Amazon faced significant criticism this year when reports cast doubt on the efficiency and autonomy of its "Just Walk Out" technology used in its Amazon Fresh and Amazon Go stores. The AI-powered system is designed to let customers pick items and leave without checking out, using facial recognition technology along with numerous sensors and cameras to identify items and automatically bill customers who have registered via an app.

However, in April, reports emerged suggesting that the system wasn't as autonomous as claimed. Allegedly, around 1,000 workers in India were manually verifying nearly three-quarters of the transactions. Amazon quickly refuted these claims, stating that the reports were erroneous and that the Indian workers were not reviewing video footage from the stores. Instead, Amazon clarified that these workers were merely reviewing the system to ensure accuracy, which is common practice for any AI system that prioritizes precision. Despite this, Amazon announced plans to reduce the number of stores using the Just Walk Out system.

This situation with Amazon exemplifies a broader issue in the tech industry known as "AI washing." AI washing refers to the practice of companies making exaggerated claims about their use of AI technologies. Similar to "greenwashing" in the environmental sector, AI washing involves overstating the role and capabilities of AI in products and services.

AI, while not having a strict definition, generally refers to computer systems capable of learning and problem-solving after being trained on vast amounts of data. A notable subset of AI that has garnered attention in recent years is generative AI, which can create new content such as text, music, or images. Popular examples include chatbots like ChatGPT, Google's Gemini, and Microsoft's Copilot.

AI washing manifests in several ways. Some companies claim to use AI when they are actually relying on less sophisticated computing. Others exaggerate the effectiveness of their AI compared to existing techniques or imply that their AI solutions are fully operational when they are not. Additionally, some firms merely add an AI chatbot to their existing non-AI software.

The prevalence of AI claims has surged. According to OpenOcean, a UK and Finland-based investment fund, while only 10% of tech start-ups mentioned AI in their pitches in 2022, this rose to over a quarter in 2023, and it is expected to exceed a third this year. Sri Ayangar from OpenOcean notes that the competition for funding and the desire to appear innovative have driven some companies to overstate their AI capabilities. This trend creates a significant disparity between companies claiming AI capabilities and those actually delivering AI-driven results.

The issue of AI washing has persisted for years. A 2019 study by MMC Ventures found that 40% of new tech firms identifying as "AI start-ups" used virtually no AI. Simon Menashy, a general partner at MMC Ventures, explains that while cutting-edge AI capabilities are now accessible for a standard software price, many firms are simply adding a chatbot interface to non-AI products rather than building comprehensive AI systems.

Dougal Dick, UK head of emerging technology risk at KPMG, highlights that the lack of a universally accepted definition of AI exacerbates the problem. This ambiguity allows companies to exploit the term broadly, leading to AI washing. This practice can have serious consequences for businesses, such as overpaying for technology and services or failing to meet operational objectives. For investors, AI washing makes it harder to identify genuinely innovative companies. Unmet consumer expectations from falsely advertised AI products can erode trust in start-ups doing genuine AI work.

Regulators are beginning to take action. In the US, the Securities and Exchange Commission (SEC) charged two investment advisory firms with making false and misleading statements about their use of AI. This indicates a growing intolerance for AI washing and suggests that more fines and sanctions may follow for those violating regulations. In the UK, the Advertising Standards Authority (ASA) has guidelines against misleading marketing communications, which are increasingly being applied to AI-related claims.

Michael Cordeaux, from UK law firm Walker Morris, notes that AI claims have become more common in advertisements subject to ASA investigation. Instances include exaggerated AI capabilities in app advertisements. Sandra Wachter, a technology and regulation professor at Oxford University, believes we are at the peak of the AI hype cycle and urges a more discerning application of AI, focusing on where it can genuinely be beneficial.

The environmental impact of AI is another overlooked issue. AI technology contributes significantly to climate change, and there needs to be a shift away from indiscriminately implementing AI to a more thoughtful consideration of its use in specific tasks and sectors.

In the long term, the problem of AI washing may diminish. Advika Jalan from MMC Ventures suggests that as AI becomes ubiquitous, "AI-powered" branding will lose its novelty, similar to how "we’re on the internet" is no longer a differentiator.

Overall, while AI holds transformative potential, the phenomenon of AI washing poses significant risks and challenges that need to be addressed through transparency, regulation, and a critical assessment of AI’s actual capabilities and benefits.