WEBVTT
00:00:22.689 --> 00:00:30.289
Hello and welcome to the Community IT Innovators Nonprofit AI Podcast, Midweek Check-In.
00:00:30.530 --> 00:00:32.689
I'm Carolyn Woodard, your host.
00:00:32.689 --> 00:00:35.090
I am not an AI expert.
00:00:35.090 --> 00:00:49.570
I'm really interested in AI and in the impact that AI is having on nonprofits, both how we're using it at nonprofits and how it's impacting the work that we do in the communities that we care about.
00:00:49.810 --> 00:01:11.329
So in this midweek podcast, every week I try to bring us a little bit of news, filtered through how this is going to impact nonprofits and philanthropy more broadly, and maybe answer some questions and give you some more resources that you can read and follow up on because it's hard with so much coming at us about AI every day.
00:01:11.329 --> 00:01:15.569
If you're reading about it, it's that big fire hose of information.
00:01:15.569 --> 00:01:18.929
So just trying to make it a little bit more manageable.
00:01:18.929 --> 00:01:19.340
So
00:01:19.340 --> 00:01:30.369
Please get in touch with me if you have something you'd like to hear about or something you have a question about, you can reach me through our website, communityit.com, or here on the podcast.
00:01:30.369 --> 00:01:31.649
So
00:01:31.649 --> 00:01:52.129
This week had a couple of resources and news stories that connect in this kind of odd way around a kind of cultural moment that is building, as we speak, a lot of resistance to AI among workers, young people, a lot of nonprofits.
00:01:52.129 --> 00:02:18.289
I'm seeing a lot on LinkedIn, and our clients at Community IT are also skeptical, and many are very resistant to using AI at all, or just very concerned about the power imbalance, the environmental impacts, and the impact on people and humanity for going to use the easy way out instead of writing our own essays or making our own art.
00:02:18.770 --> 00:02:28.370
So I wanted to share a couple of resources with you this week about AI's real effect on climate costs and kind of a framework for how to think about that.
00:02:28.610 --> 00:02:34.129
And also, yeah, this bigger question of resistance like, do you have to use AI?
00:02:34.129 --> 00:02:42.210
What if your company is telling you that you have to, or your organization has a policy and is expecting everyone to use AI?
00:02:42.210 --> 00:02:43.009
So
00:02:43.009 --> 00:02:58.370
The first was this uh news story about an eight-day festival in New York City that I came across called The Summer of Ludd, which is uh feeding into the original Luddite movement.
00:02:58.370 --> 00:03:04.210
Um, there was a conference at the new school in New York on AI and surveillance.
00:03:04.210 --> 00:03:12.289
Uh there were, you know, other activities, kind of in real life activities at this festival.
00:03:12.289 --> 00:03:15.329
There were some concerts in real life music.
00:03:15.329 --> 00:03:17.729
Uh, none of it was advertised online.
00:03:17.729 --> 00:03:20.769
There were flyers, so it was very old school.
00:03:20.769 --> 00:03:24.849
They were pasted around the neighborhood in New York where this took place.
00:03:24.849 --> 00:03:25.169
And
00:03:25.169 --> 00:03:34.769
I came across it through an NPR story about this festival, but also how the word Luddite, if you know it at all, is a political insult.
00:03:34.769 --> 00:03:39.169
It doesn't really relate to historically who the Luddites were.
00:03:39.329 --> 00:03:42.849
So a little bit of a reframe of the history.
00:03:42.849 --> 00:03:59.329
Most people use Luddite as an insult or shorthand for someone who can't work a smartphone or who prefers analog to digital, who likes old school, who likes doing things with their hands, maybe arts and crafts that are not related to technology.
00:03:59.329 --> 00:04:06.609
Sometimes even it's used, you know, for technophobe, someone who's afraid of technology, just doesn't like it, doesn't want to use it.
00:04:06.930 --> 00:04:14.289
But actual Luddites were skilled textile workers in 19th century England who, they weren't against technology.
00:04:14.289 --> 00:04:35.569
They knew what they were doing around the way the looms were working and the technology, these kind of early computers that were these reading cards that helps the loom do the work and make these very complicated patterns, which until then it was only a master weaver who would be able to, you know, handle this kind of complicated pattern.
00:04:35.569 --> 00:04:36.530
So they,
00:04:36.530 --> 00:04:45.250
The Luddites originally were against the way technology was being implemented to enrich factory owners and lay off staff.
00:04:45.250 --> 00:04:56.290
So if you have an early computer running your looms, you don't need those master weavers anymore to set up the loom or, you know, think up the patterns.
00:04:56.290 --> 00:04:58.209
There's lots of new patterns that you can do.
00:04:58.209 --> 00:05:16.129
And really all you need is the one person who punches out the cards and then you put them in the loom machine and it makes the people, the children, still had to run the looms, but the design part of it was taken out of the hand of people and put in the hands of technology.
00:05:16.129 --> 00:05:17.569
And this was, you know,
00:05:17.569 --> 00:05:20.370
The factory owners were thought this was a great idea.
00:05:20.370 --> 00:05:27.009
They could pay fewer people and they could just have the quote-unquote computer doing the setup.
00:05:27.250 --> 00:05:45.649
And the people, of course, who were the crafts people who knew it was like their experience and their understanding of weaving that enabled people to make the computer program that would do the same thing that they could do, but cheaper because it was a computer instead.
00:05:45.810 --> 00:05:57.649
So they petitioned, they marched, they organized, they had strikes, uh, they won some concessions, but the winner of history gets to describe the losers.
00:05:57.649 --> 00:06:04.050
So the fact that this term Luddites, they followed someone, you know, named Ludd.
00:06:04.050 --> 00:06:05.730
And so they were Luddites.
00:06:05.810 --> 00:06:17.170
And at the time it was kind of the ruling class and the factory owners were afraid of the people power that was being developed by the followers of Ludd, the Luddites.
00:06:17.170 --> 00:06:21.089
And so in history, they insulted them.
00:06:21.089 --> 00:06:28.689
They said they were backward, they didn't want progress, they were afraid of the technology, and that is how it's come down to us now.
00:06:28.689 --> 00:06:32.290
I'll share some resources with you in the show notes.
00:06:32.449 --> 00:06:38.930
If you want to go deeper into the history, uh, the book to read is Blood in the Machine by Brian Merchant.
00:06:38.930 --> 00:07:06.689
And he makes this argument that the Luddite story is deliberately distorting that history by people in power who gain from us believing that anyone who is against technology doesn't understand technology or just doesn't know what the tool does, or like wasn't smart enough to be an investor, or you know, is unable to understand the technology and they're a Luddite because they want something different.
00:07:06.850 --> 00:07:29.250
And so he's arguing for recapturing that term and that ethos of wanting technology to enhance humanity and to serve humanity instead of take over from humanity and lead to the wealth of a concentrated few people in power who were already wealthy.
00:07:29.250 --> 00:07:31.730
Uh, he does have a Substack as well.
00:07:31.730 --> 00:07:33.969
I'll put that in the show notes also.
00:07:34.370 --> 00:07:37.569
So you may be seeing this in your own nonprofits.
00:07:37.569 --> 00:07:46.449
And so I just kind of pulled together some of these threads that we've been talking about over the past couple of months here, uh, especially younger staff.
00:07:46.449 --> 00:08:00.209
Uh, if you have teenagers in your house, college students, or young workers at your nonprofit, people in their 20s and 30s, particularly, are pushing back on AI in real tangible ways.
00:08:00.209 --> 00:08:07.569
And this is showing up in polling and some other research that's coming out about generational attitudes toward AI.
00:08:07.810 --> 00:08:14.290
You've probably seen things about this as well, going to IRL in real life events like taking up knitting.
00:08:14.290 --> 00:08:24.370
There's like granny cool, uh granny hobbies that are coming back, spending time with people, especially after COVID and a lot of isolation with technology.
00:08:24.370 --> 00:08:35.170
A lot of younger people who came through that find it very important to make social connections to other people, to meet people, to find people who like your same hobbies.
00:08:35.170 --> 00:08:48.850
And you may have seen articles about using dumb phones or apps on your phone that can manage how addictive they are, how much you're using them, and you know, make you go outside and touch grass.
