Thematic Cautions Businesses on Products Overhyped by 'AI Washing'

AUSTIN, Texas, October 16, 2024 (Newswire.com) - In trade shows, advertisements, and even media reports, countless companies claim that their product features “AI.” Thematic, an analytics company founded and run by a renowned artificial intelligence researcher, cautions that many or most of these products are cases of “AI washing” -- overselling or overstating a product’s AI capabilities. 

“If you walk around U.S. leading trade shows in 2024, you won’t find a single booth that doesn't feature AI,” said Dr. Alyona Medelyan, CEO of  Thematic, creators of a platform that analyzes free-form customer feedback data and distills actionable insights. Medelyan’s research has been cited in thousands of academic papers.  

“The truth is, the majority of the vendors are using AI washing," she added. “Anything related to data science, robotics, machine learning, or in fact, any algorithm or automation is now labeled ‘AI.’ Just like ‘big data,’ it has become the new catch-all term. “   

AI washing is a problem because, over time, it erodes trust in true artificial intelligence, Medelyan said. “As a result, companies and users take longer to adopt truly innovative technology and they don’t reap the benefits of AI as fast as they could have,” she said. 

AI washing is not a new phenomenon. The first AI-powered expert systems emerged in the 1970s and AI capabilities improved, with initial applications limited to niche areas like speech recognition and recommendation systems. AI gained significant traction in the mid-2010s with the advent of deep learning.  

“Each time, there was the overselling and the overpromising,” Medelyan said. “As a result, the industry suffered through two ‘AI Winters,’ first in the ‘70s, and later in the late ’80s to early ’90s, when skepticism grew and funding dried up. So, what we are witnessing now is a never-ending cycle: While tech advancements are real, the tendency to inflate capabilities for commercial gain is a persistent issue. “  

Medelyan cited the example of IBM’s “Watson for Oncology.” It was promoted as a breakthrough AI for cancer treatment, capable of analyzing medical papers and patient data to provide treatment recommendations. However, its real-world performance was disappointing. An internal IBM document revealed that Watson based some of its suggestions on hypothetical cancer patients rather than real patient data. Watson’s failures eroded medical professionals’ trust in AI.  

Because "AI" has developed different meanings, it has become a marketing buzzword, but that has attracted the attention of regulators, Medelyan said.   

There are now new laws in the making to protect consumers from deceptive AI marketing. The Federal Trade Commission has reminded businesses that the false advertising rules that apply to traditional goods and services also apply to AI, and the agency sued Automators AI, which sold “AI-powered” online storefronts. The American Bar Association recommends thinking twice before advertising a product or service as AI-enabled.   

Still, Medelyan predicts there will be little change in AI washing by B2B vendors. AI solutions are often highly specialized, making it difficult to establish universal standards or assess whether a product is truly AI-driven, she said. In addition, every industry uses AI differently, which adds to the difficulty in crafting effective, sweeping regulations. “AI technology evolves so quickly that regulations can't keep up, so the rules are often outdated by the time they are enforced,” she said. 

Medelyan offers several recommendations for organizations that seek to purchase AI technology but are wary of buying overhyped products that do not deliver the promised advantages:  

Develop Awareness. Buyers need to be aware of AI washing. Just as most of us understand that snacks shaped and colored like vegetables aren’t actually healthy, buyers must be critical of AI claims. Not everything labeled as “AI” delivers genuine AI capabilities.   

Build basic understanding. It’s important to understand the difference between algorithms, Machine Learning, and the AI of 2020s. In fact, ask ChatGPT to explain it in simple terms. It can keep explaining and clarifying without judgment.  

Seek transparency. Buyers should educate themselves on the common signs of AI washing. Start by demanding transparency from vendors. Ask how their algorithms work and what data they are trained on. If a company cannot explain its AI in clear, understandable terms, it's a red flag.   

Lean into your industry. Buyers should also look out for industry standards created by independent entities. For example, ESOMAR, a non-profit market research organization, published 20 questions to help buyers of AI-based insights and research products.  

Try before you buy. If it’s AI, it should be relatively easy to try it out on your use case. For example, at Thematic, where we analyze customer feedback using AI, we always offer a free demo on a sample of feedback. Buyers should also get a good understanding of how the AI-enabled solution can work in production, not just a demo. Understand the difference between ChatGPT and similar DIY AI tools. If it’s a wrapper, the cost needs to be proportionate to the value of what the wrapper delivers.  

“While AI washing presents challenges and necessitates vigilance, let’s not forget the transformative power of authentic AI solutions,” Medelyan said. Even in the skeptical world of healthcare, we now have successful examples of AI solutions diagnosing eye diseases or identifying abnormal chest X-rays. Similarly, AI tools deliver countless benefits in data analysis, call center optimization, sales and marketing processes.   

Medelyan notes that apart from obvious benefits such as saving time, cost reduction and improved productivity, AI provides three intangible advantages that increase growth:  

• By providing easy access to insights, people can make more data-driven decisions which results in better outcomes.  
• By providing instant transcriptions and summaries of customer and prospect conversations, teams that aren’t usually customer-facing can get a better understanding of customer needs.   
• By automating mundane tasks using AI, employees can focus on high-value tasks, which makes their work more enjoyable.  

 “As we look forward, the key to unlocking AI's full potential lies in recognizing and supporting authentic AI solutions that deliver real, measurable benefits,” Medelyan said. “By focusing on transparency, substantiated claims, and genuine technological advancements, we can ensure that AI continues to be a force for good, driving progress and innovation across industries.” 

For more information about Thematic and its AI-driven insight solutions, please visit https://getthematic.com/.

About Thematic:  

Thematic is a pioneering leader in AI-powered insight solutions, making it easy for organizations to transform customer feedback and verbatims into actionable, data-driven insights. Prioritizing transparency, high data quality, and rigorous validation processes, Thematic delivers unmatched accuracy and speed in text analytics for companies ready to elevate their customer experience (CX) programs. 

Source: Thematic