AI Washing: A Rising Concern in the Expanding Artificial Intelligence Landscape

As artificial intelligence (AI) continues to dominate headlines, a growing issue has emerged: AI washing. This term, as explained by Securities and Exchange Commission (SEC) chair Gary Gensler, refers to false claims by companies about their use of AI technologies, misleading investors and stakeholders.

In March, the SEC fined two investment advisers, Delphia (USA) Inc. and Global Predictions Inc., $400,000 collectively for falsely marketing AI-enabled investment predictions. While both companies settled without admitting or denying the SEC’s findings, this case underscores a broader trend of companies overstating their AI capabilities.

This phenomenon extends beyond the financial sector. A report by analytics firm FactSet revealed that 179 S&P 500 companies mentioned AI in their earnings calls over three months, significantly above the five-year average of 73. The increase in AI references has raised concerns about the authenticity of these claims.

Michael Stewart, managing partner at Microsoft’s venture arm M12, highlighted the ease of integrating AI terminology into business pitches without substantive implementation. “There’s no sustainable competitive advantage to that,” Stewart said, warning that AI washing undermines trust between vendors, consumers, partners, and investors.

Timothy Bates, a professor of practice at the University of Michigan-Flint College of Innovation & Technology, emphasized the risks to stakeholders who may not fully understand the technology. He pointed out that applications relying on repetitive inputs without true AI learning capabilities are also examples of AI washing. For instance, a law firm’s AI assistant may falter if not based on a robust, specialized database.

Toby Coulthard, chief product officer at Phrasee, advises caution when companies broadly use the term AI. He suggests verifying the specifics of the AI technology and examining the company’s history with AI, especially if they discussed AI before the advent of ChatGPT. Companies should also clearly outline their AI ethics policies and limitations.

Bates recommends scrutinizing the AI model a company uses, whether it’s proprietary or dependent on third-party models, and evaluating their service-level agreements and key performance indicators.

At M12, Stewart and his team use a framework called the four D’s—data, dividends, distribution, and delight—to assess AI startups. They prioritize access to critical customer data, evaluate whether AI outputs contribute to the bottom line, and consider the startup’s profitability and distribution channels. Creating delightful user experiences is also crucial for customer retention.

With a plethora of AI startups in the market, distinguishing genuine innovation from AI washing is essential for investors and consumers alike. M12 emphasizes the need for strong distribution channels and exceptional user experiences to ensure a startup’s sustainable success.

As AI technology continues to evolve, stakeholders must remain vigilant against AI washing to maintain integrity and trust in the rapidly growing AI industry.