Should you use AI for travel?
Its pitfalls and perils
Across the travel world we're seeing more and more AI-generated content being pushed out into the internet. And this has made us stop to think.
A little power can be a dangerous thing. And AI is certainly incredibly powerful. It can create content - images and videos - that look incredible. Better than any human being could create on their own. It can carry out research and produce information sets that look - on the surface - to be brilliant.
But, when it comes to the world of travel, this doesn't always work out.
It's disastrous for the environment
This is actually our biggest gripe with it all, to be honest. People don't tend to know just how destructive using gen-AI is for the environment, mostly because it's not really something that AI companies want to shout about.
Using generative AI harms the environment in more ways than one:
- Environmental destruction: Building AI computer chips requires mining critical minerals like lithium, silicon and gallium. This mining creates toxic chemical runoff, which in turn poisons water sources and destroys habitats for wildlife.
- They’re energy hungry: Data centres for AI servers are predicted to consume 945 terawatt-hours of electricity every single year. For context, this is triple the annual electricity use of Pakistan, Bangladesh and Nigeria - combined.
- Water usage: The servers that AI runs on generate huge amounts of heat, particularly when they're all stacked together in data centre buildings. AI infrastructure withdraws billions of cubic meters of water - which places particular strain on drought-prone areas
- Waste product: Discard AI hardware is project to create 2.5 million tonnes of waste per year by 2030 - the bulk of which is expected to end up in low-income countries
When people are using AI multiple times a day for things like writing emails, it's coming with a huge unseen environmental toll. Using AI to generate images and videos in particular, drives this up even further. Multiply this by the thousands of people doing it on a daily basis and the result is severe damage to our planet for relatively trivial outputs.
It creates ethical issues
The AI-generated images you'll see will be made to look as perfect as possible. Usually to make you buy something or to convince you of something (usually to eventually make you buy something!)
If someone's gone to the trouble of making a fake image, their motives should probably be questioned. After all, would you want to buy something on the basis of an entirely fabricated image/video/text?
The real challenge here is the expectations it creates for travellers. If people think that it's even remotely possible to have a tiger cub perched on your shoulder like a pirate's parrot (see below!), they're in for nothing more than crushing disappointment.
These inflated expectations also have negative impacts on local hosts, too. We've heard increasing reports of people (not VetSetters clients of course - you're all much too reasonable) getting angry and aggressive with staff at safari camps when nature fails to deliver the same ‘experiences’ they've seen online.
Similarly, more and more people are engaging in intrusive - and at times downright unsafe - behaviours around wildlife, in an attempt to replicate cool-looking photos that have actually been generated by a machine.
All of this creates downstream negative impacts for the livelihoods of humans too, particularly those on the ground in lower-income countries and creative professionals; whose craft that's been developed over years is being replaced by something that can be created in seconds.

Needless to say, this is AI-generated
It makes stuff up - a lot
This is likely what everyone is most familiar with - gen-AI's propensity to simply invent nonsense, and to “say” it with the confidence of a vet client who's taken advice from their dog breeder.
The hallucination + confidence combination is particularly tricky, because you need to be in-the-know yourself to be able to catch when it's bluffing you. When it tells you that (yes is a real example!), it's obvious.
But how about when it tells you that the theme park Tivoli Gardens is in the town of Tivoli, just outside Rome. Unless you know already that Tivoli Gardens is actually 2,000 kilometres away in Copenhagen, you're likely to accept this hallucination as real. You might even book tickets to Tivoli Gardens, only to discover your error when your satnav informs you of your 23-hour journey north.
AI can't test the credibility of what it tells you. It's a sampling machine, so this issue is inherent to the way it works. It ‘predicts’ which word should follow each word, based on the dataset it has been fed. So it's pure probability that it's going to produce words that don't make real, contextual sense but that seem fine to the AI model.
My favourite analogy here is of placing a rhino in your living room. When you return later to find your house trashed, you wouldn't suggest that the rhino “just needed a better prompt”. You'd know that rhinos are inherently quite unpredictable and so letting it loose in your living room isn't something you should do.
By humans, for humans
We don't actually use any AI across VetSetters at all. Full stop. We feel it's a bit intellectually bankrupt, and even hypocritical, to extol the virtues of personal service using words and images created entirely by machines.
Does this mean everything takes 10x longer than it could? Oh hell yeah. But it's consistent with what we stand for. We also strongly believe that life isn't about doing everything as quickly as possible! If a thing is worth doing - it's worth actually doing.
Recent Posts








