There has been little to no research done on ways of improving visual editing, but there has been a lot of research in other endeavors where humans have strived for excellence, and perhaps some of it transfers here? It might even provide insights about how and why AI visual tools can improve human performance.
An individual sport where there has been a lot of rigorous research is competitive running, and it may serve as a proxy for ways of improving human performance. Running performance tends to occur in two areas:
- Body mechanics getting better and better
- The runner’s brain gets more able to handle the demands and stresses, particularly in marathon distance running events: mental fatigue plays a huge role.
If only a coach could look at how you run, tell you to change your stride this way, your cadence that way, land your feet just so, and wear these shoes, follow this “Rigid” formula, and you instantly become fast! This rigid, formulaic approach never works: among other reasons the runner requires many adjustments because no two people are equal.
The same thing was discovered when scientists tried to create robots that could run. The first attempts were driven by “Rigid” rules that instructed the robot exactly the “perfect” way to run, and that sort of worked: for one type of surface. But add a hill or descent or rain and the robot failed, no matter how fast its computers were.
It turns out “Loose” is much better than “Rigid”. Let a runner’s body determine what is the most efficient way to run at a given moment in a given place. Or, in the case of the robot, give it a few simple rules and the ability to make adjustments. What scientists found was the runner ran faster, and the robot could not deal with up and downslopes, uneven surfaces, and other challenges. Scientists in both areas found that loosely controlled movements allow for improvement, and rigidly controlled movement does not.
How do runners go from rigid, and clumsy, and slow, and low endurance to grace and fluidity and efficiency and speed and the ability to maintain through a marathon? Repetition. Runners have to get out there and run mile after mile, week after week, and subconsciously let their body make small adjustments to get more efficient, just like basketball players have to take shot after shot from the free throw line to get proficient, just like elite athletes in any sport have to use repetition to get really good at whatever their sport demands. Over time and with the right kind of practice they get to where they are executing the basics from their subconscious, and not having to think hard all the time.
What does this have to do with DAM? Consider the metadata searches we have been forced to use for the last 30 years: they are “Rigid”. You either know the magic words to put in your query, or you lose. And if the visual objects you are searching for have little or no metadata, you are sure to lose.
An AI approach to visual search is “Loose”! Our NOMAD visual search engine is trained by many millions of text and visual combinations done by millions of people from all over the world. There is no restriction on using the right words. Because some of the associations between your query and the returned image or video are “learned” from so many different people all around the world, there may be results you did not expect, that might broaden your horizons in beneficial ways and enable you to tell your visual story even better than ever before. This is only possible because NOMAD’s AI training has taken you from “Rigid” to “Loose”.
Similarly, the theme of repetition comes into play in AI. You have all heard the term “Big Data”, which basically comes down to this: if you have a huge set of data to repeatedly train your AI engine on, it will do better. If you can periodically feed it newer, more recent data, it will “learn” even more, and do an increasingly good job of understanding your visual objects. The more varied sources you can use to “train” your model, the more robust and informed it will be and the more flexibility it will have. And flexibility means “Loose”!
In elite athlete athletics, the concept of a “quiet brain” is crucial. No one can perform difficult tasks if part of your brain is screaming at you “Stop! Why are you even trying this!”. And noisy brains lead to fatigue: how have you felt at the end of a long, difficult, and challenging day at work, solving lots of problems? Could you win a marathon (or do a great creative job) feeling that way?
On the other hand, we hear elite athletics talk about “being in the zone” and “flow” where they achieve spectacular performances but are aware of essentially zero conscious effort. Remember we talked earlier about how repetitive training (with the right mind set) leads to the subconscious managing the task (do you consciously think about every part of every step you take, or do you just think “Walk across the room” and it happens?). As one expert put it:
“The true essence of skillful running is not correct movement of the limbs but a quiet brain.”*
Again, how does this apply to visual editing? If you spend your day constantly trying to “out think the DAM machine” by being limited to a metadata only search, and keep trying to come up with the right words to find the thing you want, how much of your mental capacity is left to then creatively select just the right visual object to tell your story?
Why not let the global wisdom you get from Big Data help you out? Shouldn’t the search part be easy, and then you have lots of energy for the creative activities of selecting the right object and putting it to optimum use in your webpage or document? Here again, shifting the workload from search using an AI tool like NOMAD and thus making it “Loose and easy” leaves you more mental bandwidth to choose the perfect object and use it brilliantly. Isn’t that your mission, versus fighting with the DAM machine?
Where do you want to spend your mental bandwidth? Fighting and struggling to find things, or creatively?
Hopefully this suggests how, to optimize human behavior in visual editing, AI like NOMAD is your friend and not your enemy.
- Reference:”80/20 Running” by Matt Fitzgerald