Will Your Chatbot Be A Responsible Bot?



We’re already well past the first generation of chatbots. While the early scripted examples were unlikely to cause offence, unless they were gamed or hacked, as new bots get smarter, the businesses that deploy them need to be careful over responsibility, governance and user safety.

Most businesses would apply the same levels of training, rigor, and service rules to their chatbots as they would staff on the shop floor, customer service agents and office workers. Bots are there to be helpful, kind and considerate, and must not break any laws.

However, as bots get smarter, and AI moves their scope beyond what the creators may have intended, there’s the risk that a bot could get out of hand. Microsoft recently published a set of guidelines about creating a responsible bot (PDF).

These are useful for developers and businesses looking to roll out bots in critical areas such as health, banking, travel and government, but apply just as well to customer service and traditional bots.

With great bot growth comes great responsibility

Bot growth continues to rise dramatically, and chatbots and virtual assistants are playing a larger role in people’s lives. This moves them from simple helpers to trusted guardians who can be relied on for vital assistance. Imagine an assistant that reminds vulnerable or older people to take their medicine. What happens if a cloud outage or other problems fails to provide the reminder – who is responsible?

The guidelines are designed to help with the responsible development of conversational artificial intelligence. They are based on experiences gained through talking to third-party developers, business customers and partners.

Note, Microsoft’s guidelines are just guidelines. Any business engaging in conversational AI will need to check with all their legal counsel or trade bodies in their industry that the bot complies with standard regulations, and look for any specific bot-related rulings.

While that might sound complicated, in most cases bot builders have nothing to fear, as long as they take into account the guidelines, their own common sense and the needs of the business and the customer. Still, if the bot has some consequence for the user, extra care must be taken to ensure the bot provides the correct information, makes users aware of any consequences of that advice, seeks consent where needed and so on.

As the advice explains, “In these situations, we’ve learned to pause and ask: Is this a situation in which it’s important to make sure there are people involved to provide judgment, expertise and empathy?”

The key guidelines are:

  • Before beginning design work, carefully specify how your bot will benefit the user or the
    entity deploying the bot.
  • Be transparent about the fact that you use bots as part of your product or service
  • Ensure a seamless hand-off to a human where the human-bot exchange leads to interactions that exceed the bot’s competence.
  • Design your bot so that it respects relevant cultural norms and guards against misuse.

And there’s plenty more and a lot of detail in the document to ensure that the business is covered and the end user protected from the bot failing to deliver on the above.

Any business developing a new bot or refreshing legacy services needs to take the advice on board and ensure that the next generation of bots are upstanding citizens of the digital age. Ultimately, consider if it was one of your family or friends using the bot, would they be satisfied with the service and accept the advice or information it offers?

Conversational AI, chatbots and virtual assistants still have a long way to go. But, while creators might have crafted their early versions as side-projects or business experiments, now – as usage booms and bots take on more responsibilities – all parties from designers to coders and analysts need to up their game to make sure the next bot is very fit for purpose.

Businesses looking to launch their first bot will find many services out there offering development or building tools, SnatchBot offers a do-it-yourself or expert-help pro-level service, that allows businesses of any size to develop bots for Facebook Messenger, apps, communications platforms and other touchpoints, with easy-to-use cloud analytics. You will find plenty of advice on building bots inThe Chatbot Magazine.