Blog: Three Filters for Values-Aligned AI Decision-Making

Not all AI use is equal. There’s a conversation happening in the nonprofit sector right now about AI and environmental impact. But a lot of the conversation stops at “AI uses too much energy and water,” which isn’t very nuanced and isn’t useful enough to act on as a decision maker at your nonprofit.

Here’s what we’ve come to believe at Community IT: the question isn’t whether your organization uses AI full stop. It’s whether you’re using it thoughtfully. And “thoughtfully” means more than picking the right tool, as hard as that is, and building a values-aligned organization-wide policy.

The next step in “thoughtfully using AI” at your nonprofit means building a shared understanding across your team and your leadership of what your AI use actually costs, and what it’s actually worth.

Community IT has developed a three-filter framework to help nonprofits do that analysis. These aren’t rules. They’re questions. The kind of questions that belong in your AI use policy and your ongoing, evolving, team and leadership conversations.

Filter 1: What Are You Asking AI to Do?

Not all AI tasks carry the same environmental footprint. Research shows that text-based tasks like drafting documents, summarizing meetings, answering questions, and doing research are relatively lightweight in terms of energy and water consumption. Coding and using agentic tools use more tokens that a basic chat, but they are still in the text family.

The jump from text to image or video is where there is a magnitude change. Research finds creating an image with gen AI requires about 1500x more energy than a chat query. Video generation is also more demanding than text, using at least about 2000x the energy, and is the most resource-intensive thing a consumer can ask AI to do, particularly since the video is rarely ready to go after the first prompt. Each successive prompt also requires that higher energy use, so the more edits you make, the more energy you use.

That distinction alone gives your organization something to work with. A policy that says “we use AI to draft and summarize, but we don’t use AI to generate video for social media” is grounded in something real. It’s not an arbitrary rule; it’s a values-aligned choice.

It’s also worth thinking about the arc of a project. Building an AI-powered tool or agent can be intensive upfront, but once built, ongoing use may be light. A one-time investment that saves your staff 10 hours a week is a different calculation than burning resources on something ephemeral, or that will continue to use the grid at elevated levels indefinitely.

The question to ask: What types of AI tasks does our team actually use or want to use, and what do we know about how they compare in terms of resource consumption?

How to answer? A staff survey or organization assessment can help you identify the use cases your organization is using or thinking about, and then rate them against energy use to help develop your policy.

Filter 2: Where Are You?

This filter surprises people, but it matters: the carbon intensity of the electricity powering AI data centers varies enormously by location, time of day, grid conditions, and water availability. An organization in a state with robust renewable energy infrastructure and ample water using AI on a Tuesday evening is doing something meaningfully different from one running the same task during a heat dome and drought during working hours when the grid is under stress.

This doesn’t mean you need to become an energy policy expert. But it does mean that “where are you?” is a legitimate policy question. What state are your staff in? If you have remote employees in different regions, they’re operating in different grid zones. Do you know anything about your state’s renewable energy mix or grid policies? What about where your data center water comes from? Heat domes are a useful example: they stress energy grids and data centers simultaneously, at exactly the moment when increased cooling demands are highest. That convergence is worth knowing about.

Your AI use policy doesn’t have to solve for all of this, but acknowledging it gives your team permission to factor it in. Some organizations may choose to do their most AI-intensive work at off-peak hours, or to flag high-demand periods as times to be more conservative.

The question to ask: What do we know about the energy grid conditions and water availability our AI use depends on, and is that something we want to build into our acceptable use guidelines?

How to answer? The owner of your AI policy, whether a board member, committee, or leadership, can do the research. Utilize academic and nonprofit peer organizations with expertise in grid resilience and local policies, who may have the best data and institutional knowledge of your local grid, water, and regulation issues.

Filter 3: What’s on the Other Side of the Ledger?

This is the filter that most AI-and-environment conversations may skip, and it’s essential for mission-driven organizations.

Environmental cost doesn’t exist in a vacuum. It exists in relation to benefit. Using AI to draft a grant proposal that will fund a smart grid project is a different moral calculation than using AI to generate a 5-minute video for a social media post. Both use resources, but they don’t weigh the same.

