Editorial
The Gist
- Real risks. Misrepresenting AI can lead to legal action, hefty fines, and lost customers.
- Easy isn’t enough. AI that just makes a product easier to use is classic AI-washing.
- Data is the difference. Avoid AI-washing with proprietary, high-quality data that makes your AI smarter.
You’ve probably heard the phrase “putting lipstick on a pig” to describe a situation where someone is dressing something up to make it look new—even when it’s fundamentally unchanged.
That’s one of the best ways to describe “AI-washing” in tech, which happens when someone slaps an “AI” label on existing processes without materially improving the underlying data, models, or outcomes. Sure, it might look like AI, but underneath it’s still the same old workflows and rules.
Table of Contents
The Perils of AI-Washing
We’ve seen this type of overselling before. When SaaS became popular, people joked that it was really just “Software at another Server,” essentially moving client-server applications from a corporate data center to the cloud. Eventually, we got true SaaS, but not before a lot of “cloud-washing.” Similarly, you’ve likely seen “green-washing” in retail, as companies use terms like “natural” to make their products seem more environmentally friendly than they actually are.
Common Signs of AI-Washing
A quick reference for CX and marketing leaders to identify exaggerated or misleading AI claims.
| Claim | What It Really Means | Impact on Customer Experience |
|---|---|---|
| “AI-powered” interface | Likely just a natural language wrapper over existing workflows | Little or no improvement to outcomes; customers can tell it’s superficial |
| “Smart assistant” or “AI chatbot” | Decision-tree logic or scripted responses | Frustration rises when the bot can’t solve real issues |
| “Predictive insights” | Generic third-party models with no proprietary data | Low accuracy and repetitive suggestions that feel impersonal |
| “Enterprise-grade AI” | No visibility into training data or model provenance | Inconsistent answers erode trust and drive churn |
| “Personalized recommendations” | Simple rules based on past clicks or broad segments | Generic experiences that fail to feel relevant or helpful |
When it comes to AI-washing, the risks are significant, including:
- Legal consequences: Last year, the U.S. Securities and Exchange Commission (SEC) settled charges against companies that made “false and misleading statements about their purported use of artificial intelligence (AI).” The firms were censured, ordered to cease and desist, and paid hundreds of thousands of dollars in civil penalties. The Federal Trade Commission’s “Operation AI Comply” is also targeting companies that use AI in “deceptive or unfair conduct” for consumers.
- Poor customer experience. As customers become more savvy and experienced with AI, they expect it to perform a wide range of tasks, from navigating their cars to monitoring their health—and much more. When marketers make exaggerated claims about AI, customers will likely see through the misleading claims and switch to a competitor.
- Loss of control. Putting a thin wrapper on someone else’s API may let you deliver AI, but you don’t know what data it’s trained on and what answers it’s going to give your customers. Simply adding your logo to a platform doesn’t give you any real competitive advantage; that so-called “AI” is a commodity that keeps you tied to someone else’s data and algorithm.
Related Article: The Hard Truth About Human-Like AI Conversations
The Test for AI-Washing
Ultimately, it comes down to one question: Does the platform or product have data that makes the AI smarter, or is AI just making it easier to use?
Consider a natural language front-end for a workflow. Yes, technically it’s AI—but it’s just masking the complexity of the workflow with a more accessible customer interface. When AI simply makes the product easier to use, that’s AI-washing.
AI-washed chatbots can enhance your customer experience, but they don’t actually deliver the full value of AI. If an airline kiosk uses “AI” to guide users through a decision tree, that isn’t really AI. However, the same airline might utilize true AI to review a customer’s travel history and other proprietary customer data to offer an upgrade to first class. That’s where the real value is, for companies as well as customers.
True Enterprise-Grade AI Depends on the Data
AI is only as good as the data that fuels it. As an Enterprise Strategy Group research report explained, “the better the data, the better the AI.” To determine if a product truly deserves the “AI” label, here’s what I look for in the data:
- Proprietary, first-party data. Your data is your competitive advantage. Your competitors don’t have your sales data or customer feedback—which means they can’t build an AI platform that leverages these insights. When you own (and can audit) your data, you know it’s clean, and you can avoid many privacy and regulatory concerns. External data is fine for solving common problems, but it won’t help you get as far ahead, since everyone has access to the same generic data.
- High-resolution data. Collecting data more frequently, reducing noise in the data, and using a diverse range of sources can enhance the quality of the data and the AI platform. For example, fewer gaps in the data allow for better pattern recognition, which leads to more accurate models and predictions.
- Empirical (real) data. Data grounded in reality, not hypothesis, is typically more trustworthy and valuable. Of course, you’ll still have to make decisions based on models—but using empirical data can make them more reliable.
- Contextual data. The supporting data can make all the difference in the quality of AI. If one of your POS devices crashes, for example, it’s helpful to know what the machine was doing, what apps are installed, what network it’s connected to, and other relevant data, which allows for faster mean time to resolution (MTTR). Contextual data often reveals what’s really happening, so the AI can deliver more accurate solutions.
Unfortunately, we’ll likely continue to see a fair amount of AI-washing as everyone tries to promote AI-driven products and platforms. But hopefully, we’ll get more and more “true” AI as customers continue to demand the benefits that only real AI can deliver. In the meantime, keep watching out for AI-washing—and gain a competitive advantage by turbo-charging your AI with proprietary, high-resolution, contextual data.
Learn how you can join our contributor community.