00:08:49.009 --> 00:08:53.330
You may have staff in this age group or in other areas, I mean, it's not just young people.
00:08:53.330 --> 00:08:59.330
There are plenty of other people who are also deeply skeptical about AI and about technology.
00:08:59.649 --> 00:09:02.929
So you may have people who say, you know, they're just not going to use it.
00:09:02.929 --> 00:09:16.929
And if you're tempted to see that as a training problem that you can quote unquote fix, it might be worth it at your organization or in your family to go a little bit deeper and ask some more probing questions.
00:09:16.929 --> 00:09:18.049
What are they saying?
00:09:18.049 --> 00:09:19.649
Where is this coming from?
00:09:19.649 --> 00:09:22.850
And what, you know, what's behind it?
00:09:23.009 --> 00:09:27.250
Are they against AI for specific reasons?
00:09:27.250 --> 00:09:31.330
Are they concerned about who benefits and on what terms?
00:09:31.330 --> 00:09:34.850
A lot of people want to live their values.
00:09:34.850 --> 00:09:38.850
So they don't want to shop at a place that they don't agree with.
00:09:38.850 --> 00:09:44.529
They don't want to go to a restaurant that maybe has a policy stand that they don't agree with.
00:09:44.529 --> 00:09:51.490
They want to try where possible, and it can be uh it can be difficult to do this completely.
00:09:51.490 --> 00:10:01.409
Um, but you will meet young people, idealistic people, kind of people who work at nonprofits, who are really don't want to participate in something that they don't agree with.
00:10:01.409 --> 00:10:04.370
So they may be having that reaction to AI.
00:10:04.370 --> 00:10:05.090
Uh,
00:10:05.090 --> 00:10:11.090
If you are a local person, I'm gonna stick a plug-in here of an amazing event that I read about in DC.
00:10:11.090 --> 00:10:17.970
The DC Public Library uh has is running a three-part series this month called AI and Us.
00:10:17.970 --> 00:10:21.889
How is AI impacting DC and what we can do about it?
00:10:21.889 --> 00:10:25.250
And their tagline is overwhelmed by AI.
00:10:25.250 --> 00:10:35.809
They're talking about building power among communities that are affected by AI, experiment and build connections, in real life connections to each other.
00:10:35.970 --> 00:10:44.529
And they're really focused on marginalized, uh, underserved, and what they're calling invisible communities.
00:10:44.529 --> 00:11:01.169
So many communities that we care about as nonprofits that are really being impacted by AI, but aren't part of policy making, aren't part of, you know, contacting their elected representatives, maybe don't totally know what AI is, need some AI literacy.
00:11:01.169 --> 00:11:03.330
So really interesting series.
00:11:03.330 --> 00:11:06.049
If you're in DC, I'll share that link with you.
00:11:06.049 --> 00:11:07.730
It's at the public library.
00:11:07.730 --> 00:11:08.370
And uh
00:11:08.370 --> 00:11:10.450
This Thursday, July 16th, Dr.
00:11:10.450 --> 00:11:15.649
Nicole Turner Lee from Brookings Institution is going to be speaking about AI equity.
00:11:15.649 --> 00:11:20.850
And she wrote the book Digitally Invisible, which if you haven't read it, I urge you to.
00:11:20.850 --> 00:11:22.210
It's amazing.
00:11:22.450 --> 00:11:25.570
If you're not in DC, I would say look around.
00:11:25.570 --> 00:11:52.450
You may have a similar uh lecture series or you know, webinars or local government, local organizations, non-government organizations who are interested in this and trying to build AI literacy, not just around the data centers, which there's a lot of that happening too, we've been talking about, but around the power imbalance, the AI literacy, the impact on communities, and particularly the impact on workforce.
00:11:52.450 --> 00:11:55.570
Um, and just the using our data.
00:11:55.730 --> 00:12:03.250
Like this is all stuff that we put on the internet about ourselves that then just got sucked up and monetized, basically.
00:12:03.250 --> 00:12:06.129
So these conversations are happening everywhere.
00:12:06.129 --> 00:12:07.649
Uh, show up in person.
00:12:07.649 --> 00:12:10.049
That's what this movement is all about.
00:12:10.370 --> 00:12:23.169
And this is really a bipartisan issue, demanding AI governance around data centers, around worker protections, around humanity-oriented AI.
00:12:23.169 --> 00:12:29.730
I wouldn't say it's uniting the parties in the United States, but it's definitely not something that's only in a fringe.
00:12:29.730 --> 00:12:40.850
There are very mainstream and centrist uh representatives and organizations that are also very concerned about the growth of AI.
00:12:40.850 --> 00:12:41.649
Uh,
00:12:41.649 --> 00:12:57.409
We covered it in this UN report that came out last week, the Papal Encyclical that we talked about a month or so ago, local zoning fights over data centers in you know, some very posh neighborhoods, so people are getting involved in those as well.
00:12:57.409 --> 00:13:05.490
If you've been to any of these zoning meetings, a lot of them are getting, you know, 200% turnout, have to be in the overflow room.
00:13:05.490 --> 00:13:08.610
So uh, you know, that's maybe a sign.
00:13:08.769 --> 00:13:09.970
Keep getting involved.
00:13:09.970 --> 00:13:11.889
Uh, people power is really changing.
00:13:11.889 --> 00:13:17.330
There are some more moratoriums coming online in different states and counties, even.
00:13:17.330 --> 00:13:26.529
Uh in Virginia, where I am in Northern Virginia, where there are so many data centers, there's a real movement around limiting them.
00:13:26.529 --> 00:13:28.370
So, People Power works.
00:13:28.370 --> 00:13:30.049
Uh, you want to get involved.
00:13:30.049 --> 00:13:31.570
It's uh fairly easy.
00:13:31.570 --> 00:13:35.330
Look around in your own location for groups that you can join.
00:13:35.570 --> 00:13:41.169
So, one Luddite critique of AI is about its environmental cost.
00:13:41.169 --> 00:13:44.529
And we've talked about that a lot, and it's a very legitimate concern.
00:13:44.529 --> 00:13:47.250
I'm not downplaying it for anyone.
00:13:47.250 --> 00:13:59.570
But as I've said several times, the picture from people in the environmental movement and conservation movement who use technology, there is a lot of, it's a lot more complicated.
00:13:59.570 --> 00:14:04.450
It's not just AI is bad for the climate end of story, black and white.
00:14:04.450 --> 00:14:04.929
So
00:14:04.929 --> 00:14:17.649
I wanted to share a research framework that I came across recently that is a way to help think through technology and climate, technology and AI and environment and conservation.
00:14:17.649 --> 00:14:25.809
So Jennifer Turliuk is a researcher, climate tech investor, and lecturer at MIT and Harvard.
00:14:25.809 --> 00:14:33.009
She is very well known in AI circles around this particular focus on environment.
00:14:33.009 --> 00:14:41.250
She wrote her MIT thesis on the net climate impact of AI, one of the first frameworks to look at both sides of the ledger.
00:14:41.250 --> 00:14:43.009
She's done a couple different TED talks.
00:14:43.009 --> 00:14:45.009
You can find her on YouTube.
00:14:45.009 --> 00:14:48.689
I will include some links to her work in the show notes.
00:14:48.850 --> 00:14:56.929
But her core argument is that most of the conversation right now around environmental impacts focuses on costs only.
00:14:56.929 --> 00:14:59.409
And costs are very real.
00:14:59.409 --> 00:15:07.250
Global electricity demand from data centers is expected to more than double by 2030, four years from now.
00:15:07.250 --> 00:15:12.450
And it will be roughly the same energy consumption of the country of Japan.
00:15:12.450 --> 00:15:18.610
About 60% of that increased demand is predicted to be met by fossil fuels.
00:15:18.610 --> 00:15:20.929
So that's an increase in fossil fuels.
00:15:20.929 --> 00:15:25.730
Well, we're been spending a couple of decades now trying to decrease fossil fuel use.
00:15:26.610 --> 00:15:47.409
Most people only count that operational carbon, so burning the fossil fuels, running the servers, and they don't also include the emissions from the buildings, from the generators that are being used in Northern Virginia, at least and in other parts of the country and around the world.
