AI Washing: How to assess the real value of AI

min readpublished onupdated on

With the advent of technologies like ChatGPT, artificial intelligence (AI) has become more than just a technological term; it's a significant part of our daily lexicon. This surge in popularity, however, does not imply that AI is a novel concept. AI has been a staple in consumer technology for years, evident in devices like smartphones, smart home devices, wearable fitness trackers and navigation apps. These applications, though, represent only the tip of the iceberg in terms of AI's capabilities and potential. 

AI washing
AI Washing: How to assess the real value of AI
Table of contents

The AI Bandwagon: More Than Just a Buzzword

The real game-changer in recent years has been the rise of generative AI. This type of AI goes beyond the predefined functionalities of earlier applications, venturing into the realm of creating new content, whether it be text, images, or even code. The introduction of user-friendly interfaces, as seen in GPT models, has made these advanced AI capabilities more accessible and intriguing to the general public and the corporate world.

For investors, this evolving landscape signifies both opportunities and challenges. While the potential for innovation and disruption is immense, it also requires a nuanced understanding of what AI truly is, and perhaps more importantly, what it isn't. Investing in AI technology necessitates a deep dive into the mechanisms, capabilities, and limitations of different AI models. This understanding is necessary not only for making informed investment decisions but also for assessing the long-term viability and ethical considerations of AI-driven companies.

The Spectrum of AI: From Narrow to General AI

While the innovation and potential of AI are evident, it's helpful to delve deeper into the various forms of AI to appreciate its full scope and impact, especially for those investing in this technology. 

Narrow AI: The Specialist

Narrow AI, or ANI, as the name suggests, specialises in performing specific tasks. Examples include AI systems designed for playing chess, recognising speech, or identifying objects in images. These systems are highly efficient at their designated tasks, often surpassing human capabilities. However, their expertise is limited to a narrowly defined domain. 

Broad AI: The Domain Expert

Broad AI represents a more advanced and versatile form of AI, capable of handling multiple tasks within a specific domain. This type of AI is particularly relevant in fields where complexity and variability are high. In healthcare, for example, Broad AI can not only interpret medical images but also analyse patient histories, laboratory results, and suggest treatment plans.

General AI: The Future of AI

General AI, often referred to as Artificial General Intelligence (AGI) or Artificial Super-Intelligence (ASI), is the holy grail of AI research. It aspires to replicate human-like reasoning and problem-solving abilities across a wide range of domains. Unlike Narrow or Broad AI, General AI is not restricted to specific tasks or fields. Understanding these different AI types is just the first step. Equally important is recognising how AI is integrated into products and services, which brings us to the issue of AI washing and the need for genuine AI utilisation.

Understanding AI Integration

The integration of Artificial Intelligence into devices and systems has been revolutionised by the advent of APIs provided by major players such as OpenAI, Google, and Anthropic. This technological advancement has simplified the process of embedding AI into various products and services. As a result, we've seen the emergence of what's commonly known as AI "wrappers", where AI capabilities are integrated at a surface level, often without deep, intrinsic functionality.

This ease of AI integration, however, comes with a caveat. As Kjell Carlsson, the head of AI strategy at Domino Data Lab, points out, creating transformative AI-enabled products requires a significant enhancement of in-house AI capabilities. Without this, companies might only achieve limited GenAI capabilities. This situation often leads to "AI washing", where AI functionalities, which have been available for decades, are rebranded as cutting-edge innovations without substantial improvements in their capabilities.

AI Washing: The Marketing Masquerade

AI washing" represents a significant challenge in the tech industry. It is a deceptive marketing practice where the extent to which AI is used in products or services is misrepresented. Often, companies use the term “AI” loosely to describe what are essentially basic automation or rule-based systems. These systems lack genuine learning and adaptive capabilities, which are hallmarks of true AI.

This misrepresentation is not only a disservice to consumers but also poses a significant challenge for investors in differentiating between genuine AI innovation and mere AI branding.

Startling Misrepresentations

The prevalence of AI washing is alarming, as seen in numerous startups that have exaggerated their AI capabilities to attract funding and customers. A striking example of this was a startup claiming its "human-assisted AI" could enable the development of mobile apps with minimal effort and time. This claim led to substantial investment, including nearly $30 million from AI-focused venture capital funds. However, a Wall Street Journal investigation revealed that the so-called AI was, in fact, largely reliant on good old fashioned “human intelligence” of software engineers.

