently, as the competency map shows.
The shape of the learning
Seventy-five educators answered the same four prompts alone, on their own schedule, with no one listening. What comes back is a record with unusual agreement in it. Thirty-two reached for the same handful of words to mark where AI use goes wrong: crutch, shortcut, cheating, copy and paste. Twenty-eight described checking what a model returns. Forty-one named a human capability they believe sits outside AI's reach.
That convergence is the asset. These educators arrive already holding common language about where the line sits and what teaching becomes, which is further along than most AI professional learning assumes. It also arrives untested, because a solo protocol collects 75 answers that never meet. The real disagreements are visible only by reading across the transcripts: whether struggling or confident students show deeper learning more clearly, and whether the answer to over-reliance is more practice with AI or deliberate time without it.
The most useful finding is the distance between naming and designing. Educators describe student over-reliance in detail and their own AI use as straightforward productivity, and only two connect the two. They name critical thinking more often than any other phrase and rarely say what a student would do on a Tuesday to practice it. That gap is not a shortfall in this group. It is the specific, workable thing a next badge can take up.
We're letting the worries of the technology be the ceiling of our vision.
I would argue that my prompts are well thought out and that I'm using it for the creativity and to save me time, but I also don't think I can ask more of my students than I ask of myself.
But the thinking is a big part of the work.
Five themes that surfaced
Where a number appears, it counts distinct educators out of the 75 with a recorded session, and it is a keyword floor read back in context with off-topic uses removed. The qualitative reading found each pattern more often than the keyword count catches, so treat these as minimums rather than estimates. Where a pattern resists counting, it is described rather than numbered.
At least 32 educators reached for the same small vocabulary: crutch, shortcut, the easy way out, cheating, copy and paste. The objection is rarely to the tool. It is to a student handing over the reasoning and not reading what comes back.
At least 28 described checking what a model returns, and it is the one pedagogical move that arrives with operational specifics: back-checking a claim, comparing coverage across outlets, checking AI's method against the one taught in class, asking whose perspective is missing.
Asking AI to simplify a text is the archetypal shortcut in the dominant frame. For roughly a dozen special education and multilingual educators it is the whole point, and they argue it rather than assume it. Several refuse to trade dependence on a teacher for dependence on a tool. Their evidence is the most concrete in the record.
When Session 4 asks what stands in the way, the answer is rarely students. It is colleague caution, absent district policy, approval queues, and tools blocked at the firewall. Twenty-nine described restricted or gated access. Much of that caution is well founded, and an open badge draws people already leaning in, so the careful view is described here rather than spoken.
Most visions describe keeping something intact: judgment, voice, relationships, the capacity to struggle. Roughly a dozen describe something that does not yet exist, and those come disproportionately from educators working with students already locked out of something.
Themes were built from the 285 recorded reflections and the 27 hand-written Small Wins. Transcripts were read in full, session by session, then coded and re-reviewed against the whole corpus. These are automatic voice transcriptions, and passages of ambient audio that were not the participant speaking were identified and excluded before any coding or quoting. Eleven sessions ended normally but captured no audio, and the platform-generated Ideas were treated as already-analyzed artifacts rather than as educator voice.
What this badge was built to develop
My Why of AI targeted three competencies from the AI for Deeper Learning Framework, each tied to a learning target. This is where evidence from the badge should be read first. Evidence depth reflects the richness and specificity of the available record, not a performance grade or evaluation.
Define a personal why for AI, ground it in a clear vision for teaching and learning, articulate a stance on responsible and ethical use, and identify the uniquely human skills needed.
The richest record in the badge. Every educator with a substantive transcript articulated a stance or a vision somewhere across their four reflections, and 41 named a human capability they hold to be outside AI's reach. The ethical position is stated far more often than it is operationalized: the stance arrives readily, the classroom design that would enact it rarely does.
Build foundational AI knowledge to support growth, innovation, and decisions, including AI's impact on human skill development and cognition.
Widely present and very unevenly deep, which is the finding worth carrying forward. Thirty-seven treat prompting as a practiced skill, around 17 raise bias in the systems themselves, and only about 10 discuss how models actually work. Hallucination is named outright by six, though the underlying habit of checking the output is near-universal. Others described having barely used these tools, and because the protocol is solo, nobody had any way to see where they sat relative to the rest.
Set and pursue AI learning goals grounded in your why, and use personal learning data to guide upskilling or reskilling.
