The rise of AI-generated content has sparked concerns over the authenticity of online reviews, especially in industries where customer feedback is critical to consumer decision-making.
A recent study from Originality.ai revealed that AI-generated reviews on Amazon have skyrocketed by 400% since the launch of Chat-GPT. In fact, extreme reviews, which are categorized as five-star or one-star reviews, were more likely to be AI-generated.
Another industry that is highly dependent on customer reviews is the travel industry. Finding the right airline for your needs is absolutely critical, whether it’s because you have specific in-flight requirements or unique requests when you get to the airport. Flights are also high-ticket items, so it’s important for consumers to know they’re booking with a trustworthy company.
One way to be sure that your needs will be met is to head to review sites like Tripadvisor or AirlineQuality to see what other customers think of a particular airline before you fly with them. But can these airline reviews still be trusted?
At Originality.ai we are experts in AI detection with our models detecting AI-generated content at an accuracy rate of 99%. We used our proprietary tool to help consumers find out which airlines have the most AI-generated reviews and its impact on consumer trust and airline reputations.
Airline reviews play a key role in shaping consumer choices. However, with AI tools capable of generating realistic reviews, the line between authentic and artificially created feedback has blurred.
In just a matter of clicks, an experienced prompter can get a list of reviews generated, each with a unique (enough) tone and style to come across as a different “customer.”
John Gillham, CEO at Originality.ai, explains why humans struggle to spot AI content:
“Despite confidence in their own ability, humans tend to struggle to spot AI-generated content. A study revealed that experienced teachers only accurately identified 37.8% of AI texts. Which is why AI content detectors are so important. They’re significantly more effective than humans at identifying AI content.
“There are some significant implications of not being able to detect AI content, such as the spread of misinformation, academic dishonesty, and a lack of authenticity online.”
This has raised questions about the reliability of reviews on platforms like AirlineQuality, TripAdvisor, Google Reviews, and airline-specific portals. Airlines and consumers alike must now consider whether AI content is influencing perceptions of service quality and trustworthiness.
Our study analyzed the reviews of 10 airline carriers from around the world by looking at reviews from SkyTrax, a well-known platform where travelers leave reviews of their airline experiences. We then analyzed the reviews in our Originality.ai AI detector to find out if the reviews were AI-generated or human written.
Over the last 3 years, since the release of chatGPT, our analysis found that airline reviews that used AI-generated content increased by 189%.
Starting from 2017, the dataset shows that AI-generated reviews were relatively rare but began to increase steadily across multiple airlines.
The introduction of large language models like GPT-2 in 2019 and GPT-3 in 2020 likely contributed to a rise in AI-generated content as more advanced tools for generating text became widely available.
You can see this coming to fruition in the increased AI-generated review rates starting around 2020, with a noticeable uptick in reviews flagged as AI-generated for several airlines during this period.
The release of ChatGPT in late 2022 and ChatGPT-4 in 2023 coincides with even steeper growth in AI-generated review rates, suggesting that the advancements in AI capabilities had a direct impact on user-generated content or automated systems creating airline reviews.
This makes sense, given similar increases in AI content at this time, so it shouldn’t come as a surprise that the general trend filtered its way into the airline industry as well.
Based on the visualization above, several insights can be drawn about the evolution of AI-generated content in airline reviews across different airlines over the years.
In our study we analyzed specific airlines from all over the world to find out which ones have the most fake reviews and are potentially misleading customers.
In our study, ChinaSouthern has the most fake reviews overall, more than any other airline, with a rate of 32.4% AI-generated reviews. This shows that consumers who are trusting the reviews and potentially looking to buy a ticket from the airline are not informed enough to make a trusted decision. ChinaSouthern has a 3.5 star average rating on Tripadvisor.
From 2021 onwards, we observed notable increases in the AI-generated review rates for several airlines.
Interestingly, RyanAir and Emirates both saw a drop in AI-generated content in their reviews, which could suggest that both companies have decided to tackle the issue.
RyanAir had shown a moderate increase from 2017-2023 dropping from 11% in 2023 to 6.4% in 2024. This means that consumers can trust more recent reviews of the carrier. This could be down to the UK enacting policies on combating AI-generated fake reviews online.
Similarly, Emirates, had shown a rise from 0% in 2021 to 3.2% in 2024, peaking at 7.9% in 2023 before dropping back down to 3.2%, suggesting efforts to address fake reviews.
In 2024, SouthWest Airlines recorded an AI-generated review rate of 8.7%, marking it as the U.S. airline with the highest percentage of reviews generated by artificial intelligence.
Meanwhile, United Airlines saw the most significant surge in fake reviews over the past year, with an increase of 157%. This sharp uptick underscores potential challenges for United in managing its online reputation and ensuring that reviews reflect genuine customer experiences.
The significant presence of AI-generated reviews across airlines may prompt the industry to adopt more stringent verification measures, as travelers increasingly rely on online reviews when making booking decisions.
Next, let’s take a look at the impact that AI-generated reviews may have had on customer sentiment in the airline industry.
The sentiment graph above shows that across our major airlines, customer sentiment has seen a major decline by 90% from 2014-2024.
