How to train your A.I. dragon: 3 key takeaways for the enterprise

While artificial intelligence has seen many rises and falls in the hype cycle over the years, there’s no doubt A.I., especially in its conversational form, has now hit its stride, with products like Siri and Alexa finally finding a mainstream embrace.

However, A.I. is not just for consumers. By 2020, A.I. will be as critical to businesses and customer service as the website was 20 years ago, or the mobile app was 5 years ago.

But enterprises aren’t ready to trust their brand experience and the data that comes with it to a consumer-driven technology. As we’ve seen recently, if A.I. learns as it experiences more human interaction, it can learn to curse, issue racial slurs, or become a misogynist. Especially in a customer service situation where angry customers can present a less mannered side of society, we don’t want our A.I. bots and customer service assistants to respond in kind.

Training A.I. is like training a child or a pet. Left to run wild, bad things happen. Training A.I. technology is a matter of philosophy more than programming. A company can build its A.I. technology to simply take the input it encounters and learn from it without any rules or guidance. Or the technology can come with explicit rules that reflect society’s mores. For example, businesses want an A.I. bot or customer service assistant that can de-escalate the call from an angry customer, not simply retort with equal anger.

I like to use some rules I call LIP Service to tame potential A.I. problems.

L is for Language: The first rule is for Language, specifically natural language Interaction (NLI). NLI is a form of artificial intelligence that allows people to talk to applications and electronic devices in free-format natural language, using speech, text, touch, or gesture. If NLI is going to be effectively used across an organization, it needs to be grounded in multi-lingual, multi-platform, multi-channel, multi-modal applications that seamlessly integrate into a company’s existing infrastructure. That’s the reason simple consumer-facing scenarios like Siri and Alexa have limited viability in an enterprise environment. Moreover, these solutions could pose great risks to an organization if they cannot be easily trained to follow brand and company protocol in both friendly and tense customer interactions.

I is for Information: The second rule is about Information or data. With every customer or employee interaction, a company gets mounds of data that can help it learn and understand more about customers and its business operations. When a business hands over its A.I. technology to a third party, it loses control of the data that comes with A.I. interactions. As many have said, data is the new oil. It’s a gusher of information that is immensely valuable, both monetarily and beyond. To give it away for free is foolish. If a business can’t control the data generated from its A.I. solution, it’s better off not doing A.I. at all.

P is for Platform: Finally, the third rule is that a company must consider the platform upon which it will build its A.I. services. In a fast-moving economy like A.I., innovations will churn quickly. Much like the mobile phone market, new and improved applications and devices are introduced constantly. But also like the mobile phone, if the underlying OS and network isn’t solid, having the latest device doesn’t matter. The A.I. platform should be built for the future, so that as A.I. applications and services change, the enterprise can easily adapt without the costly burden of switching out the underlying technology.

As your company navigates the new A.I. world, it is critical to remember the rules of LIP Service as you employ the right tools to properly train your A.I. dragon:

1. L – Language, specifically natural language interaction
2. I – Information, means keeping control of your data
3. P – Platform, which must be built for the future

LIP service is a great foundation for maintaining control of your brand and customer experience while taking advantage of today’s burgeoning AI demand.