5 min read
Medicine was the first ‘art’ that I came across. To consider interpreting the scientific principles that regulate the human body an art is a bit unnerving, but it’s really about as much of an art today as it was 100 year ago. Even with genomic sequencing and fancy MRI scans, the diagnosis and treatment of disease often comes down to patterns, experience, and even gut feelings.
patterns in medicine
There are textbooks that can teach you how to read an EKG of the heart, but an experienced cardiologist can identify a heart attack with just a quick glance at the squiggle on the piece of paper below. How is that possible? Along those lines, an experienced dermatologist can decipher skin cancer from a mole with near certainty just by looking. The radiologist can spot a pneumonia on an x-ray without straying away from their Twitter feed. The internist can pick up a heart murmur within a few seconds of listening to muffled sounds through a stethoscope. The examples are seemingly endless. The human capacity to learn and identify patterns is phenomenal.
These types of skills, usually developed somewhere around Malcolm Gladwell’s 10,000 hour rule, are found in surgeons too. For example, through a small incision I can identify a small target on the back of your spine and blindly place a screw in the bone that curves delicately around the spinal cord itself, using only my “mind’s eye” and “the feel of the screw” as navigation. It doesn’t just sound scary, it actually is! Of course, that’s not how operations are done in modern day. Like most advances in medicine, artificial intelligence was born out of the need to do things better and to make things safer.
patterns in art
Traditional visual art is also classically a study in pattern recognition, not unlike medicine. There are patterns in the subjects, shapes, brushstrokes, color, and more. Even a non-expert can recognize broad categories of stylistic patterns based on popular works. van Gogh’s Starry Night or Picasso’s Guernica are easy to spot. However, the actual elements that go into creating them are much more difficult to define and replicate, if not impossible. Unlike medicine, the evolution of generative art from traditional art is more likely born out of creativity, rather than necessity.
AI in medicine
Nowadays, there is computer software that can interpret patterns of a heart EKG, and also improve upon itself over time. Similarly, algorithms exist to decipher chest x-rays, brain CAT scans, and even pathology microscope slides. All of this can be done with similar or better results than a human, and much faster. In my line of work, I cannot remember that last time I placed a screw in the spine without the use of a computer robot. The algorithm suggests the starting point and trajectory, then drives the screw into place, all while learning from itself and having X-rays autonomously guide it along the way. [Think of it like a real life production version of Tesla Full Self Driving.] It’s proven to be safer and more accurate, so I think it would be unusual not to use the technology. It’s also made a complicated task relatively straightforward.
While a lot of people are worried about doctors being replaced by AI robots, or that medicine is losing its ‘art’, my strong belief is that AI is a very powerful tool to help improve outcomes. Those doctors that can understand the utility of AI and how to integrate it into the care of patients will be the most successful. It is possible that the definition of the ‘art of medicine’ may lie in how to successfully apply these new tools, as much as it once was tied exclusively to pattern recognition. I don’t think anybody can envision hospitals hiring a 12-year-old kid and giving them AI software to run the cardiology unit. The human component of expertise remains vital, no matter how much we integrate with AI.
AI in art
Along these lines, there is a palpable sense of apprehension with regards to the inclusion of art generated by code, let alone AI. The popularity of on-chain based generative art (not AI - eg. ArtBlocks platform, Autoglyphs, etc.) tells me that the definition of art is quickly getting redefined. Even though my viewpoint is biased, code can clearly generate something complex and/or visually pleasing, albeit unique from traditional art aesthetics. Much of the uniqueness comes from the capability of blockchain-based art to permit 1) motion 2) interactiveness 3) randomness and 4) provenance/ownership in ways that were not previously possible. AI adds additional elements to the generative process, and also increases the depth and flexibility of genres seen in the output. One has to look no further than the BrainDrops platform or the Nvidia AI art gallery to get a sense of the diversity and creativeness in AI-based visual art that exists today. Similar to AI in medicine, the human component is still inseparable. Even Botto would be in a slump without feedback from humans!
When I see Pindar Van Arman’s autonomous painting robot, I can’t help but think of the robot that I use to put screws into the spine. The two are more similar than their intended purposes may seem. AI will not put traditional art out of business, nor will it take over the entire category of generative based art. As in medicine, AI is a tool that adds, rather than competes, with the possibilities around us.
about me: artist, surgeon, and neuroscientist in no particular order, with artificial intelligence permeating all three. I'm influenced by the reactions different people have to illness, despair & suffering, and the ability to overcome odds.
connect with me on twitter @wetpotatobrain