For nonprofits, many of whom are chronically understaffed or under-resourced, this filter is especially important.

If AI is helping you cover the work of five people that you’ve never had the budget to hire, that benefit belongs in the calculation. If AI can help you reach more constituents more effectively, or otherwise improve your mission delivery, that belongs in the calculation. If your organization serves communities that are disproportionately impacted by AI for any reason – workforce, climate change, bias, intellectual theft – the alignment (or misalignment) between your AI use and your mission belongs in the calculation too.

We’re not suggesting that any benefit justifies any cost. We’re suggesting that using your organization’s values from your theory of change, and considering the communities you serve and the commitments you’ve made, give you a framework for making those calls with integrity.

The question to ask: When we weigh the environmental cost of an AI task against its benefit, does the use align with our mission and our values, or does it cut against them? How do the “pros” stack against all the “cons”?

How to answer? Will be up to your organization broadly and will be an ongoing conversation. Your Executives and Board members need to lead this conversation. Don’t overlook the institutional knowledge and values owned by your program and frontline staff to contribute to this evaluation. The good news? Having this conversation transparently can help surface exciting AI use cases and benefits you may not have considered that should be weighed against the risks and wider environmental and cultural costs.

Putting It Together

These three filters work together. An organization that primarily uses AI for text tasks (Filter 1), is located in a state with a clean energy grid using lots of renewables (Filter 2), and is using AI to extend the capacity of an understaffed team serving vulnerable communities (Filter 3) is in a very different position than one generating AI images for marketing in a drought-prone coal-heavy grid region for reasons of convenience more than mission.

That doesn’t mean the second organization should stop using AI. It means they may have more work to do to get their use policy to a place they feel good about, and these filters are meant to give you a place to start.

We drew on a range of research in developing this framework. A note on how it fits with our other resources: our AI policy template covers the organizational layer – the risks, values, and norms your leadership and staff translate into guidelines. This three-filter framework helps you make good day-to-day choices within that policy, as a staff together, as the wider risks of AI to society continue to evolve.

If your organization is building or revisiting an AI use policy, this kind of structured, values-based analysis is exactly what we help nonprofits think through. Our Mission-Aligned AI Adoption Model gives you a broader framework for assessing where your organization is in its AI journey and how to move forward intentionally.

You can also explore our full library of AI resources for nonprofits including our AI policy template, recorded webinars, and AI for Nonprofits Guide built specifically for the nonprofit sector.

Is Your IT Partner Meeting the Moment?

As your organization navigates challenges, you deserve a support team that is as committed to your mission as you are.

Would you like to see how our human-centered approach can stabilize your IT environment? If you have questions about how to align your technology with your mission, we are here to help.

Community IT has been serving nonprofits exclusively for 25 years. We offer Managed IT support services designed for organizations that want a partner, not just a provider. For a fixed monthly fee, we provide the proactive planning and ongoing IT strategy you need to ensure your technology is always an asset and opportunity, not a liability.

We think your IT partner should be able to explain everything clearly, without talking down to you or using unnecessary lingo. If you’re ready to gain peace of mind and clarity about your IT roadmap in our challenging environment, let’s talk.

As advocates for using technology to work smarter, we’re practicing what we recommend. This article was drafted with the assistance of AI, but the content was reviewed, edited, and finalized by a human editor to ensure accuracy and relevance.

Photo of Hoover Dam by Ryan Thorpe on Unsplash

Blog: Three Filters for Values-Aligned AI Decision-Making

Filter 1: What Are You Asking AI to Do?

Filter 2: Where Are You?

Filter 3: What’s on the Other Side of the Ledger?

Putting It Together

Further Reading

1. AI energy consumption by task type

2. The video generation energy gap specifically

3. Grid carbon intensity use tool

4. Water use, the grid, and local connections

5. Nonprofit-specific framing from Community IT:

Is Your IT Partner Meeting the Moment?

Webinar: AI in Practice for Nonprofits