--- title: "No Product Reviews Can Be Trusted Now That AI Can’t Be Reliably Detected – 2024" canonical: https://romancescamsnow.com/ai-reviews-can-be-trusted-now-2024/ type: post author: "SCARS Institute Editorial Team" published: 2024-04-29 modified: 2024-04-29 categories: ["♦ HOW SCAMS WORK", "2024", "AI - Artificial Intelligence", "Fake Products / eCommerce Fraud", "Insights", "Technology Of Scamming"] summary: "" organization: "SCARS Institute - Encyclopedia of Scans - World's #1 Anti-Scam Website - Romance Scams NOW" organization_url: https://romancescamsnow.com/ llms: https://romancescamsnow.com/llms.txt --- # No Product Reviews Can Be Trusted Now That AI Can’t Be Reliably Detected – 2024 # No Product, Service, or Business Reviews can be Trusted Now that AI can't be Reliably Detected ## With Each Passing Year, Fundamental Trust is being Eroded by Generative-AI ### AI and Scams - How Scams Work - A SCARS Insight ## Article Abstract Undetectable AI-generated reviews pose a significant threat to the credibility of online consumer reviews and commentary, as revealed by a new study from Yale School of Management professor Balázs Kovács. Published in the Springer journal Marketing Letters, Kovács' research unveils the alarming reality that AI-generated business reviews can deceive both human readers and AI detectors alike. In two experiments involving over 300 participants, it was found that humans struggled to differentiate between real and AI-generated reviews, with accuracy levels barely surpassing chance. Even leading AI detectors failed to identify AI-generated content, raising concerns about the erosion of trust in online reviews. These findings have profound implications for businesses, consumers, and policymakers, emphasizing the urgent need for enhanced authentication mechanisms and regulatory measures to combat fraudulent practices. As consumer trust hangs in the balance, stakeholders must prioritize transparency and authenticity to safeguard the integrity of online consumer feedback. ## Undetectable AI-Generated Reviews Challenge Trust in Online Platforms: A Wake-Up Call for Consumers and Businesses For the last 20 years, online reviews have wielded significant influence over consumer decisions, but a new study from Yale School of Management professor Balázs Kovács uncovers a troubling reality: AI-generated restaurant reviews can seamlessly deceive both human readers and AI detectors, posing a serious threat to trust in online platforms. Within just 2 years we have gone from the widespread introduction of generative AI, to its inability to be detected by most humans in specific contexts. While this does not mean AI is taking over the planet, it does mean that criminal use now makes it impossible to validate truth when it comes to business and product reviews! > According to Yale Insights: "If a product or business has hundreds of reviews, customers can usually trust that the “wisdom of the crowd” has generated an accurate average rating. But a poorly rated item with few reviews is likely scored lower than it deserves to be, according to research by [Yale SOM’s Balázs Kovács and his co-authors](https://insights.som.yale.edu/insights/can-you-trust-bad-online-ratings)." ## The New Study Published in the prestigious Springer journal Marketing Letters on April 14, 2024 (see below,) Balázs Kovács' research exposes the potential ramifications of advanced language models like GPT-4 on the credibility of online reviews. ### The Study's [Abstract](https://link.springer.com/article/10.1007/s11002-024-09729-3): > Online reviews serve as a guide for consumer choice. With advancements in large language models (LLMs) and generative AI, the fast and inexpensive creation of human-like text may threaten the feedback function of online reviews if neither readers nor platforms can differentiate between human-written and AI-generated content. In two experiments, we found that humans cannot recognize AI-written reviews. Even with monetary incentives for accuracy, both Type I and Type II errors were common: human reviews were often mistaken for AI-generated reviews, and even more frequently, AI-generated reviews were mistaken for human reviews. This held true across various ratings, emotional tones, review lengths, and participants’ genders, education levels, and AI expertise. Younger participants were somewhat better at distinguishing between human and AI reviews. An additional study revealed that current AI detectors were also fooled by AI-generated reviews. We discuss the implications of our findings on trust erosion, manipulation, regulation, consumer behavior, AI detection, market structure, innovation, and review platforms. ## The