Around the world in one hour, with a little help from AI

Today I travelled from London to New York, via Paris and Athens, in about an hour.

I did not leave my desk.

I spent some time playing with AI-generated images, using ChatGPT’s Create image function and a single reference photo of my own face. I placed “myself” into a handful of familiar city settings and let the system do the rest. I generated several images that might create a sense of travelling.

This was partly curiosity, partly play. But it is also very directly connected to my PhD work.

My research looks at how rural DMOs can use AI in tourism promotion in ways that are effective, ethical, and grounded in local values. Not to replace real places or experiences, but to understand what these tools actually do, how they feel to audiences, and where the boundaries should be.

Using my own face was a deliberate choice. It avoids copyright issues, ensures consent is clear, and keeps the experiment transparent. Every image you see in this post is AI-generated, and every face in those images is mine.

What I actually did:

I did not use any external tools or platforms. Everything was done inside ChatGPT using the Create image function.

I uploaded one reference photo of my face and then asked ChatGPT to help generate detailed image prompts for different locations. I did not carefully craft each prompt myself. Instead, I asked the system to generate prompts, selected the ones that made sense, and then created images based on those descriptions.

The result was a small set of cinematic, photorealistic portraits placing “me” into London, New York, Paris, Athens, and a few other imagined moments. None of them are real. All of them look plausible.

That plausibility is the point.

A simple walkthrough of the process

For those curious about how accessible this already is, here is the basic flow I followed.

I asked ChatGPT to:

"Create an ultra-realistic, photorealistic outdoor portrait using the man in the provided reference photo as the sole and exact facial reference.

His face must remain completely faithful to the reference photo — identical facial proportions, natural skin texture, realistic pores, fine lines, beard growth pattern, nose shape, eye shape and spacing, lips, cheek structure, and a natural, softer jawline exactly as in the reference. No chin dimple. No exaggeration or idealization. No glasses or sunglasses.

The scene takes place on a small dive boat at sea. The man is seated on the edge of the boat, preparing to go scuba diving.

He is wearing scuba gear partially on — wetsuit top or rolled-down suit, fins nearby, mask and regulator visible — clearly in preparation rather than already diving.

In the background, a dolphin is captured mid-jump above the water’s surface, creating a dynamic, joyful moment. The open sea and clear sky frame the scene.

The man is smiling naturally, looking down or slightly to the side as he adjusts his gear, conveying excitement and anticipation.

Lighting is bright natural sunlight with crisp reflections from the water illuminating the face naturally.

Camera style: mid-range lifestyle portrait, dynamic composition with motion in the background.

Color grading: vivid blues of sea and sky, clean highlights, warm natural skin tones.

Mood: joyful, adventurous, alive — anticipation and freedom at the edge of exploration"

I clicked the + icon next to the message field, chose Create image, attached the reference photo if needed, and asked ChatGPT to generate the image.

That was it. No post-processing, no external tools, no manual editing.

When I needed more, I used prompts like this:

"Scenario 1: London (Urban Iconic)

• Location: London, with Big Ben and the Houses of Parliament clearly visible.

• Environment: Red double-decker buses in motion, classic London street elements.

• Pose: Standing or walking near a crosswalk, body angled slightly, gaze off-camera.

• Mood: Confident, cosmopolitan, composed.

• Lighting: Soft, overcast London daylight.

• Style: Refined urban travel portrait.

Scenario 2: South Africa (Safari Adventure)

• Location: African savannah in South Africa, on an open safari jeep.

• Environment: Elephants and lions visible in the background at a safe distance.

• Pose: Seated sideways on the jeep, one arm resting on the railing, looking toward wildlife.

• Mood: Adventurous, grounded, quietly awe-inspired.

• Lighting: Warm, golden late-afternoon sunlight.

• Style: Authentic safari lifestyle portrait.

Each prompt should be highly specific, cinematic, and photorealistic, maintaining strict facial accuracy while varying setting, posture, and emotional tone."

This ease is both impressive and slightly unsettling.

Why this matters for destinations

For DMOs, especially rural ones with limited resources, tools like this are attractive. They promise speed, flexibility, and lower production costs. They also introduce new risks.

Photorealistic AI images can blur the line between storytelling and simulation. They raise questions about authenticity, trust, and expectation management. If an image looks real, audiences may assume it represents something that exists, or something they will experience if they visit.

This is where my PhD work, and the AI Readiness and Ethical Targeting Toolkit (AIRETT), comes in. I am using small experiments like this to benchmark what AI can already do, how it feels from the viewer’s side, and what kinds of guardrails DMOs need before adopting similar approaches.

The question is not whether AI can help tourism promotion. It already can. The real question is how it should be used, by whom, and under what principles.

A few early reflections

Even in this playful test, a few things stand out.

AI-generated images are very persuasive very quickly. They require far less effort than traditional production. That makes ethical reflection more important, not less. And what about the work of local photographers and their livelihood?

Disclosure matters. If an image is AI-generated, audiences should know. Transparency builds trust, especially for destinations that rely on long-term relationships with visitors and residents.

Context matters. A playful portrait experiment is different from representing a place, a community, or a cultural experience. Rural destinations in particular need to be careful about how they are portrayed and by whom.

And finally, experimentation matters. DMOs cannot make good decisions about AI if they do not understand the tools firsthand. Avoidance is not a strategy. Blind adoption is not either.