The skills of the artificially intelligent virtual assistant

By , 8 December 2016 at 12:12
The skills of the artificially intelligent virtual assistant
Digital Life

The skills of the artificially intelligent virtual assistant

By , 8 December 2016 at 12:12

Virtual assistants or VAs have been rocking our distributed world for awhile now. They allow us to hire part-time help that tackles the tasks we don’t want to or don’t have the skills or time for, while they get the flexibility of multiple clients working from everywhere. But can those professional skills be automated?

Recently we wrote about how chatbots are replacing customer service. McKinsey says 29 percent of support could be roboticized now, while Gartner predicts that number to jump to 85 percent by the end of the decade.

Will this trend continue to other repetitive actions? Can the work of a virtual assistant be automated? As artificial intelligence continues to exponentially advance, the truly virtual, virtual assistant — also known as an intelligent virtual assistant — becomes a virtual reality. Today we talk about where it is, who’s leading the space, and where it’s headed.

Can a virtual VA really replace our real-life lifesavers?

Now, where we are isn’t where we need to be. Siri in particular — Apple’s pithy natural language processing (NLP) built into the recent iOS mobile devices — isn’t so smart. She’s supposedly always listening behind the scenes (unless you opt out) but, besides setting alarms and timers, OK Google way outshines her in terms of accurate responses. Neither go to any sort of intellectual strains.

What does the intuitive intelligent VA need to be able to do? As machine learning and NLP pushes for smarter artificial intelligence, certain pieces will have to come together:

  • Context — A good VA will have the memory of previous conversations or at least this current conversation to help. When you ask Siri “What’s the weather next weekend in Barcelona?” she will answer. But then if you reply “OK, look for cheap flights” she won’t know you mean to Barcelona for next weekend.
  • Asynchronous — Like a human virtual assistant, it will understand that you are busy and will never expect commands to be continuous or contextual. A conversational chatbot will understand the context of the conversation, while if you ask it the next day to “Search for flights for my trip next weekend” the next level of intelligent chatbot will understand you meant Barcelona.
  • Integrated — Like the chatbots integrated into our systems we’re daily already using like Slack, we don’t expect to have to open up or even call on the system. We expect — like an assistant seated next to us in the physical office — them to be for us wherever we need them, which means inside whatever system or device we already are.  

What intelligent virtual assistants are out there already?

Siri, Google and Amazon Alexa are the known generalists who are just there to answer almost every question and to point you to the right search results. Beyond the categories above which go from the simpler knowledgebase responses to the conversational and intelligent, there are specialist VAs that perform one task very well. And if they can be integrated into these generalist systems or perhaps a series of integrated business VAs, then these specialist bots can be brought together to become very powerful.

Here are some of the specialist intelligent virtual assistants working for you now:

  • Amy: Simply email amy-at-x-dot-ai to schedule a meeting. Basically she automates the back and forth of timezones and schedules to find a common time between you and your recipients to suggest the perfect time for you all to meet and then adds it to your calendars.
  • RoboCoke: Are you in Hungary and want to go out? Try Coke’s virtual assistant who will get to know you and then find the perfect concert or party for you.
  • Shae: We all have the bad habit of WebMD-ing when we have some undiagnosed symptoms. Wouldn’t it be healthier (both physically and mentally) if we had someone looking out for us? Shae uses some 500 algorithms from 10,000 data points to provide very specific help for users, telling you what to eat, what exercise you should do and a whole health plan proactively looking after you.

 

Have you hired your own intelligent virtual assistant? Tell us who you trust to do what @TefDigital and @JKRiggins!

previous article

Lucy, the robot that drinks in the sun (but contaminates a tonne less)

Lucy, the robot that drinks in the sun (but contaminates a tonne less)
next article

Flamory: a second memory for saving useful information

Flamory: a second memory for saving useful information