Power of Online Reviews Online reviews have emerged over the last 2 decades as a cornerstone of consumer decision-making, offering valuable insights into the quality and reputation of products and services. However, the proliferation of AI technology introduces a new layer of complexity to this landscape. In his study, Balázs Kovács conducted two experiments involving 301 participants recruited through Prolific Academic, comprising a diverse group of native English speakers from various countries. In Study 1, participants struggled to distinguish between real Yelp.com reviews and AI-generated counterparts, with accuracy hovering around 50%. Study 2 revealed even more startling results, as participants classified AI-generated reviews as human-written 64% of the time, despite their fictional nature. ## AI Detectors Were Also Fooled Too Also, leading AI detectors designed to discern human-written from AI-generated text proved ineffective in detecting AI-generated reviews. Copyleaks, a prominent AI-text recognition tool, labeled all 102 reviews tested as human-generated, highlighting its inability to identify AI-generated content accurately. Even GPT-4, when tasked with assessing the likelihood of reviews being AI-generated, exhibited difficulty distinguishing between human-written and AI-generated reviews, underscoring the challenge posed by advanced language models. Interestingly, in SCARS own testing of Copyleaks, it fails more often to detect real human written language and flags it as AI-written when it is not - we found this happening 72% of the time. Other AI detectors were even worse. ## Implications for Review Platforms and Beyond The implications of these findings extend far beyond simply the online reviews, with the potential ramifications for businesses, consumers, and policymakers alike being that consumer trust in any reviews and commentary can vanish at any time. The ability of AI to generate convincing fake reviews or comments raises concerns about the integrity of online platforms and the vulnerability of consumers to escalating manipulation. Small businesses, in particular, may suffer disproportionately from the proliferation of fake reviews, as their reputations hinge on genuine feedback from their customers. What this study shows clearly is that review platforms need to reassess their authentication mechanisms and that policymakers need to consider stronger regulatory measures to promote transparency, honesty, and to combat fraudulent practices. > As Balázs Kovács notes, "The finding that large language models (LLMs) can cheaply and quickly generate online review texts indistinguishable from those written by humans has wide-ranging implications." What may be more disturbing is how AI can be weaponized to destroy products, services, and businesses in systematic assaults on their reviews. It may be that it is time for all of the Internet to review the usefulness and viability of platform concepts. Are review websites, after all, such a good thing? ## Summary In an era where trust is paramount, consumers must remain vigilant and critical of online content, recognizing the potential influence of AI-generated text on their decision-making processes. Ultimately, the obligation falls on both businesses and consumers to decide if this evolving landscape is worth their trust and how to use reviews with caution. By demanding transparency, accountability, and authenticity, stakeholders can work together to preserve trust in online reviews and uphold the integrity of consumer feedback. ## More About Generative-AI: - [Can You Trust Bad Online Ratings? | Yale Insights](https://insights.som.yale.edu/insights/can-you-trust-bad-online-ratings) - [The Turing test of online reviews: Can we tell the difference between human-written and GPT-4-written online reviews? | Marketing Letters (springer.com)](https://link.springer.com/article/10.1007/s11002-024-09729-3) - [AI-Generated Reviews Fool Humans and Detectors, Threatening Trust in Online Platforms (suchscience.net)](https://suchscience.net/ai-generated-reviews-fool-humans-and-detectors/) - [FraudGPT – AI For The Bad Guys (scamsnow.com)](https://scamsnow.com/fraudgpt-ai-for-the-bad-guys/) - [The Dark Side of Generative AI (scamsnow.com)](https://scamsnow.com/the-dark-side-of-generative-ai/) - [The Age Of AI (Artificial Intelligence) And New Fraud Patterns – 2024 (romancescamsnow.com)](https://romancescamsnow.com/dating-scams/artificial-intelligence-fraud-patterns-2024/) - [AI (Artificial Intelligence) and the Engineering of Consumer Trust (romancescamsnow.com)](https://romancescamsnow.com/dating-scams/ai-artificial-intelligence-and-the-engineering-of-consumer-trust/)