Below, we provide a visual representation of the average customer sentiment over time, allowing for an analysis of trends and fluctuations in customer feedback.
The blue data points represent individual days' median sentiment scores, with each point indicating the collective sentiment of reviews made on that day.
The green dashed line represents the Lowess smoothed line, which smooths out short-term variations and highlights longer-term trends. By comparing the individual data points with the smoothed line, you can identify both immediate shifts in sentiment as well as general patterns that might emerge over time.
When interpreting the sentiment graphs, a positive sentiment score (above zero) suggests generally favorable reviews, while a negative sentiment score (below zero) indicates dissatisfaction or negative feedback.
Sharp drops in sentiment might correspond to specific events (such as poor service incidents or flight cancellations), while more gradual changes could reflect broader shifts in customer perceptions over time.
If the green smoothed line shows a steady upward or downward trend, this indicates a sustained improvement or decline in customer satisfaction.
Looking closer at Air Canada, it’s easy to see a correlation between the increase in AI-generated reviews (3.0% in 2021 to 10.7% in 2024) and a drop in customer sentiment. Of course, other factors may have also influenced this, most notably the knock-on effects of COVID, but it is obvious from the data that more AI-generated reviews do not align with more positive experiences.
This situation is also similar for China Southern Airlines, which had a high AI-generated review rate of 12.9% in 2021, which peaked at 32.4% in 2023 While average customer sentiment only began to drop off in 2023, the range of sentiment drastically increased, with customers seemingly either loving or hating their experience with the brand.
Notable, AI-generated review rates have dropped to 18.8% in 2024, which will likely begin to reflect an increase in customer sentiment in 2025 as a result.
The correlation between increased AI-generated reviews and declining customer sentiment is much clearer for Delta. Once one of the most popular airlines in America, Delta's AI-generated reviews rose from 2.8% in 2021 to 9.7% in 2024, a 244% growth in total. At the same time, customer sentiment continued to drop from the highs of 2015 to the current lows of 2024.
Interestingly, Emirates went from having no AI-generated reviews in 2021 to 7.9% in 2023. However, after seeing the impact (or lack thereof) they had on customer sentiment, reducing that figure to 3.2% by 2024.
It’s important to note Qatar Airways was the only airline to remain operational during Covid, which explains the bump in sentiment seen during those years.
Despite that, Qatar Airways increased AI-generated reviews from 4.0% in 2021 to 10.4% in 2024. At that same time, customer sentiment dropped from the highs of 0.30 to 0.15.
Although RyanAir has never been the most popular of airlines, it interesting to see that their popularity has waned in the past decade, with some significantly poor customer sentiment scores in 2022/23. This timeframe also coincided with their highest percentage of AI reviews (11%), suggesting RyanAir may have been using this method to improve their scores.
It’s a very similar story for SouthWest Airlines, which has seen a consistent increase of AI-generated content, rising from 3.2% in 2021 to 8.7% in 2024, representing a 172% increase.
United Airlines, the airline with the overall lowest customer sentiment from this study, also showed a significant rise in AI-generated content, increasing from 2.5% in 2021 to 8.2% in 2024, without any major impact on the customer sentiment scores.
Spotting AI-generated reviews can be a minefield, especially as LLMs are becoming much better at mimicking human content online. In fact, multiple studies have shown that even humans who think they can detect AI content find it difficult to do so.
Jon Gillham, CEO of Originality.ai has laid out his top tips for spotting AI-generated reviews so consumers can make better choices online:
As you can see, each airline that shows an increase in AI-generated reviews also shows a downward trend in customer sentiment, suggesting that the service customers get from each airline is dropping year over year.
Of course, AI may not be the only reason for this, as there are several other mitigating factors, such as increased customer expectations, cuts in funding, and other elements.
However, a lack of authenticity and transparency in customer reviews will also be contributing to this decline as consumers become less and less trusting in these airlines and their authenticity.
One of the biggest influencers in airline choice (other than price) is brand loyalty, and this approach to AI-generated reviews could ultimately lead to some major brands losing their current market share.
To be sure that everything you are reading online is legitimate and human-written, be sure to check out the Originality.ai AI Detector.
How We Conducted the Study:
We used Originality.ai's API to detect AI-generated airline reviews by analyzing their content. Reviews with at least 50 words were scanned for an AI likelihood score and classified as AI-generated or not. We focused on the top 10 airlines by brand value, and visualized how AI-generated reviews changed over time, marking major AI model releases.
Sentiment Analysis:
We collected reviews from airlinequality.com, cleaned the data, and analyzed customer sentiment over time using TextBlob. Lowess smoothing helped highlight trends in customer satisfaction, with visualizations showing daily sentiment scores and overall trends. These insights are useful for tracking airline performance and satisfaction.
For raw data contact maddie@originality.ai
Have you seen a thought leadership LinkedIn post and wondered if it was AI-generated or human-written? In this study, we looked at the impact of ChatGPT and generative AI tools on the volume of AI content that is being published on LinkedIn. These are our findings.
We believe that it is crucial for AI content detectors reported accuracy to be open, transparent, and accountable. The reality is, each person seeking AI-detection services deserves to know which detector is the most accurate for their specific use case.
