:) If U R Able 2 Reed My Report

Artificial intelligence, it has been noted (previously in this column as well as elsewhere), is a more protean entity than commonly considered. The usual conception is that of machinery thinking and/or behaving similarly, if not superiorly, to humans…nightmare scenarios include HAL9000 and the Terminator.

AI can also refer to the phenomenon of machines not becoming smarter, but rather influencing humans to behave more stupidly than we otherwise would. Dependence on machines makes us think less for ourselves, for instance. Or, to navigate tech interfaces in pursuit of desired outcomes, we are not permitted to input that which our intelligence would want, and we thus “dumb down” ourselves for pragmatism.

For instance, a computerized order-entry system does not let a clinician enter the relevant, detailed clinical history he knows the radiologist so desperately wants, and instead the official reason for ordering CT becomes “pain.” Presto! Otherwise-smart referrer immediately looks lazy and/or stupid about his patient, while the software looks brilliant to administrators and assorted paperpushers for meeting “meaningful use” requirements.

I have been noticing an insidious new front in the AI-via-human-dumbness struggle, and it is in voice recognition. Not, as you might expect, specific to dictation of radiology reports. Rather, I refer to cell phone (I refuse to call it a “smartphone” any longer) usage.

It was, I admit, a cute phase when everybody with a new iPhone was running around with their devices, having little chats with Siri and marveling at how very clever it all was. I’m sure I was far from the only rad who quietly smirked and waited for the bloom to come off the rose…soon enough, I thought, they will all discover the aggravations of inaccurate voice recognition, get fed up, and set this toy aside.

Instead, I found myself starting to make use of cell phone VR myself, albeit on an Android. Damned if the accuracy didn’t seem better than when I was using my radiology workstation. It sure beat trying to type everything out on the little touchscreen-keyboard with my less-than-dainty digits.

Sure, I had to make corrections, and not always because the VR turned what I said into a word salad. More often, because I wanted my transmissions to be as I would have expressed myself in person or in more formal print. Sometimes it was to preserve subtle nuances of my meaning. Sometimes out of being a perfectionist, knowing for instance the difference between “its” and “it’s.” I sure wasn’t going to let this stupid little machine make me seem like a dunce by getting such things wrong.

Much like radiology-reporting software, however, the VR did not “learn” as time went on (software vendors, I am calling you out on this lie you routinely tell), and while the machinery had endless persistence in throwing errors at me, I did not have similar stamina for making corrections. So I started letting things go, and just hitting “Send” on things that I once would have considered in need of fixing. The recipients knew me, they’d get my meaning. And probably not think the less of me.[[{"type":"media","view_mode":"media_crop","fid":"31822","attributes":{"alt":"voice recognition","class":"media-image media-image-right","id":"media_crop_2807801448559","media_crop_h":"0","media_crop_image_style":"-1","media_crop_instance":"3358","media_crop_rotate":"0","media_crop_scale_h":"0","media_crop_scale_w":"0","media_crop_w":"0","media_crop_x":"0","media_crop_y":"0","style":"height: 152px; width: 160px; border-width: 0px; border-style: solid; margin: 1px; float: right;","title":"","typeof":"foaf:Image"}}]]

Now, looking through my files of sent texts, my English teachers of yore would probably assume I had willfully forgotten everything they ever taught me. It sometimes looks like this might not even be my native language-perhaps I spoke in Urdu and ran my words through Google Translate? The result of a gradual willingness to sacrifice precision in the name of expediency…and to avoid hiking my blood pressure every time I felt the need to stop and fix things.

All of which has me wondering…how many of us dependent on voice recognition to meet our productivity requirements have gradually been worn down in a similar fashion? I know my dictated reports are far from the polished documents I once generated when I read half as many studies in a day as I do now, when the transcribers of my reports were either humans or my own two hands (yes, I used to type my own rather than dictating at all).

Further, if we allow ourselves not to care, even just in the name of “that’s how things are,” will anybody else? Referring clinicians? Or are they too overtaxed in their own specialty to read anything other than our IMPRESSION dictations? How about the patients who get ahold of their reports? Might they genuinely believe, as above, that the reports came through Google Translate, and the radiologist was reading from across the globe, not even licensed here?

I can imagine a malpractice attorney, hungry for an easy target, would more eagerly go after a rad who seems like he might not be capable of writing or speaking above a grade-school level.