00:15:47.409 --> 00:15:56.450
When there isn't enough electricity on the grid, they will set up diesel generators, very stinky, noisy, and polluting.
00:15:56.450 --> 00:15:58.929
And then there's also emissions from building.
00:15:58.929 --> 00:16:10.450
So just you know, grinding up the concrete and all of the stuff that goes into building the data centers creates emissions, uh, which is bad for the environment and bad for pollution, bad for the neighborhoods.
00:16:10.769 --> 00:16:15.889
Turliuk makes the point that AI also has potential climate benefits.
00:16:15.889 --> 00:16:25.809
And so her framework tries to include those all of the cons and also include any pros that are out there.
00:16:25.809 --> 00:16:30.850
And so some of the pros that she's talked about are optimizing energy grids.
00:16:30.850 --> 00:16:44.450
So that's something that AI is really pretty good at doing, seeing the patterns and predicting when there'll be a lot of demand on the system and you know, feeding lots of inputs into it and then managing the system better.
00:16:44.450 --> 00:16:59.649
Of course, uh, also AI can do a good job of predicting solar and wind output if you have alternative energy sources, um, predictive maintenance on green infrastructure.
00:16:59.649 --> 00:17:04.690
Uh, you could use AI to help speed up permitting for renewable projects.
00:17:04.690 --> 00:17:06.690
You can use AI, you know,
00:17:06.690 --> 00:17:17.490
So there are lots of ways that AI could be useful in um decreasing the energy use of the data centers and of AI itself.
00:17:18.049 --> 00:17:24.930
The problem is that right now those costs are very concrete and immediate, and the benefits are largely potential.
00:17:24.930 --> 00:17:31.330
As she puts it, AI harms are measurable, but most benefits are still speculative.
00:17:31.810 --> 00:17:38.370
So if you're interested in checking out her net climate impact score framework, it's not a simple calculator.
00:17:38.370 --> 00:17:42.370
So you can't just put in, I want to use this AI to do this prompt and I live in this place.
00:17:42.370 --> 00:17:43.810
How much does it cost?
00:17:43.810 --> 00:17:51.970
Uh it doesn't work like that, although that would be a great tool to have, although it's very complicated, but something else that AI could help us with.
00:17:51.970 --> 00:17:53.250
Um, but
00:17:53.250 --> 00:18:02.210
Her framework, uh, it's a way to think through and weigh both sides of deploying AI for a given use case.
00:18:02.210 --> 00:18:05.650
Uh, it's useful, like governments are using it, can use it.
00:18:05.650 --> 00:18:08.850
I will share the link to it in the show notes.
00:18:08.850 --> 00:18:12.210
If you work at an environmental nonprofit, you may already be aware of this.
00:18:12.210 --> 00:18:12.530
But
00:18:12.530 --> 00:18:20.370
If you want to go deeper on it, she also has a cohort coming up in August, August 6th through 27th.
00:18:20.370 --> 00:18:22.850
It's you get a certificate at the end of it.
00:18:22.850 --> 00:18:24.770
It's uh $999.
00:18:24.770 --> 00:18:35.410
So it's not free, it's run through Maven, which is an online, you know, certification adult education professional development um platform.
00:18:35.410 --> 00:18:44.850
Uh, but it's a chance to, you know, work with this amazing um, you know, she's on the forefront of thinking about this specific aspect.
00:18:44.850 --> 00:18:56.130
So you could have a class with um Jennifer Tuliak, and um, you know, it's professional development specifically designed for climate and sustainability professionals.
00:18:56.370 --> 00:19:08.530
So if you are at a nonprofit that works in an environment, if you know somebody that you know could take this class and benefit from it from the certification, I will definitely share that link with you if you're interested.
00:19:08.850 --> 00:19:10.450
A question comes up a lot.
00:19:10.450 --> 00:19:16.370
Uh, how do I know which AI uses are having a high environmental cost?
00:19:16.370 --> 00:19:23.250
And are there any uses of AI that are quote unquote okay that have a lower environmental impact?
00:19:24.130 --> 00:19:27.890
And we've said before, you know, the companies make it pretty opaque.
00:19:27.890 --> 00:19:33.250
It's difficult to tell exactly what your prompt in your area is costing.
00:19:33.250 --> 00:19:44.690
I mean, even for pricing, they are not very transparent on what your usage is, what's a token, and is there like surge pricing, or you know, how are you being charged for this?
00:19:44.690 --> 00:19:48.370
So that's a totally different related piece of the problem.
00:19:48.370 --> 00:19:49.250
But
00:19:49.250 --> 00:19:56.050
I find it helpful to think through three different lenses about your AI use and make your own decisions.
00:19:56.050 --> 00:19:57.970
Like I keep saying, we have agency.
00:19:57.970 --> 00:20:00.450
So you can opt out entirely 100%.
00:20:00.450 --> 00:20:03.090
And that is a valid reaction.
00:20:03.330 --> 00:20:14.610
But if you're curious about using AI and thinking about using it in a way that feels okay to yourself, uh, one question to ask is what are you asking your AI to do?
00:20:14.610 --> 00:20:19.330
Not all AI use carries the same footprint.
00:20:19.330 --> 00:20:29.810
Text-based tasks like drafting an email, summarizing a meeting, analyzing, you know, asking it questions are relatively light.
00:20:29.810 --> 00:20:50.290
You know, asking Claude to do some web-based research for you or draft an article for you is um, you know, something you would probably be using that energy in some way, anyway, if you're doing your own uh web, you know, looking online for something, if you're Googling something, it's going through a data center.
00:20:51.010 --> 00:21:12.370
So text-based, you know, these are large language models, they're based on text, and the way they, you know, utilize the what they know, how they learn, how quickly they go to the data center, the packets that they come back with, the usage that you're getting charged for, that is all kind of like in their wheelhouse and is a very light application.
00:21:12.370 --> 00:21:12.930
Uh,
00:21:12.930 --> 00:21:23.170
Image generation, video generation is one of the most energy intensive uses for consumers, common people like us.
00:21:23.170 --> 00:21:31.810
Training a model is uh also a very heavy use of electricity and the AI.
00:21:31.810 --> 00:21:35.810
So using AI isn't just one thing.
00:21:35.970 --> 00:21:43.730
And you may have a project where you use it intensively to code something, but that's a one-time use for a couple of days or a week.
00:21:43.730 --> 00:21:57.170
And then after your agent is coded or your project is enabled or your skill has been created, your gem is working and functional, uh, then you you aren't coding it anymore.
00:21:57.170 --> 00:21:57.970
You're just using it.
00:21:57.970 --> 00:22:03.090
And so you're kind of migrated back into the lighter use of AI.
00:22:03.090 --> 00:22:03.730
Um,
00:22:03.730 --> 00:22:05.250
So that's one thing to think about.
00:22:05.250 --> 00:22:06.530
What are you using AI to do?
00:22:06.610 --> 00:22:10.530
And that may be something that your organization is going to look at putting in your policy.
00:22:10.530 --> 00:22:15.490
Like we don't use AI to generate images, but we do use it to summarize meetings.
00:22:15.490 --> 00:22:24.610
And we feel okay with the environmental impact that those different making that judgment call, uh, which is Okay, and which do we not do?
00:22:24.610 --> 00:22:25.410
Um,
00:22:25.410 --> 00:22:28.130
Another question to ask yourself is where are you?
00:22:28.130 --> 00:22:43.170
So we talked about this before, but most people don't know very well the carbon intensity of the electricity grid that you're on varies enormously by your location and by the time of day and by what else is going on.
00:22:43.170 --> 00:22:57.890
So if you're in the middle of a heat dome like we are having in the United States right now, that heat stress is going to make the electricity grid more stressed right when the data centers are trying to use that electricity too.
00:22:57.890 --> 00:22:58.370
So
00:22:58.370 --> 00:23:14.610
I could see in the future this becoming something more like you know, minutes used to be on your data plan for your cell phone, that uh nonprofits would feel better about doing these more intensive tasks overnight, you know, when a lot of other people aren't using the electricity grid.
00:23:14.610 --> 00:23:16.450
So that's something to think about.
