Where Are Our Jetpacks…..and Virtual Assistants? Alexa and Siri Are Just the Beginning.
To achieve mainstream adoption, new technology must cross a wide chasm. Gartner, whose Magic Quadrant is the equivalent of a Robert Parker 95-point wine rating, maps this journey in its Emerging Technology Hype Cycle.
A single chart with two decades of these prognostications provides a composite timeline of the technologies that now make up 4IR, plus many more that fizzled. It illustrates, graphically, how perilous is the run from the Peak of Inflated Expectations to mass market— many technologies vanish on the slalom down. Others that survive end up buried in the Trough of Disillusionment.
And then there’s the curious case of Virtual Assistants, which debuted in the Trough of Disillusionment — a result of Apple’s penchant for surprise, plus Siri’s early predilection for comical errors. Since then, they’ve backed up the slope, oscillated around its apex, and birthed some related buzzwords — yet, by Gartner’s estimation, they’ve regressed. What’s up with that? Nearly a decade after Apple launched Siri, why don’t we all have a personal virtual assistant?
When you think about it, this is stupendously strange. The technical underpinnings of virtual assistants — speech recognition, natural language processing, and robotic process automation — have exponentially advanced since Siri debuted. While plenty of technology never realizes its early promise, the key components of personal assistants are actually more capable than first envisioned. Plus, as digital services subsume and integrate more of daily life, the already-long list of use-cases keeps growing. Siri, along with Alexa and Google Assistant, now controls not just phones, but entire homes via smart speakers. Going from beta to ubiquitous in just five years, smart speakers, along with smart watches, are straight of out of Sci-Fi, and enable long-envisioned conversant computing.
Socrates taught that wisdom begins with a definition of terms, which is especially important with Virtual Assistants and related technology, because there’s so much related technology. Siri, Alexa and the like are Digital Assistants, which combine speech recognition with basic Robotic Process Automation (RPA). RPA automates the sort of tasky things a secretary might do, like email a spreadsheet. Speech Recognition is just that, while NLP — Natural Language Processing or Neuro-Linguistic Programming — allows computers to understand, and speak, sentences. Chatbots converse via messaging apps like Messenger, Kik or SMS, sometimes with NLP, sometimes rules-based with multiple choice Q&A. Virtual Assistants can combine all the above into a Conversational Agent designed to assist a particular human.
Conversation is the key to morphing digital assistants into full-on virtual assistants. You can talk to Siri and Alexa, but you can’t converse with them, because, while they execute commands, they don’t maintain context. Context is what enables dialog, and dialog enables the power to perform more advanced tasks, like buying plane tickets or making a dinner reservation.
Actually, the platforms underlying DAs, eg Amazon Lex and Goole Dialogflow, can maintain context. With the addition of context-capability a couple years ago, Garner (yep, them again) figured we’d soon be talking to bots more than our spouses.
“By 2020, the average person will have more conversations with bots than with their spouse.” Gartner ¹
So, why aren’t we talking more with computers? Partially, it’s (lack of) trust. Partially, it’s the same reason we’re not talking more with humans: we don’t have to.
Once upon a time — actually, for the entirety of humanity’s existence up to 20 years ago — dialog was the foundation of our existence. Life, commerce especially, required conversation. Even after King Kullen launched the self-service era, along with exchanging dollars, buyers and sellers also traded polite chit chat. But on the web, words are an optional side dish, and, mostly, we decline the combo meal.
At the office, however, things are different. If something lowers cost, or improves efficiency, it will find its place — regardless of whether or not employees love it (see Google Glass, which is back in Enterprise Edition). While the consumer Augmented Reality market slowly unfolds, workplace adoption is already well underway. So it is also with Virtual Assistants.
An obvious Enterprise use is call centers. Unlike Interactive Voice Response and legacy speech recognition, callers are fine with AI-powered Virtual Assistants — call abandonment rates are typically far lower. Virtual Assistants enable a new call handling paradigm: the “cobotic” arrangement, where AI and human agents blend dynamically, without fixed service tiers, and with flexible depth and breadth.
With AI assistants offloading calls from live agents, unsurprisingly, costs drop. What might be surprising is that results improve. In one case, when 250,000 TV game show viewers called a toll-free hotline searching for a fortune in fabulous prizes, Vocinity synthetic agents had a higher percentage of meaningful conversations than human agents. In another, analyzing 500,000 calls about residential solar power, the combination of synthetic agents and human reps delivered 500% better productivity.
Also, AI-agents can be mind-readers. Using algorithmic ESP similar to Facebook and Google ad targeting, AI agents can quickly size up their conversation partners — in as few as 100 words — to produce a “Big Five” personality profile like the one below. Other models can also discern entrepreneurial tendency (likely to start a business), environmental leaning, inclination to volunteer, entertainment preferences (for example, reading versus movies), and many other traits with commercial relevance.
If you have any doubt about the value of this sort of information, consider that a psychological profile like this one enables Facebook to collect $112 per year on the average North American monthly active user. That’s more money — without charging users a penny — than Netflix makes charging them $12.99 per month.
Let that sink in…….
Moving on, AI agents can be sneaky. Since Google made headlines by fooling unsuspecting callers when unveiling Duplex, they’ve quietly launched it as part of Reserve with Google. You may have already spoken to Duplex without knowing it. Vocinity CCA’s speak to thousands of people daily — by listening to recordings, it’s clear most have no idea they’re talking to a computer. I’m not exactly sure why this is a problem, but in case it is, California’s Bolstering Online Transparency Act took effect lats month, requiring that online sales and election canvassing bots disclose they’re not human.
So, unlike jetpacks, Virtual Assistants are here. But, like a good admin, they don’t steal the spotlight, but work behind the scenes to make their bosses life easier.
Content Label: This post is composed of 100% human-written content.
