Consumer-facing artificial intelligence (AI) agents in the form of digital assistants and chat bots have enormous potential. They are ushering in a new era of interactive and personalized consumer engagement, bringing great value for advertising, marketing, and customer services. With the integration of payments, intelligent agents could also provide an end-to-end shopping experience. This is generating a lot of excitement and activity. But service providers must be wary of the hype surrounding intelligent agents and not underestimate the challenges that accompany them.
A fast-growing ecosystem generating a lot of heat
One of the key drivers motivating consumer-facing tech companies, OTT players, e-commerce players, and others to invest in digital assistants and the AI behind them is the potential intelligent agents present to pull consumers more deeply into service ecosystems, which could lead to higher revenues. Amazon (Alexa), Google (Google Now), Microsoft (Cortana), and Apple (Siri) are increasingly embedding their digital assistants into multiple connected-product offerings, spanning mobile devices, smart TVs, connected cars, and the smart home.
Major messaging players are opening up their messaging applications to chat bots in a bid to position themselves as platforms for a wide range of digital content, commerce, and payment services. The ultimate goal for these players is to move away from a model where users interact with individual apps, to one where their platforms mediate interactions with multiple chat bots. This enables the messaging platforms to increase the range of services they offer and in doing so gain a competitive edge. At the same time, the large user base of popular messaging platforms provides chat bots with the potential to enjoy significant scale and reach. Facebook Messenger and Facebook-owned WhatsApp have both reported 1 billion monthly active users each.
Do not underestimate the challenges involved
Although AI technology is advancing quickly and intelligent agents will get smarter, the reality today is that most consumer-facing AI applications are relatively basic. This is particularly the case with chat bots. Service providers must not overplay the capabilities of intelligent agents and what they can deliver. Many consumers already have a misguided view of what AI is thanks to the supercharged scenarios imagined by Hollywood. There is a danger that consumers' inflated expectations of intelligent agents will lead to disappointment and frustration.
There has been a lot of talk about how easy chat bots are to implement compared to traditional applications. Although there is truth in this, it does not mean that chat bot development should be underestimated. In fact, the very limitations imposed by chat bots mean they require thoughtful, careful design. Chat bot interactions are very direct and require a well-designed, conversational user interface with a clear, intuitive structure. Service providers need to think carefully about whether voice, text, or both is the most appropriate for what the chat bot is trying to achieve.
Mobile advertising always faces the danger of being intrusive, but this is magnified in the context of conversational platforms such as digital assistants, messaging apps, and associated chat bots. Many consumers will simply not want to invite brands and advertisers into spaces that they consider personal. Advertising in the intelligent-agent context must be 100% relevant and well targeted if it wants to become a legitimate part of the conversation. Brands should not barge in with a hard sell, but instead demonstrate value through tailored recommendations before moving on to commerce.
AI-powered digital assistants and chat bots have the potential to become highly efficient profiling tools. This can benefit users if handled correctly, but users must be kept informed and put in control, particularly in terms of how their data is shared. New privacy rules and safeguards will be needed.
AI applications must be introduced carefully and responsibly. They also need to be stress tested to ensure that they behave as intended. Service providers should think carefully about the environment in which they release an intelligent agent. No service provider wants to find itself in Microsoft's position when its Tay chat bot went rogue on Twitter and assimilated the negative views of some Twitter users it had interacted with. As well as being thoroughly stress tested, intelligent agents must come with built-in safeguards to enable them to not fall prey to negative interactions.
Over the coming weeks, Ovum will be publishing a series of reports on intelligent agents in consumer commerce, focusing on market context, market dynamics, and commercial prospects.
Straight Talk is a weekly briefing from the desk of the Chief Research Officer. To receive this newsletter by email, please contact us.