00:23:16.450 --> 00:23:20.290
Um, you don't control which data center your query goes to.
00:23:20.610 --> 00:23:24.690
You can ask vendors about their renewable energy commitments.
00:23:24.690 --> 00:23:30.450
You can think about where you're sited, like what state are you in, and how is their electricity grid?
00:23:30.450 --> 00:23:31.970
How is their policy regulation?
00:23:31.970 --> 00:23:33.890
Are they a highly regulated state?
00:23:33.890 --> 00:23:36.130
Do they have lots of renewable energy?
00:23:36.130 --> 00:23:42.610
Uh are they a state where it's people with a solar panel on their house can sell that back to the grid?
00:23:42.610 --> 00:23:44.770
Uh in Virginia, we cannot do that.
00:23:44.770 --> 00:23:52.850
We can you can have a solar panel, but it doesn't that doesn't give you it, you can't sell it back to other people to be able to use it yet.
00:23:52.850 --> 00:23:55.170
Another thing to lobby about.
00:23:55.170 --> 00:23:56.530
But um, you know,
00:23:56.530 --> 00:24:05.490
It's something to think about and to build into your organization's individual AI acceptable use policy, right?
00:24:05.490 --> 00:24:29.730
You might be an organization working on the environment in a drought-stricken water poor area of the state or the country or just your town where your physical address is, and you decide for that reason you aren't going to use AI because it's gonna put a lot of stress on, you know, data centers are being built in places that they shouldn't be built near you.
00:24:30.290 --> 00:24:37.490
You may be working in a location that has a lot of water and has a lot of renewables and has a lot of regulation.
00:24:37.490 --> 00:24:45.330
And so you might feel more okay as an organization living your values using AI in that situation.
00:24:45.330 --> 00:24:48.690
So, again, it's gonna be really individualized to you.
00:24:48.690 --> 00:24:55.490
If you have a lot of people working remotely, you know, maybe that's something you want to talk about as an organization as well.
00:24:55.490 --> 00:25:01.970
What is the acceptable AI use uh for your staff who work remotely based on where they are?
00:25:01.970 --> 00:25:03.730
Um
00:25:03.730 --> 00:25:12.290
In Turliuk's framework, she also asks us to think about what's on the other side of the ledger.
00:25:12.290 --> 00:25:17.250
So you've thought about what are you going to use it for?
00:25:17.250 --> 00:25:23.570
A light thing or a heavier thing that takes a lot more energy, a lot more AI use.
00:25:23.570 --> 00:25:29.170
You've thought about the electricity grid, where are you, the environment, where are you?
00:25:30.690 --> 00:25:34.770
Turliuk also wants us to take into account those pros, right?
00:25:34.770 --> 00:25:39.570
To in addition to the cons and think about that offset.
00:25:39.570 --> 00:25:49.410
So if there is a potential benefit of your use of AI, you need to think about that and calculate that into your calculation, basically.
00:25:49.570 --> 00:26:15.410
So if you are drafting a grant to fund a smart grid project and you're using AI to do it in a you know efficient way, um, that's a different calculation than generating some AI art for your gala invitation or for a LinkedIn post, for example, making a cute little video that is about something that's so ephemeral on social media.
00:26:16.130 --> 00:26:25.330
So that might also be something you want to build into your policy and think about and talk about as an organization with your values in mind.
00:26:25.330 --> 00:26:34.770
So you don't have to do this kind of deep thought about every prompt, and you don't need to feel guilty about using AI to help manage your inbox, for example.
00:26:34.770 --> 00:26:41.730
You know, productivity tools that you're using are also, you know, usually pretty light ask of the AI.
00:26:42.130 --> 00:26:47.970
But it's worth asking for your policy as a mission-based organization.
00:26:47.970 --> 00:26:53.810
Is your AI policy aligned with that mission, with the communities that you care about?
00:26:53.810 --> 00:26:56.210
Uh, or is it just something that's convenient?
00:26:56.370 --> 00:27:11.730
Is it something like you might want to have a talk with the all staff about if it's something that you can do, like putting something on your Asana, for example, or your Monday, um, you could ask your AI to do it for you.
00:27:11.730 --> 00:27:24.370
But you know, if that's sending a packet out to the data center and getting it back and doing it, um, if it's something that you can just pop over and plop it in yourself and you're done, like that wasn't something you used to do and you're able to do.
00:27:24.610 --> 00:27:41.810
Maybe you want to think more about using AI for the things that you can't do easily and offsetting that use by doing things yourself where it makes sense, where it would just be like, why would I ask AI to do that?
00:27:41.810 --> 00:27:42.930
I can just do it.
00:27:42.930 --> 00:27:44.770
So, another way to think about it.
00:27:44.770 --> 00:27:53.410
So, trying to keep in mind those pros, the things that you're using, the benefits of your using of AI.
00:27:53.410 --> 00:27:53.970
You know,
00:27:53.970 --> 00:28:03.410
Maybe you are using AI to um handle part of three other people's jobs that you're doing at your nonprofit because you're chronically understaffed.
00:28:03.410 --> 00:28:07.410
That's a valid way to use AI and reach more people.
00:28:07.410 --> 00:28:18.370
Like your mission is to serve more people, and you can do that because AI is helping you with some of these day-to-day tasks or productivity tasks, then build that into your calculations.
00:28:18.370 --> 00:28:20.130
You have a benefit.
00:28:20.530 --> 00:28:29.650
You will probably still have people who say, I'm just against AI, I don't want to learn about it, I don't want to use it, it's terrible for the planet, and I'm just very upset about it.
00:28:29.650 --> 00:28:36.530
Um, and those are the people who really care about these kinds of questions, like that have a lot of nuance to them.
00:28:36.530 --> 00:28:43.410
It's not irrational to be worried about and afraid of and in opposition to AI.
00:28:43.410 --> 00:28:52.050
Um, the original Luddites weren't against technology, they were against their technology being stolen from them and used against them.
00:28:52.450 --> 00:29:07.730
So in 2026, I think that's a good thought to leave us with about how it's implemented, who benefits, and you know, can we use this to make the world better?
00:29:07.730 --> 00:29:20.290
And how involved do we need to be in real life to ensure that those policies and regulations and human-focused AI become reality?
00:29:20.290 --> 00:29:21.570
So
00:29:21.570 --> 00:29:26.130
I want to remind you we do have a template for an AI acceptable use policy.
00:29:26.130 --> 00:29:28.450
It's at our site, communityit.com.
00:29:28.450 --> 00:29:30.370
I'll include that in the show notes.
00:29:30.370 --> 00:29:32.930
Again, it's you can't just download it and use it.
00:29:32.930 --> 00:29:38.290
You do have to customize it and make it individual to your organization.
00:29:38.450 --> 00:29:42.850
And some of these questions you're gonna have to think about just even individually to yourself.
00:29:42.850 --> 00:29:47.330
What do you feel comfortable with using AI for and why?
00:29:47.330 --> 00:29:49.570
And share that with your colleagues.
00:29:49.570 --> 00:29:50.450
Um,
00:29:50.450 --> 00:29:54.850
You can find us on Reddit at r slash nonprofit IT Management.
00:29:54.850 --> 00:29:57.410
We have a webinar this week, tomorrow.
00:29:57.410 --> 00:30:01.730
It's on an AI maturity model for nonprofits.
00:30:01.730 --> 00:30:09.090
Um, so if you are at a nonprofit that's been using AI for a while, a lot of the material out there is for beginners.
00:30:09.090 --> 00:30:17.250
So this is a webinar that's gonna talk about kind of next steps, how to really deeply integrate AI into what your nonprofit does.
00:30:17.250 --> 00:30:24.210
We have some experts who are gonna be with us talking about some case studies and real life experience that they've had.
00:30:24.210 --> 00:30:26.690
If you're interested in that, please join us.
00:30:26.690 --> 00:30:27.890
You can still register.
00:30:27.890 --> 00:30:30.370
It'll be tomorrow on Wednesday.
00:30:30.610 --> 00:30:39.170
And then I'll be back here on Friday with some more nonprofit uh podcasts and next Tuesday with some more nonprofit AI.
00:30:39.170 --> 00:30:40.930
So until then, take care.
