On the hidden workload of keeping up with technology you didn't ask for

At some point in the last two years, keeping up with AI became part of your job description. Not officially. Nobody added it to your contract or gave you time for it. But the expectation arrived anyway — in professional development sessions, in admin emails, in the sideways pressure of colleagues talking about what they're using now, in the quiet guilt of not having tried the thing everyone says you should try.

That pressure has a cost. And almost nobody is accounting for it.

– THE REALITY

The headline finding you've probably seen: a 2025 Gallup survey, funded by the Walton Family Foundation, reports that teachers who use AI tools at least weekly save an average of 5.9 hours per week — the equivalent of six weeks per school year. It circulated widely. It was cited in dozens of education publications. It became shorthand for the case that AI adoption is worth the investment.

Worth reading the fine print: those savings were self-reported by the 32% of teachers already using AI weekly. They aren't representative of the 40% of teachers not using AI at all, or the 28% using it infrequently. And the survey was commissioned by an organization that has made substantial investments in AI education tools. That doesn't make the finding false, but it does mean it describes a best-case subset — the teachers who have already done the work of integrating AI into their practice — not the average experience of adoption.

A different set of numbers tells a different story. A survey by the Royal Society of Chemistry asked teachers directly whether AI had reduced their workload. Just 3% said it had greatly reduced it. Across workers more broadly, the Upwork Research Institute found that 77% of employees using AI tools reported those tools had actually increased their workload — because of time spent learning the tools, reviewing AI-generated output for accuracy, and being asked to do more work on the assumption that AI would compensate.

Those numbers aren't contradictory. They describe different points on the same curve. If you have already climbed the learning curve, integrated a handful of tools that fit your actual workflow, and built habits around using them, AI probably does return more than it costs. If you're still on the upward slope — trying tools, discarding them, being introduced to new ones before the last ones settled, building competence in your own time with no support — the math looks very different.

Most teachers are still on the upward slope. And the slope has its own workload attached.

– WHAT’S ACTUALLY COSTING YOU

The hidden costs of AI adoption don't show up in surveys about time saved, because they're harder to measure than hours. They show up in the accumulated weight of the week.

Research on AI adoption in education has identified two specific anxiety patterns teachers experience. The first is competence anxiety — the fear of not being able to use these tools well enough, of falling behind, of being visibly less capable than colleagues who seem to have figured it out. The second is role anxiety — the subtler, deeper worry that when AI can generate a lesson outline or provide feedback on student writing, something that can be considered important about the professional value of teaching is being quietly eroded.

Neither of these shows up in your planning period. Both are consuming energy.

There's also the cognitive overhead that comes before any tool gets used: evaluating whether a new tool is worth the learning curve, checking its privacy policy, wondering whether your district has approved it, deciding whether the output it generates will actually save you work or create a different kind of work when you have to verify, edit, and adapt it. Each tool you don't adopt after investigation takes time. Each one you adopt and then abandon takes more. The tool that actually works — when you find it — was preceded by several that didn't, and that cost is real even though it's invisible in the time-saved column.

This is the AI pressure tax. It's what you pay before any savings arrive.

– THE SHIFT

The question worth asking isn't am I using AI enough? That question has no good answer because it frames adoption as an obligation rather than a return on investment.

The more useful question is is this tool paying me back?

A tool pays you back when it returns more in time, cognitive load, or quality than it cost to learn, maintain, and use. A tool takes more than it gives when the learning curve, output-checking, adaptation work, and ongoing updating exceed whatever it saves. Both situations exist. Neither can be answered in the abstract — only in the context of your specific workflow, your specific students, the specific tasks you're using it for, and — rarely discussed — the consideration you place on the environmental and social impact of using AI tools, and specifically, finding a balance that doesn't force you to choose between the extra labor of trying to 'give back' or the persistent guilt of feeling like your tools are harming the planet..

This reframe matters for sustainability because it replaces a vague cultural obligation with a concrete personal calculus. You're allowed to decide that a particular tool doesn't pay back in your context, even if it works well for a colleague in a different one. You're allowed to use exactly the tools that return something to you and leave the rest. That's professional judgment, not resistance to change.

– THE FRAMEWORK

Before you adopt a new AI tool — or when you're reassessing one you're already using — run through these four questions. Budget ten minutes. This isn't a formal evaluation; it's a personal accounting.

What task is this actually for? Be specific. Not "lesson prep" — which specific part of lesson prep, how often does it come up, and how long does it currently take? A tool solving a problem you encounter twice a week is a different investment than one solving a problem you encounter twice a term.

What does it cost to use? Learning curve, ongoing time to maintain prompts or settings, time spent reviewing and editing output before it's usable, environmental impact, and any privacy or data obligations you'd need to investigate. Estimate honestly.

What does it return? Again, be specific. Time saved per use, multiplied by how often you'd use it. Quality improvement, if that's what it offers. Cognitive load reduced, if the tool handles something that currently occupies mental bandwidth.

Does the return exceed the cost, in your context? For this task, with your workload, given where you are on the learning curve for this specific tool — does it come out ahead? If yes: worth using. If no, or not yet: worth waiting until it does.

The last question is the one most adoption conversations skip. They treat the potential savings as the whole story and ignore the investment required to realize them. Your audit closes that gap.

– The Honest Part

Most of the pressure driving AI adoption isn't coming from you. It's coming from administration, from district mandates, from PD sessions designed around tools that someone else chose, from a cultural narrative that frames non-adoption as falling behind. Individual return audits don't change any of that.

When AI tool adoption is mandated — when you're told what to use and when to use it, regardless of whether it returns anything in your specific context — you're back in the same position as any other mandated initiative: doing the compliance work without the agency that would make it genuinely useful. The extraction mindset from our previous PD post applies here too. Find the one thing it actually returns to you, use that part deliberately, and don't spend energy resisting the rest.

What individual practice can do is protect your energy from the costs that are genuinely optional. The guilt about tools you haven't tried yet is optional. The three hours spent exploring a new platform on a Sunday evening because you felt like you should is optional. The pressure to be visibly current is optional. None of those are the same as genuine professional development. You're allowed to draw a line between what serves your teaching and what serves the narrative around AI adoption — and to notice which costs you're paying in service of which.

– Your Move

Pick one AI tool you're currently using or feel pressure to use. Run the return audit on it — ten minutes, the four questions, honest numbers. Does it pay back in your context right now?
If yes: keep using it, deliberately.

If not yet: put it down without guilt. It may pay back later, when you have more time to invest in the learning curve, or when the tool improves, or when your workflow creates a better fit. That's a professional decision, not a failure.

Drop your audit in the comments — which tool, what you found, what you decided. The most useful thing we can build collectively is an honest account of what actually returns something and what doesn't, in real classrooms, with real constraints.

If you had to guess, how many 'invisible hours' have you spent this year just auditioning tools that didn't end up making the cut?

---

Note: This series provides professional frameworks for managing workplace stress and workload. It is not a substitute for medical advice, diagnosis, or treatment. If you are experiencing persistent feelings of hopelessness, severe anxiety, or physical symptoms, please consult a healthcare professional or mental health provider.

This is part of Accingo's Sustainability Studio — Making teaching a lifelong career with workload and boundary focus