Though they are proliferating, that doesn’t mean that chatbots have reached their peak. They’re far from perfect, in fact, with the technology constantly evolving to improve on learning and interactions.
Econsultancy’s Marketer’s Guide to Chatbots delves into the subject of chatbots in much more depth, but in the meantime, let’s sum up some of the key challenges that the chatbot space still faces.
Finally, it seems many marketers are guilty of jumping on the chatbot bandwagon, without real consideration as to whether it will add proper value for users. One category that springs to mind is ‘gift finder’ chatbots, where the idea is to help consumers narrow down their search to find the perfect gift.
In theory, it sounds cool, but in reality the result is often limited and rather boxed-in, with chatbots asking a series of basic (and even irrelevant) questions. There are likely to be some good examples, but more often than not, it’s used as a quick marketing win, meaning users are far better off browsing comprehensive retail gift guides elsewhere.
In contrast, chatbots that have a clear area of focus and do it well tend to succeed. Skyscanner, for example, which delivers straightforward flight information in an easy to understand manner. Or Whole Foods, which offers users recipes based off of single ingredients.
Focused, valuable, and easy to use – these examples are where marketers should look for inspiration.
A lot of chatbots are designed to fill a customer service gap. In other words, to allow users to ask for help online, and take away strain on resources elsewhere. However, success here relies on effective integration with existing systems.
If the user interacts with a chatbot that exists in silo, there will be no way of accessing that user’s data – even if they’re a long-term or loyal customer. In turn, there’ll be no record of that chatbot interaction (and the end result) elsewhere.
In contrast to this, the best chatbots seamlessly take users through the customer service journey, allowing them to be passed to other touchpoints (such as telephone or email) if necessary, and drawing on information held there.
Chatbots are only as good as the data they are built on. So, if the data is poorly sourced or biased, the results can be disastrous.
Remember Taybot – the Microsoft chatbot that learnt from human behaviour on Twitter? We all know how ghastly people can behave on Twitter – it is the lurking place of endless trolls. Naturally then, the bot ended up being programmed to spout racist and hateful opinion, and was shut down within hours.
To prevent this, many tech companies have now put in place an internal code of ethics specifically created for their chatbots. This enables companies to actively teach morals and ethics to their AI, ensuring that biased data or opinion does not sway it from its own rules.
With artificial intelligence growing at a rapid rate, UNESCO announced the importance of a global code of ethics for artificial intelligence earlier this year, also suggesting that it is in the process of setting up an international regulatory instrument to help ensure the responsible development of AI.
For more on this topic, subscribers can access our Marketers Guide to Chatbots