Trends for 2019: NLP boosts the power of chatbots and voice
We’re likely to see natural language processing (NLP) reach a level of advancement in 2019 that finally makes more complex applications look both possible and appealing.
Marketing Week first predicted two years ago that “message commerce” (as we called it) would be a trend for the future, with consumers being able to make purchases, manage bookings, ask questions, get recommendations and more from within messenger apps.
Take-up has been slow – perhaps because marketers blew too much budget too early by setting up chatbots with no clear purpose. There was a rush to develop a ‘messaging strategy’ rather than a marketing strategy. (Unfortunately our name also didn’t catch on.)
But if some brands were burned by their messaging experience and didn’t go back, they will have missed the progress made by those such as KLM, Lego and Victoria Beckham that have seen solid business results from carefully planned use of Facebook Messenger.
It remains the case that most brands entering the messaging space for the first time will have fairly simple needs. In customer service, for example, 90% of queries are covered by the top 10 questions, according to Facebook, meaning that brands can achieve a lot with fairly simple automated messages.
But in 2019 we’re likely to see natural language processing (NLP) reach a level of advancement that finally makes more complex applications look both possible and appealing. This will benefit not just chatbots but also that other hugely promising but so far disappointing technology – voice assistants.
Right now, NLP still struggles to detect a lot of nuances in meaning, whether due to dialectal differences, lack of context or spelling and grammar errors. But the field is developing quickly.
In November, MIT announced its scientists had created technology that learns language more like a baby than a machine – taking in visual and situational information as well as linguistic – which could help artificial intelligence understand and approximate human communication more closely. Also in November, a global conference concluded machine translation is now “very close to the performance of human translators”.
As a result, in a year’s time we could see brands taking the first steps towards creating chatbots and voice assistants that are more capable of understanding and responding accurately to complex queries and conversational requests. The upshot, eventually, should be a much improved customer experience and the ability for brands to make the path to conversion much smoother – whatever their business objectives.
To read more of Marketing Week’s trends for 2019, sponsored by Salesforce, head here.