At Google’s annual developer conference I/O 2018, the audience watched in amazement as CEO Sundar Pichai played out the recording of a real conversation between Google Duplex, the company’s personal assistant and a hair salon employee.
Instructed by the user, the assistant had placed a call to the salon to book an appointment. What was astonishing was not just the Google assistant could call the salon and mimic a woman’s voice modulation; it could grasp the subtleties of human speech, even uttering ‘mmm, hmm’ while the employee put the call on hold as she checked for a suitable time slot.
With the premise that 60% of small businesses in the US do not have an online presence, Pichai alluded to the conversation as an example of the possibilities of expanding the reach of technology with AI.
Google Duplex, as Pichai revealed, took years in the making. But how did the technology eventually become possible in this decade?
Driven by the resurgent progress in AI and the simultaneous evolution of computer hardware, neural networks have expanded to support deep learning capabilities, where algorithms (developed as far back as the 1960’s) which were once hampered by the inadequate advancements in hardware, can now be run easily. Software application has become more democratized because of the mainstreaming of cloud computing and the enhanced ability of GPOs to handle bigger tasks. The convergence of these multiple hardware and software evolutions has brought in a new paradigm of Human-Computer Interaction, including Natural Language Processing (NLP).
NLP – the ability of a machine to understand human language is now on the path to becoming what it was meant to be – more ‘natural’, where machines are beginning to ‘comprehend’ (process) ‘think’ (execute) and ‘talk’ (respond) like humans. From toddlers to grandparents, the entire demographic arc can soon delight in getting work done from a phone or a virtual assistant home device without knowing any programming languages.
As tech behemoths like Google, Amazon and Apple take the lead in unraveling the potential of NLP, today’s Duplex, Alexa, Siri and the like are more intuitive, making technology more personal than before. Bolstered by the patronage of their bigger brethren in the field, smaller tech companies are now following suit by investing in AI driven technologies like NLP. While big companies have access to APIs and various open source components, the real differentiator is a company’s access to annotated data that allows a bot to be trained with domain-specific knowledge for a particular industry. Several small companies are specializing in specific goals, like for example, developing a chatbots for the insurance or utility sector.
Unlike a few years ago, where assistants interpreted verbal cues as data inputs and not language, developers are working to create virtual helpers that interpret the idiosyncrasies of the semantics, syntax and sentiment of human thought expressed in speech, enabling machines to deliver real-time, appropriate responses to queries. We are entering a phase where technology rarely, if ever, disappoints.
Attuned to instant gratification, consumers are also contributing to the push effect in technology disruption. As this McKinsey report analyses, consumer loyalty is slowly becoming an obsolete phenomenon; spoil for choice, today’s consumers leave brands that do not satisfy their shopping experience and move on to those that do.
Curating technology to become more customer-centric for delivering better customer experience (CX), has therefore emerged as a battleground for consumer brands. Even those businesses (think of the hair salon in the Google Duplex case) that are out of the ambit of technology reach are being impacted by growing sophistication (ergo simplicity) of the human-computer relationship.
Where does this leave the enterprise?
With the current developments, NLP’s potential to humanize a brand and make it more intuitive is irrefutable, and the business implications of NLP have become increasingly significant. According to a recent report by Orbis Research, in the enterprise setting, NLP is gradually developing ‘ubiquitously’ especially in the legal, media, healthcare, automotive, retail and education sectors. Chatbots and virtual assistants can help with a variety of critical functions like customer service, business intelligence, analytics and compliance monitoring.
While at a nascent stage of adoption at the enterprise level, the tech industry is buoyant about the possibilities of using NLP for enterprise transformation. Investing in NLP can bring about the benefits in an enterprise setting in three main domains – new service registration, technical customer support and internal data access needs:
Customer facing benefits:
- An NLP platform can function has an automated first point of contact for a brand that needs a lot of customer interaction or one that sells identical products. A bot or assistant can be programmed with domain specific knowledge that enables accurate intent classification of the query and pre-built, intuitive conversational flows to resolve incidents on the lower end of the escalation spectrum. It can also help customers choose between similar products. In case of non-resolution, the query can be escalated further.
- An NLP based marketing strategy offers competitive advantage for business that rely on online channels for growth. In the era of ‘smart search’, an NLP application that can run on social media apps and virtual assistants can provide invaluable insights on consumer behavior and sentiment to help the brand target the right set of potential customers.
Internal organizational benefits:
- Well-programmed NLP platforms can deliver critical business content ‘on the go’ to employees in a safe and secure manner. Driven by cognitive computing, the cross platform deployment of NLP application on the web, phone, virtual assistants, chatbots, robots, the cloud and other proprietary applications (including legacy systems) will ensure that the ease with which the right insights reach the stakeholders in the form of real-time analytics – charts and graphs that analyze user data traffic.
This adaptability is another differentiator that will decide the acceptance of a technology company’s NLP solution.
For instance, when an enterprise adopts bots to resolve queries about Software Developer’s Kit (SDKs), and this information is available across platforms 24/7, the developer’s productivity and time saved on a particular task significantly increases.
- By automating several mundane tasks, NLP increases the organizational efficiency of an enterprise and allows opportunities for lower level employees to upskill or reskill for higher order workflows.
With manifold increase in research in the area, NLP based chatbots are on the rise. In the enterprise setting, apart from enhancing user experience, SaaS based conversational bots provide tangible benefits like reduction in operational costs, increased productivity and customer satisfaction and overall improvement in brand perception. While we are not yet at a stage where a bot can analyze a quarterly sales spreadsheet and call out the profit numbers for a particular month rather than merely pull up the document at a voice command, we will eventually get there.