Astounding leaps forward have been made in two key technologies in the last couple years: voice recognition and artificial intelligence. Machine learning has fueled tremendous advances that are now resulting in these two technologies being used very broadly across many types of devices and in many situations.
Virtually everyone is becoming used to conversing with Siri, Alexa, and others – backed by powerful voice recognition engines to convert speech to text and generate text or speech output. At the same time, AI is at the heart of major breakthroughs in driverless vehicles, robotics, IoT devices, and other emerging technologies.
The promise of these technologies to make our world safer, improve our lives, and address chronic societal problems is no longer the province of science fiction writers. It is within our grasp. But, in the midst of all the excitement, there is a dichotomy that must be resolved: the gap between the current performance of voice and AI technologies and the levels required to achieve the benefits of driverless vehicles and other tech.
Practically everyone can relate to frustrating incidents in using voice and AI-based systems for rudimentary tasks. For starters, it is true that voice technologies have made great strides recently and the progress seems to be accelerating. However, my personal experience with these systems often produces results that are comical, sometimes offensive, and many times just downright misleading. The Bluetooth in my car cannot even correctly interpret names when I want it to dial individuals with simple, single-syllable first and last names. And we have all seen crazy text conversations posted on Facebook.
On the AI front, the use of virtual assistants is becoming more common. My recent experience with one was a trigger for writing this blog. Communicating with a virtual assistant to address a technical email issue, I was sent into endless loops and misunderstandings, always to be followed with, “Did I answer your question?” My pleas to PLEASE LET ME TALK TO A LIVE HUMAN were answered with nonsense responses, or “Please rate your satisfaction with the answer.” You can guess my response.
Similarly, if you have ever used a virtual assistant to try and schedule meetings, you have most likely come away completely flummoxed and ended up emailing or calling your colleague directly to schedule it. These types of interactions begin to give you a deeper appreciation for the nuances of human communication.
My point in describing the examples of tech shortcomings is that we are headed into a future where we will rely extensively on them in many parts of our lives. The AI behind decisions that affect your life will not all occur when you are barreling down the highway at 65 miles per hour. But much of our transportation, healthcare, entertainment, education, and many aspects of our daily lives will be highly dependent on the recommendations and decisions made by AI-based systems. And we will control the world around us largely via voice commands.
Both voice and AI technologies are improving rapidly and will surely be much improved over time. But the question becomes – how can we be confident that the AI in a driverless vehicle will make the right life and death decisions in milliseconds when the relatively simple interactions we experience today are botched so badly? Will the robot companion of my elderly father make a fatal mistake on medicine dosage? The accuracy and success of these technologies when used in the future for driverless vehicles, smart homes, and the world at large is essential. And the implications for the insurance industry are significant.
The progress of these technologies bears close monitoring so that the regulatory environment, availability, and usage of new capabilities in the connected world do not actually make the world more risky and less safe.
For now, we must live with (or suffer through) the current state of the tech. Just for fun, I actually asked Siri the title question. It was interpreted as “What’s your opinion of dry rose vehicles?” Need I say more?
This blog entry has been reprinted with permission from SMA.
Readers are encouraged to respond using the “Add Your Comments” box below.
The opinions posted in this blog do not necessarily reflect those of Digital Insurance or SourceMedia.