The blogosphere is awash with views on Apple’s Siri. Some of it effusive with its praise, some reveling in its supposed shortcomings — see this post from the New York Times Bits blog for example. One thing for sure, you could be forgiven for thinking that Apple’s natural language technology that powers Siri was the only game in town.

It is not, of course. Artificial Solutions — along with other specialist vendors – has been working much longer in the area of natural language interaction and promoting its myriad applications both in fields such as contact centers and web-based virtual assistants and, more recently, in the fast-growing consumer electronics space.

But the recent arrival of Siri has opened the eyes — or rather mouths — of a much larger mass market to the wonders of natural language interaction and, for that, the whole industry should be grateful.

I used the word “wonders” deliberately in the last paragraph. It is clear that for many consumers, natural language technology is indeed something worthy of marvel and wonder.

The ability of a smartphone to first understand and then act on its owner’s wishes expressed in everyday language is indeed impressive. Just think how many ways you could ask about the weather; ‘what’s the forecast’, ‘what will the weather be like’ ‘will I need my hat/coat/sunglasses’ and a multitude of other combinations; and NLI needs to understand each and every one of them, act on it and deliver the right answer – all in milliseconds. So it’s an incredibly clever and sophisticated technology – but it’s not magical.

Hundreds of man-years of work go into refining the algorithms at the heart of technology such as Artificial Solutions’ Teneo Mobile or Apple’s Siri. However, it is just computer code and like every other real-world computer program it’s designed to meet certain objectives and is tailored for specific applications.

So for example, try asking a contact-centre IVR system (programmed for a very tight set of requirements) a philosophical question and you won’t get very far. NLI based technologies need to be able to understand spoken word, sentiment, meaning, context, abbreviations and so much more – they’re pretty good at doing this and are getting better by the day but they’re not perfect all of the time. Indeed, as consumer applications of natural language technology become more commonplace we will inevitably hear of more cases where the technology has made an error — take for example, the case of the ad in which Siri misidentified poison oak and pulled up a photo of poison ivy instead.

However, it’s important to put all of this into context. The technological revolution has transformed daily lives and created consumer devices that would have been unconceivable a decade or so earlier. Against this backdrop of tremendous evolution in other areas, consumers wonder why computers still have difficulty understanding a “simple English conversation”.

But human conversations are not simple to understand to a computer, which is why the Turing Test continues to draw challengers — see this earlier NLInews post.

A large part of the blame for the heightened expectations surrounding natural language interaction can be placed on human nature — we all get excited about new inventions that promise to revolutionize our daily lives in some way.

To understand the phenomenon better, let’s use GartnerGroup’s Hype Cycle, which describes the introduction of any new technology and is widely used in the IT industry to aid decision-making.

In 2011, Gartner described “natural language question answering” as a high-impact trend and so the roller-coaster ride began.

This is how the hype curve works:

• Following its introduction, every promising new technology climbs to the so-called “peak of inflated expectations” — when media buzz is greatest but the number of successful implementations is usually far-outweighed by the failures.
• Then, interest wanes as implementations fail to deliver — Apple’s Siri is clearly coming close to this turning point, which Gartner cruelly calls the “trough of disillusionment”.
• As more successful implementations emerge and the new technology becomes more widely understood, second and third-generation products appear from providers and more users embrace them — the “slope of enlightenment”.
• The final stage, called the “plateau of productivity” covers the stage when the successful technology gradually enters the mainstream.

With natural language interaction we are clearly a long way from the situation where every new automobile, home or smartphone come equipped with this technology.

But Artificial Solutions and other industry players are working hard to enlighten users about the benefits of natural language technology, so speeding its adoption and ensuring its commercial success.