Is Your Sustainability Presentation Actually Hurting the Environment?
As travel and tourism companies put sustainability under the microscope, there's one environmental impact that deserves more attention: the technology we're using to communicate about it.
Sustainability has been a hot topic across industries for years. In travel and tourism in particular, it's practically impossible to attend an industry conference, read a trends report, or sit through a leadership presentation without hearing about it.
We’re discussing carbon emissions, over-tourism, waste reduction, sustainable aviation fuel, responsible tourism, and everything in between.
We're also getting more critical about the difference between companies making tangible, consistent progress and those simply getting better at talking about it. Skift, for example, has increasingly challenged the travel industry to examine the gap between sustainability promises and true action.
And having spent time in plenty of rooms discussing sustainability, I've started wondering about something else.
Are we thinking about the environmental impact of how we're communicating all this sustainability messaging?
Because I've sat through numerous presentations listening to the speaker wax righteous around reducing environmental footprints while pointing to a powerpoint that included some pretty obviously AI-generated elements.
Which always makes me want to ask: How much generative AI went into making your sustainability presentation and (more importantly) how much of that resource consumption is anyone accounting for?
AI doesn't exist in a resource-free universe
I’ll start by acknowledging I work in marketing and technology and obviously I use AI. I experiment with new tools, build things with them, and see enormous potential in what they can help businesses accomplish.
I've also used AI in conducting research for this specific article, which is not lost on me.
But I think we've gotten comfortable treating digital activity as though it happens in some magical, resource-free universe (it doesn't).
Every time we use a generative AI tool, computers somewhere are doing the work. Those computers require electricity. They produce heat and need cooling, sometimes using substantial amounts of water. Building the data centers and manufacturing the equipment requires resources as well.
According to the International Energy Agency's 2025 Energy and AI report, data centers accounted for approximately 1.5% of global electricity consumption in 2024. That demand is projected to more than double by 2030, with AI a major contributor.
That figure includes all data-center activity, not just AI. But it helps illustrate the scale of the infrastructure involved.
And the environmental implications go beyond electricity consumption.
When data centers draw electricity generated from fossil fuels, that contributes to greenhouse gas emissions and climate change. Water used for cooling can put additional pressure on local supplies, particularly in regions already facing shortages. Manufacturing the specialized chips and equipment requires mining, materials, and energy, creating additional environmental impacts.
So when we talk about AI's environmental footprint, we're talking about energy consumption, carbon emissions, water use, and the physical resources required to support the technology.
Some AI tasks use a lot more energy than others
Think about the different ways you might use AI in a typical workday.
You could ask it to organize information, write a paragraph, build a website, create an image, or generate a video.
Those might all feel like similarly simple requests from your side of the screen. Behind the scenes, they can be very different jobs with very different energy consumption implications.
A 2024 peer-reviewed study by Sasha Luccioni, Yacine Jernite, and Emma Strubell examined the energy consumption of 88 AI models performing different tasks. The researchers found substantial differences in energy use, with image generation particularly demanding.
Here's a simplified look at how common AI activities compare:
| What You're Using AI For | Everyday Example | Energy Use* |
|---|---|---|
| Sorting or analyzing information | Categorizing customer reviews or organizing spreadsheet data | Low |
| Writing or summarizing text | Drafting an email or summarizing a report | Low to moderate |
| Building websites, apps, or dashboards | Using Claude Code or ChatGPT to build a website or dashboard | Moderate to high |
| Generating images | Creating a destination image for a presentation | High |
| Generating video | Creating a 30-second AI-generated promotional video | Very high |
*This is a simplified, directional comparison. Actual energy use varies depending on the AI model, complexity of the task, number of revisions, and computing infrastructure.
To put the differences into perspective, the 2024 study found that generating 1,000 images used an average of approximately 2.9 kilowatt-hours of electricity across the models tested. By comparison, 1,000 simple text-classification tasks used approximately 0.002 kilowatt-hours.
That's a dramatic difference. And it's worth thinking about when we're generating 20 different images for a presentation, experimenting with dozens of versions of a video, or producing content that ultimately never gets used.
