When travelers ask ChatGPT or Google AI where to go in the Mediterranean, they are not getting an objective cross-section of real-world demand. Instead, they receive a curated travel itinerary shaped by digital editorial bias.
A study from Tharro (The Mediterranean Pulse) put 520 holiday scenarios—spanning five key source markets—to the two leading AI platforms. The results show an algorithm that rewards a few favorites while pushing major destinations down the list. In the 520 AI-generated travel answers, Greece, Italy, and Spain dominate the Mediterranean destination map, appearing in 94%, 88%, and 87% of answers, respectively.
After those three, visibility drops sharply. Croatia appears in 59% of answers, followed by Portugal at 43%, France at 37%, Turkey at 34%, Cyprus at 31%, and Malta at 30%.
But appearing in an AI answer is not the same as being recommended.
Italy, Greece, and Spain account for 77% of the destinations named first in the answers studied. Italy leads with 32%, followed by Greece at 24% and Spain at 21%.
For smaller destinations, the problem is often not being completely left out. It is being mentioned too late.
Typecasting the Challengers
For destinations outside the top tier, AI rarely presents a full view. Instead, it files them under repetitive categories:
- Turkey (2.7× budget/all-inclusive): Recommended mostly for cheap package holidays, while nearly invisible for luxury, culture, or food-and-wine prompts.
- Cyprus (3.2× winter sun / 2.1× all-inclusive): Strongly cited for winter warmth, families, and package deals, but almost absent from luxury or solo travel queries.
- France (2.5× luxury / 2.0× weddings): The only challenger destination that AI consistently places upmarket.
- Malta (2.6× culture): Recommended mainly as a historic or winter getaway, missing growing upmarket demand.
| Destination | Share of AI Recommendations | Share of Search Demand | AI vs Demand |
|---|---|---|---|
| Croatia | 6.3% | 11.8% | 1.9× |
| Italy | 10.7% | 17.3% | 1.6× |
| France | 5.3% | 6.8% | 1.3× |
| Portugal | 7.6% | 9.2% | 1.2× |
| Cyprus | 5.7% | 6.1% | 1.1× |
| Greece | 17.9% | 18.6% | 1.0× |
| Malta | 5.9% | 6.0% | 1.0× |
| Spain | 28.1% | 17.4% | 0.6× |
| Turkey | 12.5% | 6.8% | 0.5× |
The researchers describe the pattern as AI compressing the largest destinations while inflating the smallest ones.
Engine Disparities: ChatGPT vs. Google AI Mode
The choice of AI tool dramatically alters destination visibility. ChatGPT generates longer responses, averaging 5.7 destinations per query, giving secondary destinations a chance to appear near the bottom.
Google’s AI Mode, however, keeps lists tight, averaging 4.3 destinations. This compression hits challenger destinations hardest:
- Cyprus: Drops from 38% presence in ChatGPT to 25% in Google AI Mode.
- Turkey: Drops from 49% down to 19%.
- Croatia: Drops from 79% down to 39%.
As travelers increasingly rely on generative search engines rather than traditional query boxes, hoteliers and tourism boards face a new reality: being online is no longer enough. Winning requires securing a top-three placement within the concise, highly curated lists that AI chooses to produce.
Five Markets See Five Different Mediterraneans
The AI map also changes depending on where the traveler comes from.
Cyprus appears in 44% of Polish AI answers, but only 25% of German answers and 24% of British answers.
That creates a particularly striking contradiction for Cyprus.
The UK is its largest search market, but one of its weakest AI markets.
In other words, British travelers are looking for Cyprus, while AI is less likely to put it in front of them.
The Polish market tells a different story. Cyprus is strongly associated with all-inclusive travel, which aligns with how Polish travelers search for holidays.
Israel also produces a distinctive AI map. Cyprus appears in 36% of Hebrew-language answers, while France receives unusually high visibility.
The lesson is that there is no single AI Mediterranean. The answer changes with the question, the language, and the market.
What This Means for Travelers
For travelers, the research offers a reason to ask AI more than one question.
If you ask, “Where should I go in the Mediterranean?” you are likely to get very different answers from asking:
- Where can I find an affordable Mediterranean holiday?
- Where should I go for local food and wine?
- Which Mediterranean island is best for a romantic trip?
- Where can I find winter sunshine?
- Where can I have a quieter beach holiday?
- Where should I go for local food and wine?
- Which Mediterranean island is best for a romantic trip?
- Where can I find winter sunshine?
- Where can I have a quieter beach holiday?
The destination that disappears from the first question may become the obvious answer to the second or third.
That is particularly relevant for Cyprus, Malta, Turkey and Croatia, whose AI identities are more specialized than those of Greece, Italy and Spain.
For destination marketers and hotels, the message is even clearer.
Being mentioned is no longer enough. Being mentioned early matters.
And increasingly, the competition may not be the destination that attracts the most traditional searches. It may be the destination that AI has decided belongs on the shortlist.
The Mediterranean has always had its rivalries. Now, the robots have opinions too.