.png)
Best, cheapest, and safest have quietly become one word
Through the 5 Lenses, October 2026 Edition
This is the first article in a new October series of the 5 Lenses, starting of course, with Purpose.
Organisations are adopting AI faster than they've worked out what they want it to do and why. That's not news though, it is being discussed in forums, conferences and writings across the globe. The specific part of this I am looking at now, both in my own business as much as in clients', is smaller and more personal: its adoption through convenience. Nobody decided what, why, when and how AI should be used in their business. The tool just turned up and somehow became part of the daily workflow.
An organisation that never chose
I’m sure the scenario is familiar. A comms officer is using an AI tool to help draft newsletters. She's not doing anything wrong. It comes as part of her standard software package, it’s useful, quick, and nobody told her not to. Down the corridor, the finance manager is doing something similar with a tool that came bundled into last year's software upgrade, switched on by default before anyone thought to ask what it did. Neither of them mentioned it to the other, because neither of them thinks of it as a decision: it was already there.
Ask around, and you'll hear the same honest answer everywhere: it was already there, or a colleague recommended it, or it was free, or it came bundled with something else the organisation already pays for. None of those are bad reasons on their own. What's missing is the step before any of them: nobody asked what the organisation actually needed the tool to do, and then checked whether this was the tool that did it, at the price, with the safeguards, that mattered. Best, cheapest, and safest have quietly become one word: convenience also known as whichever tool is already open on the screen.
That's the dilemma this article explores. Not that organisations are choosing badly. That almost nobody is choosing at all. And to be clear before going further: choosing the free tool, or the one already bundled into an existing licence, can be the right call. Many organisations especially in the charity and not for profit sector have little spare capacity for a formal procurement process, and convenience is often a reasonable response to being stretched. The gap isn't using the tool that's already there. It's that almost nobody can currently say whether that was a decision or an accident.
What research helps us see
The scale of the non-decision is measurable, and the numbers are more precise than "most charities haven't thought about this."
The Charity Digital Skills Report 2026 (Zoe Amar Digital and Think Social Tech, 807 responses, the largest sample the survey has had) found that 79% of charities are now using AI in some form. On its own that sounds like a sector that has moved. Look one line down and the picture changes: only 4% describe that use as "widely used and part of our strategy or future plans." The gap between those two numbers is the gap between trying or using AI in some way, and having it widely embedded in strategy or future plans. Commentary on the same report stated: the 79% figure is close to meaningless on its own, set against a strategic-use figure that's still in single digits.
The same report asked, for the first time in 2026, whether charities had taken the specific step of agreeing which AI tools are organisationally approved. The answer was 17%. That doesn't mean the remaining charities have made no decisions at all: informal agreement, manager discretion, or an unrecorded working rule could exist in plenty of them. It shows that most charities have not yet formalised agreement on which AI tools are organisationally approved, which is a smaller, more precise claim, but the one the evidence actually supports. The risk isn't simply that a formal approval process is missing. It's that where choices aren't made visible, priorities can default to whichever interest is loudest, rather than the people closest to the work or the people the organisation exists to serve.
That default has a cost that isn't always visible on a budget line, though it isn't automatically a failure either. One nonprofit technology adviser has highlighted a pattern worth checking: organisations may be paying separately for tools such as ChatGPT or Claude without realising that an AI assistant covering some of the same everyday tasks is already available through a Microsoft 365 licence they hold. That isn't proof of bad judgement, or even of the absence of a formal process, given how fast this technology has moved. It's a story about nobody having checked what was already there before reaching for something new, which is a smaller, more fixable thing than a governance failure, if anyone notices it.
A 2024 study in Nonprofit Management & Leadership, based on twenty interviews across NGOs of varying size, adds the piece the sector conversation usually misses. Godefroid, Plattfaut and Niehaves found that NGOs often deliberately under-use available technology for certain activities, not from lack of capacity, but for reasons tied directly to the work: relationship-sensitive tasks, for instance, where a tool would cost more in trust than it saved in time. That matters here because it cuts both ways. Not adopting a tool can be the purpose-led choice. But the same study found something sharper: in international NGOs, decisions about where technology gets prioritised tend to follow donor interest rather than the interests of the field staff and the people the organisation exists to serve, and may work against the organisation's own mission while looking, from the board table, like progress. That finding is specifically about donor-driven prioritisation in international NGOs, not about a comms officer picking a chat tool. I'm borrowing the mechanism here, not claiming the exact case transfers: the same shape, an unmade decision defaulting to whichever interest is loudest, is the pattern worth watching for wherever nobody has named who a tool is actually for, even without a donor in the room.
