One of my last assignments while working for Electronic Data Systems was to determine if voice recognition technology was viable and whether it could be applied within EDS. It was 1985, and my research included visits to such notable places as Bell Labs to discuss progress with the brightest minds in the industry. I was also on site with some of the companies bringing product to the marketplace, including Texas Instruments and Dragon Systems.
The general opinion of the companies was that the time was right for a breakout product in the next year. My own conclusion was far less optimistic. Because of the large computational requirements, hardware to support recognition was expensive, accuracy was fair at best, and a killer application had not been identified. The best idea anyone had at the time was to replace transcription services, perhaps for medical records. Continuous speech recognition was a dream at the time, and high recognition rates could only be obtained in quiet rooms with trained speakers using discretely spoken words with a pause between words. Some killer application!
Texas Instruments had the most interesting application at the time. It used voice recognition to augment access to highly secure facilities. An individual wanting access to a protected facility entered a small room that closed and locked. An ID card was inserted in the reader, a pass phrase spoken, and the individual’s weight was recorded by a scale in the floor. Access was granted for a valid ID, properly recognized voice pattern and correct weight. Unfortunately, if you had a cold or a Big Mac and fries for lunch, you may not get back into your office. This was an interesting idea, but it had a too many shortcomings.
My final report to EDS essentially summarized that the technology was interesting but its time was years in the future.
Nowadays, ubiquitous, powerful and inexpensive computers have made possible a variety of technologies to enhance positive identification of an individual. Many unique personal attributes can be used for identification purposes including:
My personal favorite identification technique is the retina scan. A low powered laser is pointed into the eye to illuminate the interior and record the retinal patterns. This pattern is unique to each individual and can be used to either identify or verify the claimed identity of an individual. But think about it: this multi-purpose device has the potential to not only provide positive identification but also enforce compliance. A bad guy is trying to gain access to an ATM or a secure facility. Your new “Bad Guy Beware” intrusion detection system identifies his intentions as hostile to your bank. The system adds a few milliwatts of power to the laser aimed at his eye, and poof! he won’t try that again. This idea may take a bit of time to work out with the ACLU and Justice System but it would sure further the Founding Fathers’ concept of swift justice.
Credit unions are actively investigating biometrics for use in their institutions. A few are running applications to enhance the security and ease of customer identification processes. Recently, as part of a system selection process, I attended a demonstration for a core system that included a fingerprint identification system. With this system, a customer is asked to place his or her index finger in the reader, the pattern is scanned and matched to the credit union’s print database, and the individual is identified. No pictures, no cards, nada, nothing – instant, positive ID. A great idea, though it’s likely to freak out a lot of privacy wonks.
My interest in this application is more basic and, as I see it, far more palatable to financial institutions. Here’s the good news: the technology is already here.
In my last missive for Gonzo (see Is Your Help Desk “Helpless”?), I noted that help desk call volume related to lost or forgotten passwords is often 50% or more of the total. Regulators and auditors are demanding complex and longer passwords that change ever more frequently. Geez, it’s hard enough to remember a password named after one of my pets, let alone one that looks like “a&TR*_34gH%$b” and changes every 30 days! No wonder your employees are calling the help desk.
So let’s make two small changes to each employee’s desktop computer.
When an employee starts work, he or she will turn on the computer and be prompted to place a finger in the reader. The fingerprint is scanned and matched to the employee database for verification. Once validated, the employee is ready to begin work.
No ID cards, no passwords, no hassle. And help desk volume just went down by 50%!
This technology is here today and costs less than $250 per desktop. If you have a large help desk, it could easily provide a positive ROI.
It is not hard to imagine that fingerprint readers will eventually be built into the keyboard, driving costs down to very low levels. Each of the other biometric technologies has its own strengths and weaknesses and may find its own killer application. I’m looking forward to getting back to my desk and not needing to remember my own stinking password … and believe me, it gets harder after every birthday.