00:00:22.689 --> 00:00:30.289
Hello and welcome to the Community IT Innovators Nonprofit AI Podcast, Midweek Check-In.
00:00:30.530 --> 00:00:32.689
I'm Carolyn Woodard, your host.
00:00:32.689 --> 00:00:35.090
I am not an AI expert.
00:00:35.090 --> 00:00:49.570
I'm really interested in AI and in the impact that AI is having on nonprofits, both how we're using it at nonprofits and how it's impacting the work that we do in the communities that we care about.
00:00:49.810 --> 00:01:11.329
So in this midweek podcast, every week I try to bring us a little bit of news, filtered through how this is going to impact nonprofits and philanthropy more broadly, and maybe answer some questions and give you some more resources that you can read and follow up on because it's hard with so much coming at us about AI every day.
00:01:11.329 --> 00:01:15.569
If you're reading about it, it's that big fire hose of information.
00:01:15.569 --> 00:01:18.929
So just trying to make it a little bit more manageable.
00:01:18.929 --> 00:01:19.340
So
00:01:19.340 --> 00:01:30.369
Please get in touch with me if you have something you'd like to hear about or something you have a question about, you can reach me through our website, communityit.com, or here on the podcast.
00:01:30.369 --> 00:01:31.649
So
00:01:31.649 --> 00:01:52.129
This week had a couple of resources and news stories that connect in this kind of odd way around a kind of cultural moment that is building, as we speak, a lot of resistance to AI among workers, young people, a lot of nonprofits.
00:01:52.129 --> 00:02:18.289
I'm seeing a lot on LinkedIn, and our clients at Community IT are also skeptical, and many are very resistant to using AI at all, or just very concerned about the power imbalance, the environmental impacts, and the impact on people and humanity for going to use the easy way out instead of writing our own essays or making our own art.
00:02:18.770 --> 00:02:28.370
So I wanted to share a couple of resources with you this week about AI's real effect on climate costs and kind of a framework for how to think about that.
00:02:28.610 --> 00:02:34.129
And also, yeah, this bigger question of resistance like, do you have to use AI?
00:02:34.129 --> 00:02:42.210
What if your company is telling you that you have to, or your organization has a policy and is expecting everyone to use AI?
00:02:42.210 --> 00:02:43.009
So
00:02:43.009 --> 00:02:58.370
The first was this uh news story about an eight-day festival in New York City that I came across called The Summer of Ludd, which is uh feeding into the original Luddite movement.
00:02:58.370 --> 00:03:04.210
Um, there was a conference at the new school in New York on AI and surveillance.
00:03:04.210 --> 00:03:12.289
Uh there were, you know, other activities, kind of in real life activities at this festival.
00:03:12.289 --> 00:03:15.329
There were some concerts in real life music.
00:03:15.329 --> 00:03:17.729
Uh, none of it was advertised online.
00:03:17.729 --> 00:03:20.769
There were flyers, so it was very old school.
00:03:20.769 --> 00:03:24.849
They were pasted around the neighborhood in New York where this took place.
00:03:24.849 --> 00:03:25.169
And
00:03:25.169 --> 00:03:34.769
I came across it through an NPR story about this festival, but also how the word Luddite, if you know it at all, is a political insult.
00:03:34.769 --> 00:03:39.169
It doesn't really relate to historically who the Luddites were.
00:03:39.329 --> 00:03:42.849
So a little bit of a reframe of the history.
00:03:42.849 --> 00:03:59.329
Most people use Luddite as an insult or shorthand for someone who can't work a smartphone or who prefers analog to digital, who likes old school, who likes doing things with their hands, maybe arts and crafts that are not related to technology.
00:03:59.329 --> 00:04:06.609
Sometimes even it's used, you know, for technophobe, someone who's afraid of technology, just doesn't like it, doesn't want to use it.
00:04:06.930 --> 00:04:14.289
But actual Luddites were skilled textile workers in 19th century England who, they weren't against technology.
00:04:14.289 --> 00:04:35.569
They knew what they were doing around the way the looms were working and the technology, these kind of early computers that were these reading cards that helps the loom do the work and make these very complicated patterns, which until then it was only a master weaver who would be able to, you know, handle this kind of complicated pattern.
00:04:35.569 --> 00:04:36.530
So they,
00:04:36.530 --> 00:04:45.250
The Luddites originally were against the way technology was being implemented to enrich factory owners and lay off staff.
00:04:45.250 --> 00:04:56.290
So if you have an early computer running your looms, you don't need those master weavers anymore to set up the loom or, you know, think up the patterns.
00:04:56.290 --> 00:04:58.209
There's lots of new patterns that you can do.
00:04:58.209 --> 00:05:16.129
And really all you need is the one person who punches out the cards and then you put them in the loom machine and it makes the people, the children, still had to run the looms, but the design part of it was taken out of the hand of people and put in the hands of technology.
00:05:16.129 --> 00:05:17.569
And this was, you know,
00:05:17.569 --> 00:05:20.370
The factory owners were thought this was a great idea.
00:05:20.370 --> 00:05:27.009
They could pay fewer people and they could just have the quote-unquote computer doing the setup.
00:05:27.250 --> 00:05:45.649
And the people, of course, who were the crafts people who knew it was like their experience and their understanding of weaving that enabled people to make the computer program that would do the same thing that they could do, but cheaper because it was a computer instead.
00:05:45.810 --> 00:05:57.649
So they petitioned, they marched, they organized, they had strikes, uh, they won some concessions, but the winner of history gets to describe the losers.
00:05:57.649 --> 00:06:04.050
So the fact that this term Luddites, they followed someone, you know, named Ludd.
00:06:04.050 --> 00:06:05.730
And so they were Luddites.
00:06:05.810 --> 00:06:17.170
And at the time it was kind of the ruling class and the factory owners were afraid of the people power that was being developed by the followers of Ludd, the Luddites.
00:06:17.170 --> 00:06:21.089
And so in history, they insulted them.
00:06:21.089 --> 00:06:28.689
They said they were backward, they didn't want progress, they were afraid of the technology, and that is how it's come down to us now.
00:06:28.689 --> 00:06:32.290
I'll share some resources with you in the show notes.
00:06:32.449 --> 00:06:38.930
If you want to go deeper into the history, uh, the book to read is Blood in the Machine by Brian Merchant.
00:06:38.930 --> 00:07:06.689
And he makes this argument that the Luddite story is deliberately distorting that history by people in power who gain from us believing that anyone who is against technology doesn't understand technology or just doesn't know what the tool does, or like wasn't smart enough to be an investor, or you know, is unable to understand the technology and they're a Luddite because they want something different.
00:07:06.850 --> 00:07:29.250
And so he's arguing for recapturing that term and that ethos of wanting technology to enhance humanity and to serve humanity instead of take over from humanity and lead to the wealth of a concentrated few people in power who were already wealthy.
00:07:29.250 --> 00:07:31.730
Uh, he does have a Substack as well.
00:07:31.730 --> 00:07:33.969
I'll put that in the show notes also.
00:07:34.370 --> 00:07:37.569
So you may be seeing this in your own nonprofits.
00:07:37.569 --> 00:07:46.449
And so I just kind of pulled together some of these threads that we've been talking about over the past couple of months here, uh, especially younger staff.
00:07:46.449 --> 00:08:00.209
Uh, if you have teenagers in your house, college students, or young workers at your nonprofit, people in their 20s and 30s, particularly, are pushing back on AI in real tangible ways.
00:08:00.209 --> 00:08:07.569
And this is showing up in polling and some other research that's coming out about generational attitudes toward AI.
00:08:07.810 --> 00:08:14.290
You've probably seen things about this as well, going to IRL in real life events like taking up knitting.
00:08:14.290 --> 00:08:24.370
There's like granny cool, uh granny hobbies that are coming back, spending time with people, especially after COVID and a lot of isolation with technology.
00:08:24.370 --> 00:08:35.170
A lot of younger people who came through that find it very important to make social connections to other people, to meet people, to find people who like your same hobbies.
00:08:35.170 --> 00:08:48.850
And you may have seen articles about using dumb phones or apps on your phone that can manage how addictive they are, how much you're using them, and you know, make you go outside and touch grass.
