Will Facebook Chat Bots be Trustworthy?

OpenMarket – May 24, 2016

Gartner self service without human interactions

If the above Gartner prediction becomes a reality, then chat bots are a strong candidate for how businesses will automate self-service with their customers. Combine that prediction with the belief that customer experience will overtake price and product as key differentiators for many businesses. And now factor in the importance for businesses in providing an exceptional mobile engagement solution. All of this leads to an interesting set of choices for consumers. It begs the question: will mobile consumers primarily adopt SMS or Facebook Messenger chat bots and what are the trust and privacy considerations for consumers?

What are chat bots?

Let’s back up and provide some context before discussing this trustworthy question further. In case you haven’t heard of chat bots, they are software programs with human-like conversational abilities. Their conversational skills are powered by a form of artificial intelligence called machine learning. Machine learning enables software programs to provide conversational abilities without being explicitly programmed.  Essentially, machine learning allows chat bots to become better at understanding context so they can have more intelligent conversations with people. Read this article for more background information.

2016 is quickly becoming the “year of the chat bot.” Microsoft’s Tay and Facebook Messenger are examples in the market today.  Microsoft’s Tay started out as a naïve innocent bot that quickly became badly influenced by the Internet (no surprise). Facebook announced support for developers of chat bots in Messenger at the 2016 F8 conference last month. Dag Kittlaus and Adam Cheyer, the creators of the artificial intelligence behind Siri, Apple’s iconic digital assistant introduced in 2011, showed off their newest creation, Viv, a next generation AI assistant, that they have been developing in stealth mode for the last four years.

I would note, however, that chat bots have been around for a very long time. You can find over a thousand chat bots on chatbots.org with the oldest one, Eliza, from 1966. Eliza was named after Eliza Doolittle, the character from George Bernard Shaw’s play Pygmalion who was taught to speak with an upper class accent.

What has changed now?

The current belief is that machine learning is advancing sufficiently and more quickly to power these chat bots. Also, the pervasiveness of text messaging and mobile messaging apps make them ripe communications channels for the inclusion of chat bots.  As of last year, we’ve crossed the tipping point of having more millennials than baby boomers in the US. In fact, millennials make up over a quarter of the nation’s population, and our recent research (see infographic on why millennials would rather text than talk) shows that millennials really prefer texting with businesses.

What is the state of Facebook’s Messenger chat bots?

Some of Facebook’s target use cases are for customer service, commerce and content distribution, and they are in the early stages of development now. Messenger’s head of product, Stan Chudnovsky, indicated that they have over 2,500 merchants on Shopify and Facebook is still figuring out what use cases and behaviors to allow or disallow.  For instance, bots can be chatty and thus quickly viewed as spammy on Messenger. This is why Facebook will be building chat bot analytics systems to help spot bots/messages that are annoying users and may cause them to block messages or abandon a bot.

What about the future?

So this leads us to the central question of this blog: will consumers trust using a set of business chat bots from a single platform like Facebook Messenger that also collects a ton of personal Facebook demographic data? To explain the context for this question, let’s look at what happens today for customer service using just a single mature communications channel: voice.

When we call our financial service provider to inquire about an account status change or call our cable company to report a service issue, the phone company isn’t listening to our conversation and mining our personal data or interactions in order to create a better customer engagement experience or to advertise. Hence, there is no information sharing with the communications service provider (your phone company) when you are interacting with a set of different businesses using the same communications channel: voice.

Think about how digital channels such as Internet browsing and free hosted e-mail services are used. We see that data mining happens fairly regularly, but it is still somewhat siloed. It can be based on Internet searches, our hosted e-mail service accounts or website activity. After I login to my financial service or healthcare provider’s websites, my search engine provider (e.g. Google) doesn’t really know what the nature of my interactions are with each of those businesses.

Now consider a mobile channel like text messaging (SMS). Each business is using its own A2P (Application-to-Person) messaging program as a silo and your mobile carrier (i.e. AT&T, Verizon, T-Mobile) and the business’ SMS provider isn’t mining those messaging programs to create a digital profile of the consumer across all businesses that he or she chooses to text. Thus, the consumer’s digital privacy is protected via this walled garden of individual business-to-consumer communications just like in the world of voice communications.

