AI is becoming increasingly integrated into everyday life.
There’s been a notable amount of fake, AI-generated reviews across industries such as healthcare clinics, airlines, and even holiday shopping.
Further, there’s been a substantial increase in AI content in Google Search Results.
As AI continues to impact numerous fields, we conducted a study to determine whether AI was present in health and wellness product reviews across key consumer industries — baby formula, skincare, and health supplements.
To conduct the study we analyzed a dataset of 11,263 product reviews.
Our research aims to assess the impact of AI reviews on consumer trust, and the ethical concerns surrounding their use. These are our findings.
Looking to find out whether that review you’re reading is human-written or AI-generated? Try the Originality.ai AI Checker.
After conducting our analysis, we were pleasantly surprised.
Previous findings in our healthcare clinic review study demonstrated a substantial amount of AI reviews in healthcare clinics. 28.9% of U.S. plastic surgery clinic reviews and 20.7% of Canadian dental clinic reviews were detected as Likely AI-generated.
However, in contrast, there was a low level of AI in health and wellness product reviews.
Only 4.4% (494 reviews) of health and wellness products were Likely AI-generated.
This means that the vast majority of health and wellness product reviews are still Likely human-written at 95.6% (10,769 reviews) — at least for now.
The rate of AI healthcare reviews at 4.4% is so low that it falls within the typically normal range of false positives with AI detection.
False positives in AI detection occur when an AI detector identifies original, human-written content as AI-generated.
At Originality.ai, our Lite model offers an under 1% false positive rate. Then, our Turbo model has an under 3% false positive rate.
So, in this context, 4.4% of reviews being detected as Likely AI is just slightly higher than average false positive rates in AI detection.
Read more about our AI detection models.
Note: False positive rates can vary by AI detection model and company.
For our study, AI-generated reviews were segmented by product category:
The aim of dividing the reviews was to evaluate how prevalent AI-generated reviews are across key consumer products in the health and wellness industry.
While supplements had the highest amount of reviews that were Likely AI, skincare products exhibited a higher percentage of AI-generated reviews than baby formula.
This suggests that AI tools are more commonly used in the review process for supplement and skincare products.
The baby formula category had the lowest percentage of Likely AI reviews, which is potentially due to the sensitive nature of the product and a correlating preference for human-generated reviews.
The higher percentage of AI in supplement and skincare reviews may indicate a preference for AI content due to the abundance of similar product offerings where AI could be used to generate large volumes of reviews quickly.
Looking at the bigger picture, beyond AI reviews in sensitive consumer health and wellness goods, the presence of AI reviews is on average trending upwards across several industries.
Since the launch of GPT-2, GPT-3, and GPT-4, AI reviews have steadily increased.
The rising trendline of AI reviews is depicted in the graph below.
To provide better context as to the presence of AI reviews across industries, let’s take a comparative look at how health and wellness product reviews stack up with the other studies we’ve conducted at Originality.ai.
The notable percentages of AI in healthcare clinic reviews across hospitals, dental clinics, and plastic surgery clinics in both the U.S. and Canada raise a number of ethical concerns. Read the full study on AI healthcare reviews here.
We found that there is a notable seasonal increase in shopping reviews that were Likely AI over the holiday season. Read the full study on AI holiday shopping reviews here.
Some of the top airlines demonstrate high percentages of AI reviews, for comparative purposes, we are highlighting a few of the most popular airlines here. Read the full analysis of AI in airline reviews here.
Next, we investigated how AI-generated reviews impacted the average star ratings of products and their popularity (number of reviews).
We compared AI-generated reviews to human-written reviews in terms of their influence on product ratings.
AI-generated reviews often had slightly higher star ratings compared to human-written reviews, but the difference was marginal.
AI-generated reviews pose significant ethical concerns, especially in sensitive product categories like healthcare products.
The lack of transparency around AI-generated content can mislead consumers, erode trust, and potentially manipulate consumer purchasing decisions.
AI-generated reviews can potentially inflate product ratings, especially if used by companies to manipulate consumer perceptions, particularly for expensive and critical healthcare products.
This report has highlighted the growing role of AI-generated reviews in health and wellness consumer goods.
The analysis shows that while AI-generated reviews are prevalent, for the present, key health and wellness products are still primarily human-written across supplements, skin care, and baby formulas.
Although this study found that AI health and wellness product reviews fell within the range of false positives in AI detection, the presence of any AI-generated reviews for products like supplements or baby formula is a cause of concern both ethically and for establishing consumer trust in brands.
This study analyzed supplement, skincare, and baby formula reviews for AI-generated content. 11,263 product reviews were collected across these industries.
Reviews were classified based on the AI_Likelihood score, where reviews with a likelihood above 50% were categorized as AI-generated, and those below were considered human-generated.
Dataset available upon request.
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.