Sixty-eight educators completed Session 4, where the priority challenge and the AI learning goal are named. Roughly half named something concrete enough to act on, from prompting as a taught discipline to building a tool. The rest answered by reading back a competency title from the framework the prompt had just asked them to open, which is an instrument effect worth designing around rather than a finding about the cohort. Educators who located the challenge in conditions beyond their own classroom tended to name the broadest goals.
Explore the full framework map
All six strands and 33 competencies, showing where this cohort's evidence landed beyond the three the badge targeted.
These reflections touched competencies well outside the three the badge was built around, which is worth seeing but not worth grading the badge against. Every rating below was set by reading the full transcript corpus competency by competency, discounting passages where a participant was reading the framework, the session prompt or the Session 1 infographic aloud rather than speaking for themselves.
Evidence depth reflects the richness and specificity of available evidence, not a performance grade or evaluation. Bold items mean documented evidence is present. Lighter items mean the pattern is not yet surfaced in this cohort's data. Two patterns are worth naming. The record is strongest where an educator evaluates AI and thinner wherever a designed student task is required, which is why CT3, CT4 and LM1 sit lower than the stance competencies above them. And AL4 reads low for a reason that is itself a finding: this cohort discusses creativity constantly, almost always as something to protect from AI rather than something to do with it. PH3 is reported rather than observed, since a solo badge gives peer exchange nowhere to happen inside the platform.
sideby's AI for Deeper Learning Framework is informed by the UNESCO AI Competency Framework for Teachers, the colleague.ai AICE Framework, Teach AI with EC + OECD & code.org frameworks for AI learning, and the Hewlett Foundation Deeper Learning Competencies, evolving as AI evolves and shifts what this technology can mean for deeper learning and education.
Where evidence is strongest, session by session
This view reads across the whole framework rather than the three competencies the badge was built around, so it picks up everything the cohort touched. Four guided solo reflections make up My Why of AI, and each drew on a different part of the map. Evidence depth reflects the volume and specificity of transcript text recorded for each session, not a performance grade or evaluation.
Session counts come from the platform export and are distinct participants, not runs. Eleven sessions across the badge ended normally but captured no audio. They are counted as sessions but contribute no qualitative evidence.
What got written down
Twenty-seven Small Wins were written by 20 educators during the badge window, 24 logged as Successes and 3 as Lessons Learned. They are the only place in this badge where someone chose to write something rather than answer a prompt. Private means visible only to the person who wrote it; launched means shared with the community, and five were. The six below are summarized rather than quoted, so the wording is ours and the substance is theirs.
An educator advocated for AI literacy to be added to the ninth-grade unit where students already learn to tell primary from secondary from reliable sources. The reasoning was that students are using AI to find information regardless, so the useful move is to say explicitly where it fits, what its limits are, and why verification still applies. This is the clearest instance in the record of the verification theme becoming a curriculum decision rather than a classroom habit.
A student was treating an AI response as a primary source. The educator showed them how it could be useful for generating ideas and locating possible sources, then required them to go to the original and evaluate it. The student used the AI response as a starting point and found a credible source to support the work.
In a building where English is not the first language for many students, an educator described watching those students grow more resourceful with translation over two years, moving class materials into Spanish themselves rather than waiting. The documented win is the shift in who initiates. One student asked permission to use a translation tool for assignment directions and got a yes.
An educator used AI to build a differentiated lesson for students with IEPs and 504 plans, drawing on prior assessment data to match level and learning style. They noted it was something they had since shared with colleagues, which is the theme of this badge in miniature: the practice traveled, the documentation did not.
An educator recommended AI for brainstorming experiment designs in a senior science course and found that simple prompts produced near-identical ideas across many students at once. Their stated takeaway was that prompt specificity has to be taught first, before the tool can do anything useful in an inquiry task.
An educator tried to bring an AI tutor into a math classroom and found students could not reach it because the platform was restricted. What they took from it was that accessibility and school approval need confirming before a tool goes anywhere near instruction. This is the access theme showing up as a concrete classroom outcome rather than a complaint.
Six of 27 entries are shown. Small Wins carry no badge attribution in the platform, so these were filtered by person and date. Six of the 27 belong to educators who were also active in another sideby badge during the same weeks, and may reflect that work rather than this one. Only three of the 27 were logged as Lessons Learned, and they are among the most useful entries in the set.
Cohort voice across the badge
Every line below is exactly as an educator said it in one of the four solo reflections, from a group that included classroom teachers from elementary through high school, special educators, multilingual specialists, instructional coaches, librarians, and school and system leaders. These are voice transcriptions, so the false starts and run-ons are theirs and are left as recorded.
