The Most Advanced AI in the World Is Worthless Without This One Key Element
AI is here to stay, and there’s a simple reason why: proven ROI.
Businesses in multiple industries are incorporating AI because the upside is tangible. According to an Accenture report, implementing AI solutions boosted profitability by an average of 38 percent across 16 industries. For the U.S. economy as a whole, this boost could lead to an additional $14 trillion in gross value added by 2035.
AI offers businesses more than just incremental improvement; it’s a transformative technology. By using AI to automate repetitive and time-consuming tasks, you can free up valuable human resources for higher-value work. AI also enables better processes in lead generation, marketing, and product development workflows, which eventually leads to more innovative products — and a better customer experience.
AI will become a critical component of the customer experience as traditional methods of customer segmentation become less and less effective. By using AI, companies will be able to scale while still creating more individualized experiences, such as custom incentives or rewards based on current location-based services, which are quickly becoming the customer expectation.
But even the most groundbreaking technology won’t be any good if no one can use it, and in many cases, user experience is the limiting factor when it comes to successful AI implementations.
Users Are Being Left Behind
AI is an emerging technology with a lot of experimentation still in progress. Many companies are trying to figure out the standards as they go, using customers as lab rats. Such companies often lack a clear vision for how individuals or businesses will use their product. Worse, they don’t recognize the problem. According to Capgemini, 75 percent of organizations think of themselves as customer-centric, but only 30 percent of consumers share that view.
Companies are learning that the tech they’re investing in doesn’t automatically create a good user experience. Take home conversational assistants, for instance. An IDC report predicts that the adoption of cognitive systems that mimic the human brain, like the systems found in virtual assistants, will increase AI revenues from $8 billion in 2016 to more than $47 billion by 2020. In the race to create the next voice user interface VUI, companies invested less in the user experience, thinking it would be easier to design because there’s no visual component.
In reality, VUIs are harder to design for the exact same reason — the assistants are currently too reliant on a specific set of voice commands, pronunciation, and syntax. One usability study by the Nielsen Norman Group found that conversational assistants were “close to useless for even slightly complex interactions.”
In a number of industries, companies are racing to generate revenue from this emerging technology, creating complicated interactions for consumers. From business intelligence dashboards to your smartphone, every new technology comes with a learning curve for users. But AI often works so quickly and so dramatically that it causes widespread consumer-side problems that tech teams never foresee.
Missing Out on Trust
AI only works with continuous feeding of data, and data has been shown to introduce various demographic biases into models. If that data does something to break consumer trust — if a ride-sharing algorithm routes drivers away from certain parts of the city or distributes gigs to drivers in an imbalanced way, for example — then the relationship can be difficult to repair. Customers value human qualities like morality and fairness, but AI algorithms don’t always deliver. When Microsoft deployed a conversational chatbot powered by real-time AI on Twitter, for example, the bot was almost immediately corrupted by trolls. Bias in AI is a prominent enough issue that Google even created a tool to test for it.
As evidenced by the many security breaches (Best Buy) and invasions of privacy (Facebook), companies are forgetting that trust is part of the customer experience. Customers are hesitant to work with companies that aren’t open about how they treat user data. It’s easy to see why only 20 percent of consumers ”completely trust” organizations to keep user data private even though 78 percent say that data privacy is “extremely important” to them. Trust is part of the customer experience, and if AI can’t earn that trust, companies have a lower chance of earning true loyalty.
That’s why companies that look at the customer experience when integrating AI have an opportunity to pull ahead of competitors. While those in the industry have known about the internal advantages offered by AI and data automation for some time, mainstream consumers are just starting to recognize how much potential AI has to improve their everyday lives. Businesses that focus on user experience in their AI products have the most to gain in the years ahead.
Curing What Ails AI
In order to build a better user experience, product teams must hit on a bare minimum of three key areas with an AI product.
1. Focus on ease of use
If customers need a degree in machine learning to use an AI product, there’s a problem. AI should deliver quick wins and be easy for users to figure out. It’s also important that products have low barriers to entry to encourage habitual usage. As norms are established, more complexity can be introduced. (This applies to VUIs, for example, as well as business intelligence dashboards and other outputs of AI.) If you must, make products that deliver immediate value first and then consider a different business model for more advanced features when customers know you can deliver what you promise.
2. Prove that customers can trust you
Customers want to know that their data won’t ever be compromised or misused, and a good customer experience will help you communicate that. Reliability and consistent uptime are also critical to earning consumer trust: If you want customers to believe in your product, it needs to be available all the time, but in a way that doesn’t invade privacy (e.g., VUIs that are only listening for a certain number of seconds at a time). A better customer experience is the first step toward earning trust, and that type of relationship will help boost the business outside of traditional marketing channels.
3. Eliminate possible bias
Ensure that AI doesn’t output data that excludes important areas of your target consumer base by introducing biases into the model. AI systems are built to find patterns and may correlate inputs from outcomes, so be sure to consider and test the limitations of AI. Know the lines where abuse of the system may create unexpected or undesirable output. Remember that AI cannot distinguish right from wrong or truth from lies. This is why people have begun to worry about the information they receive from Facebook’s personalized news feed algorithm.
With so much potential upside and amazing possibilities in AI, it’s easy for businesses to forget about the humble users — especially when they’re trying to justify an investment in the technology. But the customer experience is what will ultimately separate successful AI projects from unsuccessful ones. Even if their projects aren’t as flashy, tech teams that can deliver a product that’s easy to use and easy to trust will ultimately be the ones that get the most out of their AI investment.