It is month six of your district’s artificial intelligence initiative, and the AI ROI numbers on your screen are not the numbers you wanted. Usage is flat. The pilot campuses look a lot like the campuses that never got the tool. Somewhere in the building, a board member is preparing a question you do not have a comfortable answer to yet: was this worth what we spent?
You have two options, and both feel risky. Kill the initiative now and cut your losses. Or defend it with a “trust the process” argument that starts to sound thin the third time you make it. Here is a better option: figure out which situation you are actually in before you decide anything.
Amara’s Law: You Are Probably Early, Not Wrong
Amara’s Law states that you overestimate the short-term impact of new technology and underestimate its long-term impact. It explains why so many district AI investments look disappointing at exactly the six-month mark and then look inevitable at the 18-month mark.
Picture a district like yours, a year into an AI-assisted feedback tool for writing instruction. At month six, usage is inconsistent. A few enthusiastic teachers use it weekly. Most touch it once and quietly go back to their old workflow. The dashboard looks like a failed pilot.
But teacher practice change follows its own timeline, and it is slower than technology adoption curves suggest. The first six months are rarely about the tool. They are about teachers rebuilding habits, testing whether the feedback actually helps students, and deciding the tool is worth the friction of learning it. That work does not show up as usage numbers. It shows up as a slow bend upward in month 10 or 12, right after most districts have already pulled the plug.
The Caveat: Not Every Slow Starter Is a Late Bloomer
Here is the uncomfortable part. Amara’s Law can become an excuse to fund something that was never going to work. Not every flat month six is a story waiting to bend upward.
The test is not whether usage numbers moved. It is whether practice moved. Are teachers who touched the tool doing anything differently, even a little, even inconsistently? Are the conversations in team meetings referencing it unprompted? If the answer is genuinely no across the board, you are not looking at a slow bloomer. You are looking at a tool that never found a foothold, and Amara’s Law will not save it.
Goodhart’s Law: When the Metric Stops Meaning Anything
Goodhart’s Law states that when a measure becomes a target, it stops being a good measure. Districts fall into this trap constantly with adoption metrics, and seat license logins are the clearest example.
In a district like yours, the technology office might track daily logins as its primary adoption metric because it is the easiest number to pull. A campus where every teacher logs in each morning, glances at the dashboard, and closes the tab looks like a success story. A campus where a smaller group of teachers logs in twice a week but genuinely integrates the tool into lesson planning looks like it is underperforming.
The login count rewards the wrong behavior. It cannot distinguish between a habit and a routine, and it will actively mislead you if you let it drive funding decisions. If you are measuring adoption, measure something closer to actual use: assignments created, feedback cycles completed, or teacher-reported changes in practice. Logins tell you who opened the door. They do not tell you who walked through it.
Three More Laws Worth Knowing
Sturgeon’s Law says that 90% of everything is mediocre, and your inbox proves it weekly. In a district like yours, the technology director might keep a running count: 14 AI tool pitches in a single grading period, most from vendors nobody on staff has heard of and half promising to solve the same problem the district already pays for elsewhere. That is not a sign the district is being targeted or that its vetting process is broken. It is the normal base rate of a fast-moving, underregulated market. The useful move is to stop treating each mediocre pitch as a fresh disappointment and start treating the 90% failure rate as the baseline you are filtering against. A vetting process built around that expectation looks for reasons to say yes quickly, rather than exhausting itself trying to fairly evaluate everything that lands in the inbox.
Brooks’s Law says that adding people to a late project makes it later, and district technology offices run into this every time a rollout stalls. When a writing tool’s adoption lagged at a campus like yours last fall, the instinct was to pull in two more instructional coaches to run extra training sessions. But the coaches were not the bottleneck. Teachers were not confused about how to use the tool. They were unconvinced it was worth the extra 10 minutes per assignment, and no amount of additional training addressed that. The coaches spent weeks reteaching a skill that was never the actual barrier, while the real question, whether the tool’s value was clear enough to justify the time cost, went unanswered. Before staffing up a struggling rollout, find out what is actually stalled. More hands help with a training gap. They do nothing for a value gap.
Hofstadter’s Law says it always takes longer than you expect, even when you account for Hofstadter’s Law, and a timeline like yours is a tidy example. The original plan called for full staff onboarding by October, padded from an initial August target because the district had learned from past rollouts to expect delays. October came and went with about 60% of staff onboarded. The padding was not wrong. It was simply not enough padding, because the sources of delay, staff turnover, a mid-year schedule change, a competing curriculum adoption pulling attention away, are the kind of thing no timeline fully anticipates in advance. That is not a planning failure worth punishing. It is what timelines for anything involving human behavior actually do, and building in a second layer of slack, beyond the slack you already added, is a more realistic response than treating each slip as evidence the plan was flawed.
The Law of Diminishing Returns: Your Fourth Platform Is Not Helping
The Law of Diminishing Returns states that each additional unit of input yields less benefit than the one before it, and your AI tool stack is a clean example. A third platform might add real capability. A fourth one, layered on top of two that are already underused, mostly adds a login screen and a renewal invoice.
Run the math on what you are actually paying for right now. If your district is paying licensing costs on three platforms and only one is seeing consistent use, the other two are not idle assets waiting to be activated. They are recurring costs with no offsetting benefit, quietly renewing every year until someone notices.
What to Do With This at Month Six
You do not need to guess whether your initiative is a slow bloomer or a dead end. Before your next budget or renewal conversation, pull three things: a practice-change signal, not just a usage number; a real count of what platforms you are paying for and which ones show actual use; and an honest answer to the caveat question above.
If you want help running that check, TCEA’s AI Investment Audit walks district leaders through exactly this kind of review, and it pairs directly with our earlier post, “Does Your AI Support Your District Improvement Plan? A Practical AI ROI Check.” If you read that post when you launched your initiative, month six is the right time to revisit it. Reach out to Bruce Ellis at TCEA to talk through what your own numbers are actually telling you.