This incident is not an anomaly. According to a study by MMC Ventures, which analysed 2,830 European startups, a startling 40% of companies claiming to be AI startups had minimal actual AI utilisation. This finding, along with heightened regulatory oversight, highlights the critical need for investors to perform thorough tech due diligence when evaluating AI-centric companies.

The Regulatory Lens: SEC and FTC on AI Washing

Regulatory bodies like the Securities and Exchange Commission (SEC) and the Federal Trade Commission (FTC) have become increasingly vigilant about the practices of AI washing. Their concern stems from the potential for AI washing to mislead investors, consumers, and the market at large, leading to unfair practices and a distortion of the competitive landscape. The SEC, primarily concerned with protecting investors and maintaining fair, orderly, and efficient markets, has taken a clear stance against AI washing, particularly in the context of public company disclosures. SEC Chair Gary Gensler has emphasised the importance of truthful and accurate representations of a company's AI capabilities. His comparison of AI washing to greenwashing is particularly powerful given the SEC's track record in combating greenwashing – thus providing a useful blueprint for its approach to AI washing.

FTC attorney Michael Atleson has outlined the agency's stance in a blog post, emphasising the importance of truthful labeling of AI-powered products. The FTC expects companies to avoid exaggeration in their AI claims and to ensure that any performance assertions are scientifically supported and applicable universally, not just under specific conditions.

For tech investors, this regulatory landscape means that due diligence must extend beyond financials and into the veracity of a company’s AI-related statements. Companies that overstate their AI capabilities or potential could face regulatory scrutiny and financial penalties, affecting their valuations and investment attractiveness.

AI Due Diligence: A Critical Factor for Investors

For dealmakers, the risk of inflated valuations due to misrepresented AI capabilities is a pressing concern. Beyond financial repercussions, there's the potential for reputational damage and regulatory scrutiny.

To navigate this complex terrain, it's imperative for investors to engage tech due diligence experts who specialise in assessing AI technologies. These experts offer an in-depth understanding of Artificial Intelligence and its practical applications, helping investors to discern genuine AI innovations from mere AI branding.

Key Areas of Focus for Tech Experts

Technology due diligence in AI encompasses several critical areas:

Authenticity of AI Capabilities

Tech experts begin by assessing whether the technology in question truly possesses AI capabilities. This involves understanding if the AI is merely an extension of basic automation or rule-based systems, or if it embodies advanced machine learning techniques.

Critical questions include:

  • What level of machine learning is being used? (e.g., supervised, unsupervised, reinforcement learning)
  • What type of data does the AI learn from, and how does it adapt to new data or feedback?
  • Can the AI system improve its performance over time, and is there scientific evidence to support performance claims?

Adaptability and Robustness

Another vital aspect is the AI system's ability to handle unexpected or out-of-distribution data. Tech experts evaluate whether the AI can adapt to new situations or if it requires retraining. This assessment is necessary for understanding the AI's long-term viability and adaptability to changing market conditions.

Risk Assessment

A comprehensive risk assessment focusing on what might happen if the AI does not function as predicted is essential. This includes scrutinising the safeguards in place to prevent biased or discriminatory decisions and the potential impacts of these risks on the company's operations and reputation.

Conclusion: A Call for Prudent AI Investment Strategies

For investors and dealmakers, the allure of AI-driven innovation is undeniable. However, in an era where AI washing is prevalent, it's vital to exercise discernment. Understanding the true extent of AI integration and its practical applications is key to making informed investment decisions. By prioritising responsible AI practices and thorough tech due diligence, investors can navigate the complex AI landscape with confidence.

Contact us for more information on AI evaluation

References:  (last visited 25 January 2024). (last visited 25 January 2024). (last visited 25 January 2024). (last visited 25 January 2024). (last visited 25 January 2024). (last visited 25 January 2024). (last visited 25 January 2024). 

Kristin Avon Senior Legal Officer Vaultinum
Kristin A.Kristin is a registered US attorney specializing in the areas of IP and technology law. She is a member of Vaultinum’s Strategy and Legal Commissions charged with overseeing and implementing the policies and processes related to the protection of digital assets.

Recommended for you