While AI is making a revolutionary impact, some vendors falsely label AI to their solutions to boost sales or misunderstand the technology.
Although AI has tremendous potential, it is still widely misunderstood and misrepresented. Take Oral-B’s Genius X toothbrush, for instance. They marketed their product as a toothbrush with AI abilities. Yet, research indicated that only smart sensors were used, not AI.
Many customers and even SMBs connect AI with the science fiction version of a robotic device or an advanced computer that’s faster than any human. Since the technology or algorithm is too complicated to fathom, many assume it to be AI and market it with a picture of Ultron or Terminator. Marketers and advertisers must find the line between intelligent branding and misdirection. Years into its advancements, it is yet to be fully explored. Apart from customer misdirection, it also critically affects the AI perception in the global market.
When AI was new, most tech companies declared themselves as AI companies without fully understanding the technology or possessing relevant expertise. It could not have become worse, or so was thought.
Enter: AI Washing
While some companies misunderstood the technology, many technology vendors misused customer’s lack of knowledge. They labelled their non- AI products as AI or rebranded their older services as AI-driven. Even Gartner cautioned software and technology vendors to use the term in their sales and marketing pitches wisely.
What’s in a name?
There was greenwashing, cloud washing, so AI washing was an expected turnout.
It is always important to be familiar with popular buzzwords that might make a huge difference in business decisions. Sometimes, even if the brand is aware of fake AI-driven products in the market, the rumours or inside information with certain buzzwords like AI washing might not be taken seriously. It is important to keep track of the latest trend names. It all began with greenwashing.
Greenwashing made its first appearance in the 1990s when NGOs began to expose the unfriendly environmental practices of industry giants. It was formed by merging the two words ‘green’ and ‘brainwashing.’ AI washing follows the same origin story.
While ‘fake AI’, ‘pseudo-AI’ and ‘weak AI’ are other common buzzwords companies are aware of, ‘software snake oil’ is a similar buzzword that should send a red signal.
In the 19th century, salespeople sold ‘snake oil’ as an elixir of life but the panacea did not consist of any medicinal properties nor did it have the oil extracted from snakes. It was one of the cons of the century and since then the term snake oil refers to all false promises and fake products.
In the recent times of advanced technology, the word has evolved into software snake oil. While experts believe that the technology industry holds many forms of snake oils from far-fetched assurances to vapourware, the biggest software snake oil that exists today is AI.
Also Read: The Analytical Ladder of Success
Getting Away with an AI SmokeScreen
Experts believe that developers are usually eager to use the term AI as it usually guarantees funding. Even if they do not have the expertise, they are tempted to AI wash. During the dawn of AI, many capital-hungry companies started services in the name of AI but behind the screen, human intelligence was at play. Even today, several companies use labour marketplaces like Amazon Mechanical Turk to outsource AI work.
For instance, a company named Edison Software claimed to have AI engineers going through emails sent by hundreds of users with their identities redacted. In reality, it was humans who did the task. Other companies such as X.ai and Clara had their covers blown for having human beings pose as chatbots for over 12 hours a day. A business expense management app Expensify confessed to having been using low-paid workers to be the brand’s “smart scan technology.”
Two years ago, another startup Expertise.ai that raised almost $30 million for an AI- document scanning and audio transcribing application, had overseas developers do its dirty work. As unethical as it all sounds, the fake-it-till-you-make-it strategy of tech companies was a painstakingly disastrous time for employees, and customers and investors were often duped.
Is AI Winter Coming?
If AI washing continues, people will stop investing in AI technology and the decline in investments can cause a technological and economic breakdown — leading to the fallout of AI completely. A major issue of AI washing that would concern investors is the easy and fragile entry points for hackers to breach customers’ confidential information. Moreover, outsourced human involvement can certainly create gaps and loopholes that might loosen the security protocols. There have already been two winters in the twentieth century, and a third in our century could be catastrophic.
Watch Out for the Telltale Signs
Today, several industry leaders are sceptical. Experts strongly suggest B2B and B2C customers be wary of products, services, or solutions that claim to be driven by AI. Technology experts have guaranteed effective strategies and red flags that can help customers determine if it’s genuine or another act of AI washing.
‘Minimum data training required’
If a marketer states that a limited amount of normalised data is required for model training, companies need to take a second look at the company. AI-based solutions usually demand a substantial amount of data to perform accurately. Even basic ML models require thousands of data points and AI requires much more than that. Hence, a lack of adequate data requirements is a sign of AI misuse.
‘If it has to work, it needs business rules’
Experts verify that if the vendor shows signs of “If this, then that,” the solution might not be AI-based. AI is an advanced technology that wouldn’t require hand-holding, to say the least. Less- intelligent technology like automation requires business rules.
‘We are not sure’
The more an AI solution is deployed with massive data sets, the more learning capabilities it builds. If a vendor fails to demonstrate examples about how the solution has improved over time, it might not be a genuine AI solution. Moreover, if the vendor fails to adequately explain the working capability of the model, it is another sign of AI washing. Simple questions like “What kind of AI is this?”, “How much human interaction is required?”, and “What exactly does it do?” can provide a deeper understanding of the vendor and the offered service or tool.
‘We do not share technical details, but we have happy customers’
It takes time for a firm to be an AI company. If the self-claimed AI firm has leveraged actual AI, their biggest proof would lie in their case studies. On asking for references, the vendor might point to the success stories on their website, but experts urge customers to take a closer look. Sometimes the success stories talk about leveraging AI but do not provide details about the data, training algorithms, or the working model of the AI. It’s a possible indication the supposed AI firm is indulging in AI washing.
How informed are the vendors?
Facebook is said to have over 100 employees dedicated to AI research. As fascinating as it sounds, all global vendors cannot be expected to be deep-pocketed. While tech giants have the funds to hire the best, businesses looking at SMB vendors must be cautious. An educated background in AI or Deep Learning (DL), certified expertise, or a self-taught valid education is vital for every vendor selling an AI-driven product. The company should study the vendor and his team of data scientists, mathematicians, architects, and engineers who create AI models and teach them to analyse situations and produce future actions with data insights. Sometimes, ideation and visualisation are more important than the mechanics of the model itself.
Also Read: Data Analytics, Powering the IoT Evolution
The Benefit of the Doubt
Some technology experts allow certain leeway. As there are over 200 disciplines that AI covers, a vendor using even one branch is technically selling an AI-driven product. Experts reckon that some advanced analytics are now classified as AI and even a little ML makes the cut. They believe it is hard to draw a technical line. The bigger picture still remains that further AI washing episodes could hamper further technological advancements.
According to an Allied Market Research report, the AI market that generated over $4 billion in revenue in 2018 is predicted to reach over $53 billion by 2026. Today, among Forbes’ top 25 Machine Learning startups for 2020, 11 are branding under the .ai domain name. No wonder every technology vendor wants to be known as an AI solution provider, even if they have to take unethical measures like AI washing.
Companies need to understand that AI is not supposed to be a Wizard of Oz technique. While the technology is advanced, its full capacity is still being explored, and it has its limitations. While AI is the headline of the future, companies need to carefully tread over the hype and view AI as it should be. Experts hope AI washing will soon become a forgotten stain in business history.