What if a consumer primarily uses Facebook Messenger with chat bots for all of their business communications?

Given that all these chat communications are digital, the information is already digitized and ready for massive automated data mining to create digital profiles with predictive analytics. Facebook has indicated they will offer a data analytics service to businesses so a company can get insights (i.e. mine an individual’s interactions with a brand) just like businesses can do with other individual channels. But I fear that Facebook already has a ton of demographic data for about 900 million consumers and they will combine that demographic data with a consumer’s business interactions to build detailed digital profiles like we’ve never seen before in any industry or from any single communications channel provider.

This would represent an unprecedented level of digital consumer profiling by a single public company.

What are the safeguards for consumers? What about consumer’s right to be forgotten by Facebook chat bots? How will consumers be able to establish their own firewalls or how much information they are willing to share?  What is the risk for a business if they use Facebook digital profiles to try to enhance the consumer experience with their customers?

Assuming that millennials will be the early adopters of chat bots, some people believe that this generation doesn’t care about their digital privacy. Furthermore, many folks assume millennials are accustomed to sharing their private information online to the degree that they will happily adopt Facebook Messenger chat bots as their predominant channel for automated self-service business communications. I personally don’t subscribe to that belief when it comes to chat bots. One recent survey shows that millennials do care about privacy and securityEighty percent say it is vitally or very important that PII (Personally Identified Information), financial, and/or medical data be shared only with authorized parties. According to the latest survey and analysis by the National Telecommunications and Information Administration (NTIA), Forty-five percent of Americans are concerned enough about privacy and online security to refrain from conducting business transactions online. Twenty-three percent of Americans in general are growing increasingly concerned about data collection by online services. Twenty-two percent had major concerns about loss of control over personal data including when using mobile devices like smartphones.

I believe that SMS business-to-consumer chat bots will differ from Facebook Messenger chat bots in how they address this very important privacy and consumer trust issue. Public trust in the US government remains historically low and governmental oversight generally lags the speed of technology evolution, so there is likely to be an oversight gap with Facebook-based chat bots. In addition, another recent survey shows that consumer trust in businesses with their personal data safety is abysmal. One compelling stat shows that “only about two in 10 Americans say they have a lot of trust in the companies they regularly do business with to keep their personal information secure.” So why should consumers trust Facebook with an unprecedented ability to store their personal demographic information as well as see into their personal interactions via chat bots with all of the businesses that a consumer values?

How would SMS chat bots differ from Facebook Messenger chat bots?

Each business’s SMS interactions with a consumer are essentially a walled garden, meaning that only the individual business is monitoring and mining the consumer’s interactions with the business based on the consumer’s consent. Neither the mobile phone carrier nor the SMS provider is monitoring the individual consumer’s interactions across brands via SMS.

There is no single SMS communications provider that can tie into the consumer’s demographic data and monitor all their business communications on the scale that Facebook could easily do with their chat bot ecosystem.

How can businesses protect their brand reputation?

While brand name businesses will likely experiment with Facebook Messenger chat bots for marketing and commerce use cases because they could potentially reach 900 million Facebook Messenger consumer users, they should be conscious of the risk they are taking due to the immaturity of this new automated customer experience on the horizon. Likewise, consumers should be wary of what Facebook will do or what data retention and privacy protections will be put into place given the nascence of using Facebook Messenger for A2P messaging.

The SMS ecosystem is a much safer choice for businesses to explore since A2P messaging is a more trustworthy, mature, and accepted communications channel to start with the adoption of chat bots. As an example, text banking in the US has been widely available in the market for many years. Therefore, adopting a chat bot for that use case is a natural progression for financial service providers to adopt a chat bot for that use cases in order to provide a better customer experience without risking their brand reputation associated with using Facebook Messenger or exposing private data.  If your business wants to learn more about OpenMarket’s SMS chat bot and self-service messaging solutions, feel free to contact us.

See all blogs

Related Content