A.I. and speech advances bring virtual assistants to work
- 20 March, 2018 22:43
ORLANDO, Fla. – Speech recognition technologies have improved so much in recent years – thanks to cloud computing and advances in machine learning – that the virtual assistants created by Amazon, Google and Apple have quickly become popular with consumers.
So it should come as little surprise that the underlying natural language technology is making inroads at work, too.
“I would say that it [enterprise adoption] is in early stages now, but there are certainly basic capabilities here today,” Jon Arnold, of J Arnold & Associates, said at the Enterprise Connect conference last week.
The main uses for speech recognition in the office will, at least at first, revolve around improving employee productivity and automating workflows.
Thanks to advances in artificial intelligence (A.I.) techniques, the accuracy of speech recognition systems has improved significantly, with Google and others passing the 95% accuracy mark.
“It has been improving for a long time, but what matters is that it has passed the threshold at which it is on par with human speech recognition,” said Arnold, during a keynote presentation at Enterprise Connect. “So the machines have caught up big time.”
For businesses, there are four main ways of accessing speech technologies, said Arnold.
- Speech-to-text can be used for purposes such as email dictation and becomes more accurate the more it is used.
- Text-to-speech has benefits in mobile settings, such as the ability to create personal podcasts to review work documents or notes during a commute.
- Speech recognition also allows for conversational interactions via virtual assistants, issuing commands such as finding documents or creating calendar entries.
- And finally, speech analytics can allow for “sentiment analysis” in the enterprise, which can be useful in interview or training situations.
Four milestones in the rise of speech technology
Arnold pointed to four key technological milestones that have helped popularize speech technologies and push them forward. The first was the launch a decade ago of Google Voice Search, which used speech recognition to speed up searches on mobile devices.
Then in 2011, Apple launched Siri, which allowed for more conversational interactions and pushed speech technology toward greater mainstream awareness. The same year, IBM Watson gained a huge amount of attention for its TV appearance on game show Jeopardy!
In 2015 Amazon launched Echo and its Alexa virtual assistant, a sleeper hit for home use that has proved to be wildly popular.
And more recently, Amazon set its sights on the workplace with the unveiling of Alexa for Business. It is not the only company to aim virtual assistants specifically at business: Cisco’s Spark Assistant, Microsoft’s Cortana, Voicera’s Eva, Nuance Dragon and IBM Watson Assistant also now have business uses.
A disruptive technology on the move
During a keynote presentation at Enterprise Connect, Dave Ibitski, Amazon chief evangelist for Echo and Alexa, said that voice interfaces would be the “next major disruption in computing,” with clear applications with business settings.
“It is that idea of ambient computing, the idea that at any time I could just say ‘Alexa start my meeting,’ ‘Alexa how are my sales figures?’ or ‘Alexa I forgot to shut off the projector in the conference room please shut that off for me.’
“It is very natural, it is very spontaneous,” Ibitski said. “And all of this is happening because of the advancements we have seen in what is called NLU, or natural language understanding. And that is the difference – it is understanding context.”
Collin Davis, general manager of Alexa for Business, said virtual assistants are already helping employees get work done.
“What we are finding is a really interesting shift is happening, where voice is offering up almost another dimension of multi-tasking, where workers sitting at their desk can use Alexa almost as a vocal multi-taskerto be able to get information quickly without losing focus,” Davis said.
“You could be working on a report and you need to know how many deals closed last quarter without having to reach into your pocket or find an app or switch websites – you just get the information that you need.”
There are now more than 30,000 Alexa skills, and Davis said that a “growing community” of skill developers are focused on building skills for the workplace, and software vendors are voice-enabling existing applications.
The decisions companies face
Businesses looking at speech recognition systems for the workplace have several questions to consider before moving ahead, said Arnold. The first is figuring out which company to partner with. The major cloud providers - Amazon, Google, Microsoft and IBM - all provide speech technologies, but each has its own strengths.
And identifying a strong business case for investment is key, he said, spelling out some of the questions companies should consider.
“What balance should you strike? Who needs to be driving that decision? What business problems are you trying to address? Are you doing a tech refresh or are you really thinking about digital transformation, the bigger picture and what A.I. brings?”
Even with recent technology advances, Arnold noted that the technology is not perfect. But it doesn’t need to be for businesses to gain value.
“Speech recogntion doesn't need to get to 100% accuracy, it is good enough now," Arnold said. "So don't think we still have a few years to go until it gets really good enough. It is not 100% and it never will be, and it doesn't need to be.
“So don't aim for perfection, aim for understanding the technology and what it can do.”