Each attempt may seem effortless to us while the computers doing the work are consuming very real resources.
What about the environmental footprint of marketing itself?
This is especially relevant to my own industry. Digital marketing had an environmental footprint long before generative AI became mainstream.
Websites, digital advertising, video streaming, cloud services, and the technology supporting them all require energy and infrastructure.
Now we're adding generative AI to that ecosystem.
Think about how much content an organization produces, whether via social posts, ads, web/landing page copy and imagery, email campaigns, videos, presentations, sales materials, and more. And also think about the agencies, freelancers, production companies, and technology vendors involved.
How much generative AI is being used across all that work?
How much is generated, revised, discarded, and regenerated?
And if an organization is measuring its environmental footprint, is that content production being accounted for within its measurements?
I'm not suggesting AI-generated marketing assets are necessarily among the largest contributors to a travel company's environmental footprint. Compared with aviation, hotel operations, or transporting thousands of people to an international conference, they may be relatively small.
But shouldn't we at least acknowledge that they have impact more than we are, and start roadmapping how to measure them?
Especially when some of these same organizations are asking us to scrutinize the environmental impact of everything else.
Technology can be part of the solution too
The same technology that requires energy and resources can also help us use them more efficiently.
AI can help optimize electricity systems, improve transportation planning, forecast demand, reduce waste, and identify inefficiencies.
In travel, that could mean more efficient operations, better resource allocation, or less unnecessary consumption.
Sometimes using AI might even eliminate the need for something with a much larger environmental footprint.
The International Energy Agency has explored these opportunities alongside the technology's growing energy demands.
So this is clearly a nuanced topic. We have tools with enormous potential to help solve problems, including environmental ones, while simultaneously creating new resource demands.
How do we decide when the tradeoff is worth it?
Maybe sustainability needs a limitations section
I come from an academic background, and one of the things I remember vividly about writing research papers was the limitations section.
You could spend an entire paper building your argument, explaining your findings, and defending your conclusions.
Then you had to turn around and explain everything that might be wrong with them.
Here's what we didn't measure / here's what we don't know /here's where our assumptions might not hold / someone else published the exact opposite of what we’re saying. (etc, etc)
You didn't have to abandon your argument because it had limitations, but you couldn’t publish without at least acknowledging them.
And I wonder whether our sustainability conversations could use a little more of that thinking.
Organizing a conference or presenting about sustainability is an important step, as is investing in responsible tourism practices. But if you're using generative AI to make your message more compelling or your operations more efficient, that's worth factoring in and challenging yourself to consider.
What resources did those choices require? What do we know about their environmental impact? What don't we know? And are we comfortable with the tradeoffs?
Those questions make the sustainability conversation more credible.
So what does responsible AI use actually look like?
For many organizations, choosing not to use AI isn't realistic or necessarily desirable. The technology has enormous potential to improve how we work and help address some of the challenges we're discussing.
But responsible use should involve more than assuming that anything that saves time is automatically a net positive.
Maybe it starts with some fairly ordinary questions.
Do we need to generate this image/video, or could we use something we already have?
Are we generating endless variations because they improve the work, or just because we can?
Do we know anything about the environmental practices of the AI providers and creative partners we're hiring?
And if we're making sustainability claims, are we willing to acknowledge the resource costs associated with the technology we're using to communicate them?
I don't expect organizations to calculate the emissions associated with every image or paragraph it generates (plus we don't necessarily have the tools or transparency to do that accurately yet).
What we CAN do is start paying attention, asking better questions, and recognizing that digital consumption is still consumption.
Because when we're sitting in a room discussing environmental responsibility while looking at a presentation full of AI-generated imagery, I think it's fair to ask:
Are we applying the same scrutiny to how we're communicating about sustainability as we are to the things we're asking everyone else to change?
A world without contradictions is unrealistic, so I’m not expecting this to have a tidy solution. But I do hope to see more acknowledgment that those contradictions exist.
And maybe that's a reasonable place to start.