00:08:49.009 --> 00:08:53.330
You may have staff in this age group or in other areas, I mean, it's not just young people.
00:08:53.330 --> 00:08:59.330
There are plenty of other people who are also deeply skeptical about AI and about technology.
00:08:59.649 --> 00:09:02.929
So you may have people who say, you know, they're just not going to use it.
00:09:02.929 --> 00:09:16.929
And if you're tempted to see that as a training problem that you can quote unquote fix, it might be worth it at your organization or in your family to go a little bit deeper and ask some more probing questions.
00:09:16.929 --> 00:09:18.049
What are they saying?
00:09:18.049 --> 00:09:19.649
Where is this coming from?
00:09:19.649 --> 00:09:22.850
And what, you know, what's behind it?
00:09:23.009 --> 00:09:27.250
Are they against AI for specific reasons?
00:09:27.250 --> 00:09:31.330
Are they concerned about who benefits and on what terms?
00:09:31.330 --> 00:09:34.850
A lot of people want to live their values.
00:09:34.850 --> 00:09:38.850
So they don't want to shop at a place that they don't agree with.
00:09:38.850 --> 00:09:44.529
They don't want to go to a restaurant that maybe has a policy stand that they don't agree with.
00:09:44.529 --> 00:09:51.490
They want to try where possible, and it can be uh it can be difficult to do this completely.
00:09:51.490 --> 00:10:01.409
Um, but you will meet young people, idealistic people, kind of people who work at nonprofits, who are really don't want to participate in something that they don't agree with.
00:10:01.409 --> 00:10:04.370
So they may be having that reaction to AI.
00:10:04.370 --> 00:10:05.090
Uh,
00:10:05.090 --> 00:10:11.090
If you are a local person, I'm gonna stick a plug-in here of an amazing event that I read about in DC.
00:10:11.090 --> 00:10:17.970
The DC Public Library uh has is running a three-part series this month called AI and Us.
00:10:17.970 --> 00:10:21.889
How is AI impacting DC and what we can do about it?
00:10:21.889 --> 00:10:25.250
And their tagline is overwhelmed by AI.
00:10:25.250 --> 00:10:35.809
They're talking about building power among communities that are affected by AI, experiment and build connections, in real life connections to each other.
00:10:35.970 --> 00:10:44.529
And they're really focused on marginalized, uh, underserved, and what they're calling invisible communities.
00:10:44.529 --> 00:11:01.169
So many communities that we care about as nonprofits that are really being impacted by AI, but aren't part of policy making, aren't part of, you know, contacting their elected representatives, maybe don't totally know what AI is, need some AI literacy.
00:11:01.169 --> 00:11:03.330
So really interesting series.
00:11:03.330 --> 00:11:06.049
If you're in DC, I'll share that link with you.
00:11:06.049 --> 00:11:07.730
It's at the public library.
00:11:07.730 --> 00:11:08.370
And uh
00:11:08.370 --> 00:11:10.450
This Thursday, July 16th, Dr.
00:11:10.450 --> 00:11:15.649
Nicole Turner Lee from Brookings Institution is going to be speaking about AI equity.
00:11:15.649 --> 00:11:20.850
And she wrote the book Digitally Invisible, which if you haven't read it, I urge you to.
00:11:20.850 --> 00:11:22.210
It's amazing.
00:11:22.450 --> 00:11:25.570
If you're not in DC, I would say look around.
00:11:25.570 --> 00:11:52.450
You may have a similar uh lecture series or you know, webinars or local government, local organizations, non-government organizations who are interested in this and trying to build AI literacy, not just around the data centers, which there's a lot of that happening too, we've been talking about, but around the power imbalance, the AI literacy, the impact on communities, and particularly the impact on workforce.
00:11:52.450 --> 00:11:55.570
Um, and just the using our data.
00:11:55.730 --> 00:12:03.250
Like this is all stuff that we put on the internet about ourselves that then just got sucked up and monetized, basically.
00:12:03.250 --> 00:12:06.129
So these conversations are happening everywhere.
00:12:06.129 --> 00:12:07.649
Uh, show up in person.
00:12:07.649 --> 00:12:10.049
That's what this movement is all about.
00:12:10.370 --> 00:12:23.169
And this is really a bipartisan issue, demanding AI governance around data centers, around worker protections, around humanity-oriented AI.
00:12:23.169 --> 00:12:29.730
I wouldn't say it's uniting the parties in the United States, but it's definitely not something that's only in a fringe.
00:12:29.730 --> 00:12:40.850
There are very mainstream and centrist uh representatives and organizations that are also very concerned about the growth of AI.
00:12:40.850 --> 00:12:41.649
Uh,
00:12:41.649 --> 00:12:57.409
We covered it in this UN report that came out last week, the Papal Encyclical that we talked about a month or so ago, local zoning fights over data centers in you know, some very posh neighborhoods, so people are getting involved in those as well.
00:12:57.409 --> 00:13:05.490
If you've been to any of these zoning meetings, a lot of them are getting, you know, 200% turnout, have to be in the overflow room.
00:13:05.490 --> 00:13:08.610
So uh, you know, that's maybe a sign.
00:13:08.769 --> 00:13:09.970
Keep getting involved.
00:13:09.970 --> 00:13:11.889
Uh, people power is really changing.
00:13:11.889 --> 00:13:17.330
There are some more moratoriums coming online in different states and counties, even.
00:13:17.330 --> 00:13:26.529
Uh in Virginia, where I am in Northern Virginia, where there are so many data centers, there's a real movement around limiting them.
00:13:26.529 --> 00:13:28.370
So, People Power works.
00:13:28.370 --> 00:13:30.049
Uh, you want to get involved.
00:13:30.049 --> 00:13:31.570
It's uh fairly easy.
00:13:31.570 --> 00:13:35.330
Look around in your own location for groups that you can join.
00:13:35.570 --> 00:13:41.169
So, one Luddite critique of AI is about its environmental cost.
00:13:41.169 --> 00:13:44.529
And we've talked about that a lot, and it's a very legitimate concern.
00:13:44.529 --> 00:13:47.250
I'm not downplaying it for anyone.
00:13:47.250 --> 00:13:59.570
But as I've said several times, the picture from people in the environmental movement and conservation movement who use technology, there is a lot of, it's a lot more complicated.
00:13:59.570 --> 00:14:04.450
It's not just AI is bad for the climate end of story, black and white.
00:14:04.450 --> 00:14:04.929
So
00:14:04.929 --> 00:14:17.649
I wanted to share a research framework that I came across recently that is a way to help think through technology and climate, technology and AI and environment and conservation.
00:14:17.649 --> 00:14:25.809
So Jennifer Turliuk is a researcher, climate tech investor, and lecturer at MIT and Harvard.
00:14:25.809 --> 00:14:33.009
She is very well known in AI circles around this particular focus on environment.
00:14:33.009 --> 00:14:41.250
She wrote her MIT thesis on the net climate impact of AI, one of the first frameworks to look at both sides of the ledger.
00:14:41.250 --> 00:14:43.009
She's done a couple different TED talks.
00:14:43.009 --> 00:14:45.009
You can find her on YouTube.
00:14:45.009 --> 00:14:48.689
I will include some links to her work in the show notes.
00:14:48.850 --> 00:14:56.929
But her core argument is that most of the conversation right now around environmental impacts focuses on costs only.
00:14:56.929 --> 00:14:59.409
And costs are very real.
00:14:59.409 --> 00:15:07.250
Global electricity demand from data centers is expected to more than double by 2030, four years from now.
00:15:07.250 --> 00:15:12.450
And it will be roughly the same energy consumption of the country of Japan.
00:15:12.450 --> 00:15:18.610
About 60% of that increased demand is predicted to be met by fossil fuels.
00:15:18.610 --> 00:15:20.929
So that's an increase in fossil fuels.
00:15:20.929 --> 00:15:25.730
Well, we're been spending a couple of decades now trying to decrease fossil fuel use.
00:15:26.610 --> 00:15:47.409
Most people only count that operational carbon, so burning the fossil fuels, running the servers, and they don't also include the emissions from the buildings, from the generators that are being used in Northern Virginia, at least and in other parts of the country and around the world.
