Speech Technologies – Trends, Applications, and Their Dark Side
At Enterprise Connect 2020, Jon Arnold (Principal, J. Arnold and Associates) outlined upcoming trends in speech technologies. These technologies include a mix of AI, automatic speech recognition, machine learning, and natural language generation. Use cases are not necessarily person-to-machine and vice-versa but are increasingly machine-to-machine through the internet of things.
Present speech technologies, Arnold states, mostly derive their utility from collecting data about specific measures. Given that speech technologies are mostly found in contact center solutions, examples of these measures are call recording and length of call. However, as speech technologies reach their capacity for conversational AI beyond interactive voice response (IVR), other uses will include automation and intelligent communication. These have applications far beyond the contact center. Such applications might include more sophisticated voice assistants or accurate translation technology.
Key drivers in this space are often found in the consumer market, with enterprises following suit from employee demand – Amazon Echo led to Amazon for Business, for instance. With the surge in remote work due to COVID-19, we can expect that the proliferation of advanced speech technologies in the workplace will be accelerated. While AI-driven speech is a new realm for enterprises, the growth of smart devices having ambient interfaces will support more speech-centric workflows and authentications. This latter application could be used for improved biometric authentication, including one’s ID for meetings, signoffs, building access, and fraud detection.
However, these benefits may come with some costs, especially with regards to privacy and security. If speech technologies are gathering data through conversation, questions must be raised about how this data is used. Will information gathered through conversations be deleted, shared, or sold? Will smart speakers always be “awake,” collecting information on everything we say and do? How will this apply for keeping a separation between personal and workspaces, especially given many people are working from home? Laws should not be playing catch-up, but general data protection regulations for this type of data are not yet universal.
Further concerns arise with deepfakes. As AI-driven speech quality becomes more sophisticated, it will become increasingly difficult to tell apart real human speech from machine-generated speech. With advances in video editing for manipulating a person’s face and body movements, the combination of AI-driven speech with video could create serious problems. Should a world leader be a victim of a deepfake, what assurances can we have that such deepfakes can be differentiated from reality before dangerous consequences result? These questions will increasingly be at the heart of legal and political disputes.
For further information, watch the session Speech Technologies: Innovations and Use Cases (available online until December 31, 2020 at Enterprise Connect 2020).
Our Take
Despite the potential gains for AI-driven speech technologies, the present reality is that most organizations aren’t currently in a position to leverage the benefits of AI or even know how these benefits can be achieved. It can be hard to tie the rapidly changing and growing field of AI to your industry and organization, as well as determine if these technologies are worth the time and investment.
Source: SoftwareReviews Virtual Assistants & Chatbots Data Quadrant. Accessed September 3, 2020.
A good place to start for analyzing the market and discovering what capabilities are available is SoftwareReviews’ data quadrant for virtual assistants and chatbots (image above). While these solutions don’t encompass the entirety of speech technology, this marketspace is becoming more established. To prepare potential use cases and a strategy for these technologies in your organization, use Info-Tech’s Prepare for Cognitive Service Management resource. It may be that your organization can reap some rewards from being an early adopter.