So I'm not sure what the answer to that is, because that would be just us pretending that what is being taken away is somehow not human in the first place.
It's like having a mirror that always tells you you look nice.
And honestly, I feel like I'm so far behind most of the other teachers in this program because they're always talking about, you know, using cloud and chat GPT and using all of these different AI programs that I have barely even heard of, much less used in any way.
I think students are very afraid of making mistakes and that's why they use AI to have everything perfect, but in this world we learn when we make mistakes.
They're using AI to access learning.
I would like to be more of a partner than a policeman.
We have already done the hard work of using our problem-solving and our critical thinking to perform tasks throughout our careers that we are now using AI to offload.
If they can participate in classroom discussions, then they feel like they belong in the class and they're not just occupying a seat in that classroom.
I think that we also, the teachers, have that problem, and we just rely in some way, probably because we don't have time.
it's a great tool to use while you're learning it's not the tool to use when you're showing me what you've learned
This is a problem of the scaling, because the more we rely on AI to perform many of our job tasks, the less skilled we become in that area, because we offload that cognitive task to AI tools.
And so the way that the chat is encouraging, and I always seem to be right in the questions that I ask it, that is dangerous.
My role would shift from doing the work for students to teaching students how to work for themselves.
Because obviously it's not going anywhere, um, despite the fact that some people, sometimes myself included, would like for it to go away, um, because there's just so many ethical considerations, privacy issues, um, but it's not going away, right?
So that is very worrisome to me about what I appreciate is some caution and wanting them to move forward in a thoughtful way. They may be ending up doing harm to students when trying to not cause harm.
If we told a kid to write a research paper, we wouldn't ban them from Google.
A lot of the kids that I work with, when I've ever mentioned AI, they're really adamant against using it.
For me, that moment felt like deeper learning, not because they necessarily got the answers correct, but because they were owning their math and supporting each other.
If you can articulate it, verbalize it, teach somebody how you are thinking, you master that content.
I need to learn how to build agents and use co-work, but it's blocked.
Since AI can give you answers but they cannot detect what does your heart desires.
It's me and 24 students all working on, like, an Adobe InDesign assignment, and I can't help 24 students.
When I think about my long-term goals with my students in an AI-influenced world, my vision isn't actually about the technology, it's about who they're becoming as thinkers and humans.
So I'm out of I'm working out of both excitement and fear, I suppose, at this point.
From one badge to the bigger questions
My Why of AI is a completed badge, and its record is useful well past the people who took it. What follows is what it demonstrated about open, self-directed learning around AI for deeper learning, where the badge itself goes next, and the questions it puts to the field.
The range in this cohort was enormous. One educator had built their own math tool, signed a data agreement, and made it reachable by teachers and students through district sign-on. Another said plainly they had barely heard of the tools their colleagues kept naming. Both finished all four sessions, and the record holds useful thinking from each.
Classroom teachers sat alongside special educators, multilingual specialists, coaches and system leaders. The most concrete evidence came from those working with students already locked out of something, and it reframed the central question: shortcut or access route depends on the student in front of you.
Launching is how one educator's thinking reaches another, and the entries here deserve a wider audience: a ninth-grade source unit rebuilt to include AI literacy, a student taught to treat an AI answer as a starting point. Inviting people to launch at the end of every session, not just once, would put far more of this in front of the community.
Many educators reflected across the whole summer, from late July into September. Nothing in the current sequence invites them to look back at what they said in Session 1, so the shifts that happened over those weeks stay largely invisible in the record. Worth considering how a badge like this could capture them.
At least 28 educators described a checking routine in operational detail. Across every session, one person asked how they would measure whether it worked. Here, at least, the practice runs well ahead of the evidence.
It appears in nearly every transcript as the thing that matters. One educator named the problem directly: it almost becomes meaningless because like everything is critical thinking
.
Asking AI to simplify a text is the archetypal cheat in the dominant frame and the core access move for special education and multilingual educators. Applied without care, the dominant frame gets this exactly backwards.
Twenty-nine described restricted or gated access, including educators whose own stated learning goal is blocked on their district's network. The gap between what students reach at home and in class showed up repeatedly.
Across the field, conversations about what students need from AI are still mostly held among adults, and this record reflects that. It is also where the shift begins: two educators here named learning from their students directly as what they wanted to do next.