00:15:47.409 --> 00:15:56.450
When there isn't enough electricity on the grid, they will set up diesel generators, very stinky, noisy, and polluting.
00:15:56.450 --> 00:15:58.929
And then there's also emissions from building.
00:15:58.929 --> 00:16:10.450
So just you know, grinding up the concrete and all of the stuff that goes into building the data centers creates emissions, uh, which is bad for the environment and bad for pollution, bad for the neighborhoods.
00:16:10.769 --> 00:16:15.889
Turliuk makes the point that AI also has potential climate benefits.
00:16:15.889 --> 00:16:25.809
And so her framework tries to include those all of the cons and also include any pros that are out there.
00:16:25.809 --> 00:16:30.850
And so some of the pros that she's talked about are optimizing energy grids.
00:16:30.850 --> 00:16:44.450
So that's something that AI is really pretty good at doing, seeing the patterns and predicting when there'll be a lot of demand on the system and you know, feeding lots of inputs into it and then managing the system better.
00:16:44.450 --> 00:16:59.649
Of course, uh, also AI can do a good job of predicting solar and wind output if you have alternative energy sources, um, predictive maintenance on green infrastructure.
00:16:59.649 --> 00:17:04.690
Uh, you could use AI to help speed up permitting for renewable projects.
00:17:04.690 --> 00:17:06.690
You can use AI, you know,
00:17:06.690 --> 00:17:17.490
So there are lots of ways that AI could be useful in um decreasing the energy use of the data centers and of AI itself.
00:17:18.049 --> 00:17:24.930
The problem is that right now those costs are very concrete and immediate, and the benefits are largely potential.
00:17:24.930 --> 00:17:31.330
As she puts it, AI harms are measurable, but most benefits are still speculative.
00:17:31.810 --> 00:17:38.370
So if you're interested in checking out her net climate impact score framework, it's not a simple calculator.
00:17:38.370 --> 00:17:42.370
So you can't just put in, I want to use this AI to do this prompt and I live in this place.
00:17:42.370 --> 00:17:43.810
How much does it cost?
00:17:43.810 --> 00:17:51.970
Uh it doesn't work like that, although that would be a great tool to have, although it's very complicated, but something else that AI could help us with.
00:17:51.970 --> 00:17:53.250
Um, but
00:17:53.250 --> 00:18:02.210
Her framework, uh, it's a way to think through and weigh both sides of deploying AI for a given use case.
00:18:02.210 --> 00:18:05.650
Uh, it's useful, like governments are using it, can use it.
00:18:05.650 --> 00:18:08.850
I will share the link to it in the show notes.
00:18:08.850 --> 00:18:12.210
If you work at an environmental nonprofit, you may already be aware of this.
00:18:12.210 --> 00:18:12.530
But
00:18:12.530 --> 00:18:20.370
If you want to go deeper on it, she also has a cohort coming up in August, August 6th through 27th.
00:18:20.370 --> 00:18:22.850
It's you get a certificate at the end of it.
00:18:22.850 --> 00:18:24.770
It's uh $999.
00:18:24.770 --> 00:18:35.410
So it's not free, it's run through Maven, which is an online, you know, certification adult education professional development um platform.
00:18:35.410 --> 00:18:44.850
Uh, but it's a chance to, you know, work with this amazing um, you know, she's on the forefront of thinking about this specific aspect.
00:18:44.850 --> 00:18:56.130
So you could have a class with um Jennifer Tuliak, and um, you know, it's professional development specifically designed for climate and sustainability professionals.
00:18:56.370 --> 00:19:08.530
So if you are at a nonprofit that works in an environment, if you know somebody that you know could take this class and benefit from it from the certification, I will definitely share that link with you if you're interested.
00:19:08.850 --> 00:19:10.450
A question comes up a lot.
00:19:10.450 --> 00:19:16.370
Uh, how do I know which AI uses are having a high environmental cost?
00:19:16.370 --> 00:19:23.250
And are there any uses of AI that are quote unquote okay that have a lower environmental impact?
00:19:24.130 --> 00:19:27.890
And we've said before, you know, the companies make it pretty opaque.
00:19:27.890 --> 00:19:33.250
It's difficult to tell exactly what your prompt in your area is costing.
00:19:33.250 --> 00:19:44.690
I mean, even for pricing, they are not very transparent on what your usage is, what's a token, and is there like surge pricing, or you know, how are you being charged for this?
00:19:44.690 --> 00:19:48.370
So that's a totally different related piece of the problem.
00:19:48.370 --> 00:19:49.250
But
00:19:49.250 --> 00:19:56.050
I find it helpful to think through three different lenses about your AI use and make your own decisions.
00:19:56.050 --> 00:19:57.970
Like I keep saying, we have agency.
00:19:57.970 --> 00:20:00.450
So you can opt out entirely 100%.
00:20:00.450 --> 00:20:03.090
And that is a valid reaction.
00:20:03.330 --> 00:20:14.610
But if you're curious about using AI and thinking about using it in a way that feels okay to yourself, uh, one question to ask is what are you asking your AI to do?
00:20:14.610 --> 00:20:19.330
Not all AI use carries the same footprint.
00:20:19.330 --> 00:20:29.810
Text-based tasks like drafting an email, summarizing a meeting, analyzing, you know, asking it questions are relatively light.
00:20:29.810 --> 00:20:50.290
You know, asking Claude to do some web-based research for you or draft an article for you is um, you know, something you would probably be using that energy in some way, anyway, if you're doing your own uh web, you know, looking online for something, if you're Googling something, it's going through a data center.
00:20:51.010 --> 00:21:12.370
So text-based, you know, these are large language models, they're based on text, and the way they, you know, utilize the what they know, how they learn, how quickly they go to the data center, the packets that they come back with, the usage that you're getting charged for, that is all kind of like in their wheelhouse and is a very light application.
00:21:12.370 --> 00:21:12.930
Uh,
00:21:12.930 --> 00:21:23.170
Image generation, video generation is one of the most energy intensive uses for consumers, common people like us.
00:21:23.170 --> 00:21:31.810
Training a model is uh also a very heavy use of electricity and the AI.
00:21:31.810 --> 00:21:35.810
So using AI isn't just one thing.
00:21:35.970 --> 00:21:43.730
And you may have a project where you use it intensively to code something, but that's a one-time use for a couple of days or a week.
00:21:43.730 --> 00:21:57.170
And then after your agent is coded or your project is enabled or your skill has been created, your gem is working and functional, uh, then you you aren't coding it anymore.
00:21:57.170 --> 00:21:57.970
You're just using it.
00:21:57.970 --> 00:22:03.090
And so you're kind of migrated back into the lighter use of AI.
00:22:03.090 --> 00:22:03.730
Um,
00:22:03.730 --> 00:22:05.250
So that's one thing to think about.
00:22:05.250 --> 00:22:06.530
What are you using AI to do?
00:22:06.610 --> 00:22:10.530
And that may be something that your organization is going to look at putting in your policy.
00:22:10.530 --> 00:22:15.490
Like we don't use AI to generate images, but we do use it to summarize meetings.
00:22:15.490 --> 00:22:24.610
And we feel okay with the environmental impact that those different making that judgment call, uh, which is Okay, and which do we not do?
00:22:24.610 --> 00:22:25.410
Um,
00:22:25.410 --> 00:22:28.130
Another question to ask yourself is where are you?
00:22:28.130 --> 00:22:43.170
So we talked about this before, but most people don't know very well the carbon intensity of the electricity grid that you're on varies enormously by your location and by the time of day and by what else is going on.
00:22:43.170 --> 00:22:57.890
So if you're in the middle of a heat dome like we are having in the United States right now, that heat stress is going to make the electricity grid more stressed right when the data centers are trying to use that electricity too.
00:22:57.890 --> 00:22:58.370
So
00:22:58.370 --> 00:23:14.610
I could see in the future this becoming something more like you know, minutes used to be on your data plan for your cell phone, that uh nonprofits would feel better about doing these more intensive tasks overnight, you know, when a lot of other people aren't using the electricity grid.
00:23:14.610 --> 00:23:16.450
So that's something to think about.
