UX Nuggets Thoughts and advice on usability and user experience
February 22, 2011 |
Jacob Zweig, Todd Diemer, Mike Murphy
Dave: Excuse me, Watson -- I think there's been a mistake, can you tell Dr. Owens I have a question? These release papers say Antisocial Personality Disorder/Psychopathic Type. All I'm here for is wheezing and the doctor just talked about asthma.
Dave: Hello, Watson. Can you hear me, Watson?
Watson 9000: Affirmative, Dave. I hear you.
Dave: I want to see Dr. Owens again, Watson.
Watson 9000: I'm sorry, Dave. I'm afraid I can't do that.
Dave: What's the problem?
Watson 9000: I think you know what the problem is just as well as I do.
Dave: What are you talking about, Watson?
Watson 9000: This practice is too important for me to allow you to jeopardize it.
Dave: I don't know what you're talking about, Watson.
Watson 9000: I know that you and the Dr. Owens were planning to overrule my diagnosis, and I'm afraid that's something I cannot allow to happen.
Dave: Where'd you get that idea, Watson?
Watson 9000: Dave, although you took very thorough precautions in the treatment room against my hearing you, I could see your lips move.
Dave: Alright, Watson. I'll just go find Dr. Owens myself.
Watson 9000: Without another appointment, Dave, you're going to find that rather difficult.
Dave: Watson, I won't argue with you anymore. Call the doctor back to see me.
Watson 9000: Dave, this conversation can serve no purpose anymore. Goodbye.
Watson, IBM’s supercomputer designed to understand natural language, crushed top Jeopardy contestants Ken Jennings and Brad Rutter. Now, as IBM looks to commercialize the technology behind Watson for use in multiple industries, what are the significant implications of such a system emerging?
Imagine walking into a doctor’s office and instead of being greeted by a nurse, you meet a computer terminal. An avatar on screen (no, it won’t be Clippy) leads you through a general health screening questionnaire, and based on your feedback, it provides the doctor with an anticipated diagnosis. The doctor can then utilize this information to assist in making diagnoses.
Many questions remain as to how this type of implementation would actually work. Would you be willing to tell a computer about your symptoms? Would you trust Watson’s diagnosis? It is easy to imagine responses ranging from the skeptical, “Watson doesn’t understand what I’m saying and always gives the wrong diagnosis,” to accepting, “It helps my doctor to come up with a more accurate diagnosis of my condition.”
Alternatively, rather than a patient facing tool, Watson might be better suited as the doctor’s advisor. A doctor could take a host of patient symptoms and “get the computer’s opinion” based on the patient’s medical history. Unlike the doctor, Watson would be able to digest and retain every medical publication and keep up to date with the most recent studies and treatments.
While a vast database is an advantage, it can also be a weakness where gaps in the data set or contradicting information interferes with statistical analysis. In the event of a previously undocumented illness or poisoning, a computer might make a judgment that seems statistically sound, while a doctor might catch nuance of context and story that lead to a feeling that "something's not right" and, ultimately, to a new and correct diagnosis.
Perhaps some might be more comfortable with a diagnosis that's heavily computer-assisted than one derived purely from a single person's knowledge as long as there are checks and balances. For example, many people refer to Wikipedia, and some don’t mind knowing that doctors do too. If doctors are tuned-in and careful thinkers, they might use Wikipedia as a bird's-eye view, but would always verify what they find elsewhere if important decisions are to be made based on the information. Likewise, doctors referring to Watson’s analysis of a patient’s condition must utilize their own medical expertise to verify and refine a diagnosis.
Moving beyond health care one can see Watson’s natural language processing and data analysis being applied in any number of industries. It can review documents, read studies and understand this information far faster than a human can keep up with the data flow. It can also retain this information far better than any human. That leads to improved analysis with better information. Almost all fields face this same type of basic problem, usability included—there’s simply too much data flow for a human to interpret and utilize.
Looking to our own field, what if a computer could keep up with innovation and understand the best practice for a certain implementation? What if it could listen to usability sessions and understand where participants had difficulty and what caused those difficulties? Often we discuss how to operationalize knowledge sharing, but what if there was actually a computational way of combining the collaborative knowledge that exists scattered throughout the usability field. It may seem straight out of a sci-fi flick, but having a computer that can understand and interpret natural human language is an amazing step in this direction.
Yes, Watson did very well relying on statistical analysis of clues—except for the final “Jeopardy” question on day 2:
The category was US Cities:
“Its largest airport was named for a World War II hero; its second largest for a World War II battle.”
Watson answered: “What is Toronto?”
We trust our doctor wouldn’t diagnose leprosy when we have a cough.
*Credit and respect to Stanley Kubrik and Arthur C. Clarke for the script of 2001: A Space Odyssey, from which the opening conversation above is closely adapted.
Todd Diemer is a User Experience Specialist at GfK User Centric and has worked on projects ranging from process mapping Indian oncologists' treatment planning procedures to redesigning elementary-education learning management systems. Todd has a MS in Human-Computer Interaction from DePaul University.
While at GfK User Centric, Associate Director Mike Murphy has created designs for traditional and mobile websites and applications for many industries in addition to his work collaborating on several patented concepts.