From Apple’s Siri, to Microsoft’s Cortana, to Amazon’s Alexa, talking AI assistants now willingly, and most of the time accurately, respond to our commands. 33 million voice-activated devices were installed in the U.S. by the end of 2017, according to VoiceLabs, and Gartner predicts that by 2020, 30% of our interactions with technology will be through “conversations” with smart machines.
Voice is the new interface. It’s a radical transformation of the design and development of computers, after years of relying on eyes and hands, screens and touch, for our interactions with technology. Voice adds many opportunities for interacting with computers at the time and place of our choosing, with new locations and activities including driving, exercising, cooking, and operating machinery.
The rapid progress in natural language processing (NLP) over the last five years has been behind the development of much improved software which “understands” human speech and can respond at a high level of accuracy and relevance. The same machine learning methods—deep neural networks—that spurred the progress in NLP have also contributed to a significant expansion of the range of activities the software can support.
The word error rate for voice-recognition systems dropped from 43% in 1995 to just 6.3% today, on par with humans. Other artificial intelligence (AI) domains have demonstrated similar performance acceleration in recent years. In 2016, Google improved Google Translate more than it did in the previous 10 years combined by incorporating deep neural networks into the software, reducing errors by 60%. Also last year, human parity has been reported for the first time for conversational speech recognition and an AI system watching video of a person speaking and matching text to the movement of their mouth achieved 93.4% accuracy, up from the 79.6% achieved by the previous state-of-the-art system.
The combination of improved AI and the new voice interface has led to a race to establish a foothold in the emerging smart home market and on mobile phones. As with many other tech developments in the last 15 years, we have seen a consumer-led transition to new applications, use cases, and interfaces. It’s a huge business opportunity, as the number of people worldwide using AI voice assistants is projected to increase from 504 million last year to 1.8 billion by 2021.
A potentially even larger business opportunity, however, is the market for AI voice assistants in the enterprise. That market has recently been energized by Amazon launching Alexa Voice Service Device SDK toolset, making its AI voice assistant available on more third-party devices, and by Amazon’s and Microsoft’s joint announcement that their respective AI voice assistants, Alexa and Cortana, “will begin talking to each other later this year.” Combining their respective strengths in the consumer and enterprise markets, Amazon and Microsoft will provide their customers with an AI voice assistant that will deliver both the best online shopping experience and access to work email and calendar.
There are however, specific enterprise applications that can benefit from the work of AI voice assistants. The two leading examples, so far, are customer service and IT operations automation.
Research firm Tractica estimates that more than 23,000 AI voice assistants will be deployed for customer service applications between 2016 and 2022. First and foremost, they are and will be installed where customer interactions are standardized and repetitive in nature, in places such as banks, airports, retail stores, and shopping malls. AI voice assistants can provide many benefits when they work in collaboration with human agents responding to customers online or on the phone. These benefits include:
- Reducing the workload of human agents by providing self-service option
- Increasing the speed of response to customers by managing some or all of the interaction
- Expanding self-service to new channels such as mobile apps
- Assisting in training new customer service representatives
- Generating new revenue streams through improved customer conversion and satisfaction
- Collecting and analyzing data on good and bad customer interactions
- Collecting and analyzing data on customer concerns and queries which could serve as a basis for proactive customer service.
For customers that are very comfortable with technology, AI voice assistants may provide an additional benefit as these customers would rather talk to an avatar than speak to a human agent.
Augmenting the work of customer service representatives with AI voice assistants is an example of improving the efficiency and effectiveness of the enterprise in its interactions with the outside world. But these assistants can also improve the productivity of internal work processes, such as IT operations.
As information technology increasingly drives all enterprise activities and is more and more responsible for its survival and success, the work of the IT team is becoming more complex and demanding. Integrating AI voice assistants with this work helps meet the new IT challenges in a number of ways. For example:
- At the IT help desk, AI assistants look up an issue or a question in a database and respond to user queries with the correct answer. If further help is needed, the AI assistant can create a new support ticket
- For ongoing IT operations, AI assistants help monitor applications, systems and networks as well automate repetitive tasks. The AI assistant can escalate and assign an issue to a specific member of the IT team and provide updates as needed
- When IT systems go down, the AI assistant can notify IT administrators and even attempt to restart them again
- Collecting data continuously, the AI assistants become the system of record, notifying the IT team of observed anomalies or any unauthorized changes in the IT environment.
AI voice assistants help enterprises manage better the rising complexity of their interactions with the outside world, efficiently and accurately handling requests and inquiries from existing and potential customers, suppliers, and partners. They also work alongside the enterprise teams managing internal processes and workflows, helping with monitoring, analyzing and responding in a timely manner. In the near future, we will see the work of AI voice assistants applied to a broad range of enterprise activities, performing routine tasks, making automated decisions, and augmenting the work of knowledge workers.