00:23:16.450 --> 00:23:20.290
Um, you don't control which data center your query goes to.
00:23:20.610 --> 00:23:24.690
You can ask vendors about their renewable energy commitments.
00:23:24.690 --> 00:23:30.450
You can think about where you're sited, like what state are you in, and how is their electricity grid?
00:23:30.450 --> 00:23:31.970
How is their policy regulation?
00:23:31.970 --> 00:23:33.890
Are they a highly regulated state?
00:23:33.890 --> 00:23:36.130
Do they have lots of renewable energy?
00:23:36.130 --> 00:23:42.610
Uh are they a state where it's people with a solar panel on their house can sell that back to the grid?
00:23:42.610 --> 00:23:44.770
Uh in Virginia, we cannot do that.
00:23:44.770 --> 00:23:52.850
We can you can have a solar panel, but it doesn't that doesn't give you it, you can't sell it back to other people to be able to use it yet.
00:23:52.850 --> 00:23:55.170
Another thing to lobby about.
00:23:55.170 --> 00:23:56.530
But um, you know,
00:23:56.530 --> 00:24:05.490
It's something to think about and to build into your organization's individual AI acceptable use policy, right?
00:24:05.490 --> 00:24:29.730
You might be an organization working on the environment in a drought-stricken water poor area of the state or the country or just your town where your physical address is, and you decide for that reason you aren't going to use AI because it's gonna put a lot of stress on, you know, data centers are being built in places that they shouldn't be built near you.
00:24:30.290 --> 00:24:37.490
You may be working in a location that has a lot of water and has a lot of renewables and has a lot of regulation.
00:24:37.490 --> 00:24:45.330
And so you might feel more okay as an organization living your values using AI in that situation.
00:24:45.330 --> 00:24:48.690
So, again, it's gonna be really individualized to you.
00:24:48.690 --> 00:24:55.490
If you have a lot of people working remotely, you know, maybe that's something you want to talk about as an organization as well.
00:24:55.490 --> 00:25:01.970
What is the acceptable AI use uh for your staff who work remotely based on where they are?
00:25:01.970 --> 00:25:03.730
Um
00:25:03.730 --> 00:25:12.290
In Turliuk's framework, she also asks us to think about what's on the other side of the ledger.
00:25:12.290 --> 00:25:17.250
So you've thought about what are you going to use it for?
00:25:17.250 --> 00:25:23.570
A light thing or a heavier thing that takes a lot more energy, a lot more AI use.
00:25:23.570 --> 00:25:29.170
You've thought about the electricity grid, where are you, the environment, where are you?
00:25:30.690 --> 00:25:34.770
Turliuk also wants us to take into account those pros, right?
00:25:34.770 --> 00:25:39.570
To in addition to the cons and think about that offset.
00:25:39.570 --> 00:25:49.410
So if there is a potential benefit of your use of AI, you need to think about that and calculate that into your calculation, basically.
00:25:49.570 --> 00:26:15.410
So if you are drafting a grant to fund a smart grid project and you're using AI to do it in a you know efficient way, um, that's a different calculation than generating some AI art for your gala invitation or for a LinkedIn post, for example, making a cute little video that is about something that's so ephemeral on social media.
00:26:16.130 --> 00:26:25.330
So that might also be something you want to build into your policy and think about and talk about as an organization with your values in mind.
00:26:25.330 --> 00:26:34.770
So you don't have to do this kind of deep thought about every prompt, and you don't need to feel guilty about using AI to help manage your inbox, for example.
00:26:34.770 --> 00:26:41.730
You know, productivity tools that you're using are also, you know, usually pretty light ask of the AI.
00:26:42.130 --> 00:26:47.970
But it's worth asking for your policy as a mission-based organization.
00:26:47.970 --> 00:26:53.810
Is your AI policy aligned with that mission, with the communities that you care about?
00:26:53.810 --> 00:26:56.210
Uh, or is it just something that's convenient?
00:26:56.370 --> 00:27:11.730
Is it something like you might want to have a talk with the all staff about if it's something that you can do, like putting something on your Asana, for example, or your Monday, um, you could ask your AI to do it for you.
00:27:11.730 --> 00:27:24.370
But you know, if that's sending a packet out to the data center and getting it back and doing it, um, if it's something that you can just pop over and plop it in yourself and you're done, like that wasn't something you used to do and you're able to do.
00:27:24.610 --> 00:27:41.810
Maybe you want to think more about using AI for the things that you can't do easily and offsetting that use by doing things yourself where it makes sense, where it would just be like, why would I ask AI to do that?
00:27:41.810 --> 00:27:42.930
I can just do it.
00:27:42.930 --> 00:27:44.770
So, another way to think about it.
00:27:44.770 --> 00:27:53.410
So, trying to keep in mind those pros, the things that you're using, the benefits of your using of AI.
00:27:53.410 --> 00:27:53.970
You know,
00:27:53.970 --> 00:28:03.410
Maybe you are using AI to um handle part of three other people's jobs that you're doing at your nonprofit because you're chronically understaffed.
00:28:03.410 --> 00:28:07.410
That's a valid way to use AI and reach more people.
00:28:07.410 --> 00:28:18.370
Like your mission is to serve more people, and you can do that because AI is helping you with some of these day-to-day tasks or productivity tasks, then build that into your calculations.
00:28:18.370 --> 00:28:20.130
You have a benefit.
00:28:20.530 --> 00:28:29.650
You will probably still have people who say, I'm just against AI, I don't want to learn about it, I don't want to use it, it's terrible for the planet, and I'm just very upset about it.
00:28:29.650 --> 00:28:36.530
Um, and those are the people who really care about these kinds of questions, like that have a lot of nuance to them.
00:28:36.530 --> 00:28:43.410
It's not irrational to be worried about and afraid of and in opposition to AI.
00:28:43.410 --> 00:28:52.050
Um, the original Luddites weren't against technology, they were against their technology being stolen from them and used against them.
00:28:52.450 --> 00:29:07.730
So in 2026, I think that's a good thought to leave us with about how it's implemented, who benefits, and you know, can we use this to make the world better?
00:29:07.730 --> 00:29:20.290
And how involved do we need to be in real life to ensure that those policies and regulations and human-focused AI become reality?
00:29:20.290 --> 00:29:21.570
So
00:29:21.570 --> 00:29:26.130
I want to remind you we do have a template for an AI acceptable use policy.
00:29:26.130 --> 00:29:28.450
It's at our site, communityit.com.
00:29:28.450 --> 00:29:30.370
I'll include that in the show notes.
00:29:30.370 --> 00:29:32.930
Again, it's you can't just download it and use it.
00:29:32.930 --> 00:29:38.290
You do have to customize it and make it individual to your organization.
00:29:38.450 --> 00:29:42.850
And some of these questions you're gonna have to think about just even individually to yourself.
00:29:42.850 --> 00:29:47.330
What do you feel comfortable with using AI for and why?
00:29:47.330 --> 00:29:49.570
And share that with your colleagues.
00:29:49.570 --> 00:29:50.450
Um,
00:29:50.450 --> 00:29:54.850
You can find us on Reddit at r slash nonprofit IT Management.
00:29:54.850 --> 00:29:57.410
We have a webinar this week, tomorrow.
00:29:57.410 --> 00:30:01.730
It's on an AI maturity model for nonprofits.
00:30:01.730 --> 00:30:09.090
Um, so if you are at a nonprofit that's been using AI for a while, a lot of the material out there is for beginners.
00:30:09.090 --> 00:30:17.250
So this is a webinar that's gonna talk about kind of next steps, how to really deeply integrate AI into what your nonprofit does.
00:30:17.250 --> 00:30:24.210
We have some experts who are gonna be with us talking about some case studies and real life experience that they've had.
00:30:24.210 --> 00:30:26.690
If you're interested in that, please join us.
00:30:26.690 --> 00:30:27.890
You can still register.
00:30:27.890 --> 00:30:30.370
It'll be tomorrow on Wednesday.
00:30:30.610 --> 00:30:39.170
And then I'll be back here on Friday with some more nonprofit uh podcasts and next Tuesday with some more nonprofit AI.
00:30:39.170 --> 00:30:40.930
So until then, take care.
