Ebony and ivory in harmony

Time changes our expectations. Well, not time exactly. It's more the clash between our initial expectations and reality.

Eventually, we come to see that what we thought would be black or white actually is not, in fact, could not be.

More amazing than how many shades of gray are actually possible - and I'm convinced there are more shades than the rods in our retina can perceive - is our resilience in believing that things can, in fact, be either black or white. We do it over and over. The paperless office. Filmless radiology.

The phenomenon goes well beyond concepts to the roots of what support them.

Take a recent development. A "correctionist" is a highly evolved form of the "transcriptionist." The developers of speech recognition initially said transcriptionists would be wiped out by the technology. Algorithms would listen to doctors and turn spoken words into typed ones. (I'm told now you don't even have to speak real slow, as ... you ... used ... to ... have ... to .... do.)

But there's still one problem. Nobody can guarantee that the computer will get all the words right all the time, no matter how slowly you speak or how well you enunciate. The makers of speech recognition software have always known this. Nothing is infallible.

But for years, they had been telling radiologists how easy it is to look over the speech-recognized text created from what they dictated and correct the few wrong words that pop up.

Unfortunately, doctors typically aren't very good at proofreading. Like most people, they tend to see what they expect to see. Consequently, if they have just dictated a report, they will expect to see the report as they dictated it. This raises the onerous possibility that the word that's wrong will be glossed over with the one that should be there. And that's not good.

The labor-saving angle is perhaps the biggest chink in the techno argument for voice recognition. Whose labor is being saved, if the doctor, who is accustomed to dictating a report and moving onto the next case, has to take time out to copyedit the report? Not the doctor's. No. It's the transcriptionist's time.

This is where the highly evolved form of professional - the correctionist - comes in. The correctionist knows what to look for in radiology reports, listens to the verbal dictation, and proofreads the typed text against the voice, making whatever changes are needed.

I predict this idea is going to catch on. Radiologists pushed to adopt voice recognition programs are going to love it. Transcriptionists facing extinction are going to embrace it.

Correctionists, in fact, are all but inevitable. If you look back on history, there aren't many times when technology has completely replaced people. Automatic pinsetters come to mind, but even now - 50 years after the first pinsetters were installed - you see guys walking down alleys, a foot in the gutter, fetching the bowling pin that got away or dislodging the one that jammed up the works.

In a techno sapien hybrid of the Peter Principle and Darwinism, machines rise to their level of incompetence. People adapt.

There are no politicians preaching "no human left behind," no engineers lobbying to exclude the human workforce. What happens happens.

And you can bet the result will not be black or white.