When good technology goes bad

September 12, 2005

It’s not surprising that I was greeted by a software-driven voice when I called a major provider of voice recognition software recently. What better way, I thought, for the company to show confidence in its technology than to put VR where its mouth is?

It's not surprising that I was greeted by a software-driven voice when I called a major provider of voice recognition software recently. What better way, I thought, for the company to show confidence in its technology than to put VR where its mouth is?

A moment after being asked: "Which person or department would you like to speak with," I shot back "marketing." My digital greeter repeated the department name perfectly, transferring me to the appropriate extension where, after several rings, I left a voice mail.

Having not quite achieved my goal, which was to get answers to my questions, I decided to call the company again. This time I asked for public relations and, after repeating the department name perfectly, my digital router correctly transferred me. Again, after the requisite number of rings, I left a voice mail.

With my curiosity piqued, I called back a third time, wondering what would happen if I worked more closely with the software, as it was the only thing apparently working at this company. I learned that I could get driving directions. I could get a list of the departments. Or I could ask for the front desk.

Doing the latter instantly made human contact. The operator, having picked up on the first ring, dialed the appropriate extension and, after a few rings, I left a voice mail.

Still smarting from my 20-minute wait on a government "hotline" (a wait that had been punctuated by reminders that I should not hang up, as I could expect an even longer wait if I did so), this was - overall - a pleasant experience. The VR technology had not only routed me correctly but had answered my questions as well as anyone could. Clearly, the VR interface performed better than the annoying "push 1 if ... push 2 if ... ad infinitum" that passes for most automated answering services today. But in the end, unfortunately, the result was about the same.

It's hard to believe that there was a time when I could telephone someone and get through to him or her, or to someone taking calls for that person, who was tracked down or flagged when seen. Now when I get these folks, usually located at the "front desk" or "reception," I am asked if I'd like to go into voice mail. This just scratches the surface.

I receive messages - some of which are time-sensitive - sent as e-mails when a 30-second phone call would quickly and conveniently do the job. I am dismayed to find, when catching up on e-mail at the end of the day, a message from someone I spoke with that morning proposing that I call a source, by telephone, at 2:30 this afternoon - which, by the time I get the message, is long past.

I am sorry to say that we as a nation have been disabled by the very technology that promised to empower us. Even sadder is the fact that our technology has not fallen short so much as we have.

Voice mail, and particularly automated messaging technology, has been employed primarily by managers as a labor-saving device. We the users have made the situation worse by delegating personal responsibility to these technologies, allowing our voice mails and e-mails to fill up like so many invisible stacks of digital messages. Rather than enhancing our communications, as they could if we used them to their potential, these technologies have slowed us down. I fear the same will happen with VR-based medical reporting.

Voice recognition as the medium for transcribing radiology reports is catching on. A key driver is its capacity to cut costs. Whole legions of transcriptionists, the backbone of radiology reporting since the advent of transcription equipment, are getting pink slips. Some are being rehired as correctionists, who find and fix VR-imposed errors. But not always. There are options.

Some places are tempted to hire less skilled workers to fix mistakes. Or they may do away with this level of QA altogether, making radiologists edit their own reports. This is especially worrisome, as it cannot help but slow down workflow.

Unless VR is intelligently used to ensure that medical applications boost productivity as well as cut FTEs, it will fall prey to the same problems that now beset more broadly applied communications technology. The promise of enhanced productivity will give way to newfound inefficiencies. And the reporting process that diagnosticians depend on will stumble, leaving physicians - and patients - on hold, wondering how things ever got so bad.