Chatbots and AI- Part 1: not married yet, but soon

More and more, we see chatbots being rolled out by banks, airlines, ecommerce platforms… anyone who has contact with customers.

And more and more, we are not sure when we are talking to a real human being, or an ‘intelligent’ machine, when we click on “live chat” on any web or mobile platform.

Why not? Customer service has always been a very (and increasingly) costly part of operations, for which companies are keen to optimise (read: cost cutting). Huge industries were built in places like the Philippines, India, and Yogyakarta in Indonesia.

Nonetheless, recruiting, training and managing a large and remote workforce is still costly in both monetary and management terms.

And of course, the advent of AI and headlines made by make corporate decision makers to seriously look at the possibility of automating customer service.

Benefits of chat bots

Lots of companies are offering such type of chat bot services to reduce operational cost or to use it as a novelty in their customer offering.

Compared to traditional automated customer service methods such as FAQs, form filling or phone voice menu – chat bots are definitely a more engaging alternative. They are:

• More natural for customers or other intended users;
• More versatile, allowing more branches on the decision tree;

And compared to voice calls or human chats, they are:

• More flexible in terms of engagement time (e.g., you do not have to get a time where both the agent and customer can take the phone).
• Scalable (e.g. no more hassle to arrange extra shifts during peak hours)
• Allowing automatic data capturing and analysis, thus allowing quick improvements.

Not intelligent (yet)

Most chat bots are good at responding to simple, often transactional, questions or instructions.

In fact, a lot of the customer service requests are simple – such as checking credit card balance, or enquiring about the status of parcel delivery.

And if you think about the logic: “How much does an iPhone8 cost” or “Order me an Uber” are rather straightforward.

They do not require complicated technology, let alone AI, to deal with. What is needed, however, is that data is captured and stored properly.

To date, most chat bots have NO INTELLIGENCE built behind it.

Natural Language Processing (NLP) is not there yet. Ask Siri anything more complicated, and you will know what I mean.

But AI is on the horizon

Why? Well, maybe voice recognition & facial recognition should offer clue.

Both were for a long time unreliable and unusable; but both had some serious breakthrough over the past few years. To quote an industry expert, “the most significant change is from 95% reliability to 99% reliability.” “With 95% you still can’t trust the system to perform serious tasks, but with 99% you can.”

Now voice recognition has enabled personal assistants such as Siri, Alexa and Google Home to flourish. And facial recognition is (pretty accurately) used in many scenarios, from Facebook tagging to mobile banking authentication.

Investors and practitioners are betting that, maybe, intelligent chat bots will reach that point soon.