Smoke and mirrors: How ‘AI Washing’ is fooling the tech industry

Anywhere you turn and look, you hear about AI. It’s like that trending summer song you hear at the gym, the local shop, and work, and when you get home, your wife has it playing in the living room.

Although AI is a huge, transformative technology that will change the world, in 2024, the tech industry will be infused with AI. It has huge potential, and its benefits outweigh the negatives; so much of the talk about it is hype and hot air.

Only in 2023 did investments in AI worldwide surge to 140 billion dollars, which means that if you do not have AI in your business, you are a step behind your competitors. Some founders feel that unless they mention AI in their pitch, this might put them at a disadvantage, regardless of its role in their solution,” said Ayangar , part of the Open Ocean team.

But behind all the AI gold rush hype lies a growing problem: Not all that glitters is AI.

What is AI washing?

AI washing is a marketing tactic companies employ to exaggerate the amount of AI technology they use in their products. AI washing aims to make a company’s offerings seem more advanced than they are and capitalize on the growing interest in AI technology.

So, the company implies that its AI models and algorithms are more powerful, useful, or flexible than they are, overrating the AI’s capabilities.

Another way marketers attempt to AI-wash their products and services is when they downplay the amount of human input involved – either by the service provider or the user. Essentially, what’s important is not whether something is called “AI,” or even whether it has a particular computational architecture but whether that computing power is doing something valuable you and I couldn’t do otherwise.

Why is this a serious issue?

From a less critical lens, this tactic could seem harmless or easily swept under the rug without serious repercussions. Well, it isn’t, and I’ll tell you why. It can stifle or obscure true innovation as real AI breakthroughs need help to be noticed amidst the hype and noise of many exaggerated claims. In a 2019 study , it was found that 40 percent of new tech firms that described themselves as “AI start-ups” used virtually no AI at all.

On an even deeper level, AI washing hides monoculture in the industry. A monoculture in computer science is when a group of computers run identical software.

So essentially, companies showcasing their own “unique” AI models might use the same technology but employ different marketing strategies to promote their products. This could lead to a future financial crisis if many financial institutions rely on the same underlying models. It can also lead to inflated expectations around what AI can achieve, which could result in businesses setting unrealistic goals and targets.

Amazon’s “Just Walk Out” technology

Many companies offer tools that claim to automate the creation of videos, copy, and content. However, these tools often require significant human input to generate an acceptable-quality output.

For a specific example, Amazon’s Just Walk Out technology promised to use cutting-edge technologies like facial-recognition cameras, sensors on the shelves, and, of course, “AI.” The promise of the “Just Walk Out” stores was that customers would not need to queue for a cashier, scan their items, or pause on the way out. They could take what they needed, walk out the door, and the benevolent all-seeing eye of technology would seamlessly price their goods, charge their account, and send them a receipt.

Now, it has started to get interesting, as it was reported that Amazon was employing people from India to watch the cameras and label the footage of shoppers. A team member who worked on the technology said that actual humans–albeit distant and invisible ones, based in India–reviewed about 70 percent of sales made in the “cashier-less” shops as of mid-2022.

Just Walk Out was first introduced in 2016 , presenting Amazon’s biggest and boldest innovation in grocery shopping. Eight years later, Amazon decided to discontinue its service.

If big brands, like Amazon, are being found out for AI washing, this poses a serious trust problem. One of the biggest barriers to greater adoption of AI is a general need for more trust in accuracy. Receiving accurate output from generative AI tools is no small engineering feat. Not all solutions that brand themselves as enterprise-ready generative AI tools can deliver the accuracy needed at work.

How to spot AI washing

Brands want to invest in AI across their business, making them vulnerable to AI-washing and illicit activity in many areas. Luckily, there are information-gathering questions you can ask potential companies.

• It’s important to ask if a company can demonstrate their software technology out of the box, especially those that claim to have generative AI. The answers here will reveal potential performance gaps and provide an immediate gauge of whether the software is real. The world is moving on from slides with AI accelerators, and a true software partner should be able to demonstrate their solutions convincingly.

• Some companies see traditional and generative AI as additional tools to an already robust software or technology. Companies don’t want to be tricked into investing in just the original standard platform but something with a “Now we use AI!” sticker. It also helps determine if the company has open-source tools or API calls connected to third-party services (for reasons that are not just pragmatism). It also reveals if the solution can be fully customized to the company’s needs; if not, its capabilities are just repackaged tools that have been democratized in the market.

• A key differentiator of a great AI platform is the ability to integrate solutions and AI modeling into your organization’s cloud. CPGs recognize the need to share data with sources outside their organization to gain AI-powered insights. A true AI platform gives the toolkit to customers; data doesn’t need to be removed from a brand’s technology framework or customer environment. Suppose a solution partner wants to have customer data and deploy that data in a sandbox or cloud. In that case, it suggests that the partner is a technology services company building something from scratch on a contract basis.

• An honest AI platform can harmonize these diverse datasets and deliver synchronized results in one place, backed by a limited number of data scientists, primarily through continuous monitoring and learning in the technology itself. Brands must realize that having a team of data scientists monitoring AI performance means no AI solution. After harmonizing the data, the generated insights and results should be explainable through predictive models, optimized recommendations, or rich voice output.

The path forward

So, where do we all go from here? Well, if you ask Douglas Dick , UK head of emerging technology risk at accountancy giant KPMG, he will tell you that the issue is because the term “AI” does not have a standard definition. He argued that the ambiguity of AI’s definition is the reason for the emergence of AI washing.

Everyone can agree that supervisory bodies will be important to regulate the use of AI in products. The US Securities and Exchange Commission (SEC) has charged two investment advisory firms with false and deceptive statements about AI usage, indicating a lack of leeway in AI washing regulations, potentially leading to more fines and sanctions.

The UK’s Advertising Standards Authority (ASA) has criticized AI claims in commercials, stating that they are becoming more common. The ASA found a paid-for Instagram post about an app, “Enhance your Photos with AI,” to be misleading and deceptive, highlighting the increasing scrutiny of AI claims in advertising.

Sandra Wachter, a professor of technology and regulation at Oxford University and a leading global expert on AI, tried to be positive. She said, “AI’s practicality and environmental impact are often overlooked in discussing its application in various industries.” She argues that AI contributes more to climate impact than aviation and should be focused on specific jobs and industries where it can be useful rather than unthinkingly implementing AI everywhere.

Advika Jalan, head of research at MMC Ventures, predicts that ‌AI washing may eventually subside as AI becomes increasingly ubiquitous, making it less of a differentiator in branding, similar to saying, “We’re on the internet.”