That's close to the whole article in one finding, read as a warning rather than a proof: a decision that was never actually made isn't neutral. It tends to default to whoever's interest happens to be loudest, or closest, or already holding the licence, and that may not be the interest of the person the organisation was set up to serve.
The Purpose lens breaks down differently at different levels
At board level, the gap is invisible precisely because nothing looks wrong. AI use is happening, the annual report can say so, and nobody at board level ever sees the moment where a comms officer picked a tool because it was free. This isn't a mark against any particular board: the gap is structural, not a failure of attention, which is why even strong, well-run boards miss it. The board's job isn't to approve every tool. It's to ask whether anyone has agreed what "the right tool for this job" would even mean here, and to notice that 17% figure is a description of most boards' own organisation, not someone else's. The better question isn't "are we using AI safely?" It's "has anyone been asked to decide which tools we use, or has that decision simply never come up?"
At manager level, this is where the actual choosing happens, by default, one small decision at a time. A team lead sees a tool doing something useful and lets it spread through the team, because stopping to ask whether it's the right tool feels like slowing down work that's finally getting easier. The failure mode here is not recklessness. It's that raising the question feels like friction with no clear owner to take it to. The better question for this layer: if someone asked me next week why we use this particular tool, would my honest answer be more than "it works"?
At individual and frontline level, this is lived as relief, not risk. A tool that saves an hour on a Tuesday is not going to be interrogated by the person it's helping, and that's a reasonable response to being stretched. The people closest to the work are also the ones with the least standing to change tool policy, even when they can see the gap most clearly. But standing to decide isn't the same as having nothing to do. The one thing worth doing at this level costs almost nothing: say out loud, to whoever's closest to owning it, which tool you're actually using and for what. That's not the same as fixing the gap. It's making sure the gap is visible to someone who can.
Convenient can be chosen
Convenience is not the enemy here. For a stretched charity, choosing the tool already included in a licence may be the most sensible answer available. It saves money, avoids another contract, and lets people get on with work that matters.
The question is whether that convenience has been made visible as a decision. Can someone say: this is the tool we use for this kind of task, these are the boundaries on what goes into it, this is why it's good enough for now, and this is who will revisit it if the use changes? That's not a procurement exercise. It's a proportionate act of judgement, small enough to fit in a ten-minute conversation. Without it, the tool that happens to be open on the screen can start shaping priorities, practice, and risk without anyone having intended it to.
Using the Purpose lens in practice
If capacity is tight, this doesn't need to start as a tool review. Name the tool already in use, name one task it's being used for, agree what must not go into it, and name the person who can answer questions about it. Then write those four lines somewhere the next person can find them, and add a date to look again if the use changes. That's enough to turn an accident into a starting decision.
Questions to ask this week
Can anyone say what the tool we're already using helps with, what it must not be used for, and who can answer questions about it?
Is there one low-risk tool and one named use we could make explicit this month, without waiting for a full review?
Before paying for another tool, has anyone checked whether a licence we already hold covers the need?
If this tool is shaping how work gets done, whose time, voice, or relationship might be affected, including the people we exist to serve?
If the use is likely to continue or grow, who will check in six months that the answer is still true?
What this lens cannot do alone
Purpose can tell you that a decision was never made. It can't tell you who has the authority to make it now, or what happens when the answer is uncomfortable.
We are using AI to help us ___, for ___, and a person remains responsible for ___.
If you can't finish that sentence for your own organisation right now, you have your starting point. That includes the comms officer and the finance manager from the start of this piece: neither of them needs to fix this alone, but either of them can be the one who says it out loud first.
Purpose can show us that a choice has gone unowned. Next month, People asks who was actually in the room when it wasn't.
References
Ramsay, N. (2026). Charity Digital Skills Report 2026. Zoe Amar Digital and Think Social Tech.
Charity Digital Skills Report 2026: detailed findings. Charity Digital Skills. View report
One Hundred Nights (2026). Microsoft 365 Copilot Chat: The Free AI Tool in Your Nonprofit's Subscription. View article
Godefroid, M.-E., Plattfaut, R., & Niehaves, B. (2024). Identifying key barriers to nonprofit organizations' adoption of technology innovations. Nonprofit Management & Leadership, 35(1), 237–259. View Report
*This article was co-created through a human-in-the-lead process using several AI models as thinking partners. It reflects our commitment to ethical, transparent, and accountable use of AI, where human judgement, curiosity, and oversight remain central.*
This article was co-created through a human-led process using several AI models as thinking partners. It reflects our commitment to ethical, transparent, and accountable use of AI, where human judgement, curiosity, and oversight remain central.