Much has already been written about chatbots, Artificial Intelligence (AI), machine learning and the impact all this is going to have on organisational life, jobs and customer expectation.
In fact, we have done such a great job getting excited about possibilities we are now at the peak of inflated expectations unable to distinguish between solid functionality and fantasy. Gartner predicts the inevitable trough of disillusionment as the next chapter.
For instance, recognising words is one thing. Making sense of them is quite another.
Real-world speech recognition rates at Google improved from a word accuracy rate of just over 75% in 2013 to this year’s benchmark of 95%. When chief AI scientist Andrew Ng moved from Google to Baidu, he also moved the dial to an outstanding 99% level of word accuracy.
“In a few years everyone will be using speech recognition. It will feel natural. You’ll soon forget what it was like before you could talk to computers.” ~ Andrew Ng
But holding a conversation is more than recognising words. It’s about about navigating the ambiguities of everyday language using accumulated experience to decipher a person’s actual meaning. People learn how to do that. At some point so to will our virtual equivalents. Till then we have to rely on the craft of dialogue design to compensate when things just don’t make sense to their still embryonic abilities.
Think Big. Start Modestly
So it makes sense to tread carefully and sort fact from fiction if you want a solid start to your intelligent assistant strategy.
Be clear before you start exactly how this will improve the effectiveness of your delivery model. Certainly customers who are being exposed to the kind of low effort engagement exemplified by the likes of Amazon and Google now expect the same smartness from you.
AI has played a core role in making that happen. The mistake of course is to immediately start modelling your own ambitions against them.
Instead you need a realistic roadmap that puts you on the right path. Something that begins simple, delivers benefit, does not cause too much risk to the organisation or personal reputation, is affordable and has an immediate impact.
To quote Bob Weir, you should aim to get it ‘just exactly perfect’.
This blog post has been re-published by kind permission of Martin Hill-Wilson