Does Machine Learning Cause The Chasm Between Knowledge and Understanding and provides us with the tools for our next step in the evolutionary process?
“Deep Patient” is an initiative to predict patient’s medical future through Artificial Intelligence. An apt representation of the effects of machine learning could be through a program that may be oblivious of the fact that knocking someone on the head could have a dizzying effect for example, or that diabetics could gobble a huge bar of chocolate in a single sitting. The program may not even be aware of the fact that the arm bone and the wrist bone are interconnected. The program’s actions and responses are based on the data that researchers fed into the program a couple of years ago. The data may have been baseless and discombobulated medical records of several thousands of patients.
However, upon analysis of how these blind bits are interconnected, the program surprisingly and indeed exceeding expectations was able to diagnose the probability of individual patients being susceptible to certain diseases, comparatively accurately; in some cases surpassing physicians in flesh and blood. The occurrence of some diseases was unpredictable or in other words, they were undiagnosed and therefore untreated until the advent of machine learning as new and emerging technology.
If one’s query to one’s physician is explaining the reason behind Deep Patient’s thought that it might be prudent for the physician to prescribe statins or undergo preventive surgery to patients, the doctor may not have an answer, but there is absolutely no reason to jump to the conclusion that the doctor isn’t technically savvy or smart enough. Deep Patient is a kind of artificial intelligence known as deep learning synonymous with machine learning that seeks and finds relationships between pieces of data, unbeknown what that data might represent.
Once the relationships among the data were found, a network of information points was assembled with a weighting ascertaining the likelihood of the interconnected points ‘firing’ which led to affecting the interconnected points akin to firing a neuron in a brain.
An example of understanding the reason behind Machine Learning’s thought is the likelihood of patient developing schizophrenia and a doctor internalizing those infinite points and their connections and weightings. The sheer volume of interconnected complex relationships is mindboggling though. As a patient, one has the prerogative of rejecting Machine Learning’s unverified conclusions but there is a risk, as, in reality, a “black box” is used in diagnostic systems that are incapable of explaining their predictions as, in some instances though, they are far more precise in comparison with human doctors.
This truly is looking at the future in the face, in all its hue and color transcending from one domain to the other. The navigation system of one’s phone, predictive typing, language translation, recommending music and so on depend on machine learning already.
With different forms of machine learning turning into an increasingly hi-tech tool its myriad functions may transcend human comprehension or in other words simply mindboggling. If, for example, the number of probable chess moves was to be subtracted from Go moves, the remainder isn’t even comparable with innumerable atoms in the universe. Nonetheless, Google’s AlphaGo program based on A.I. outsmarts the human brains that are the cream-of-the-crop of human brains, although its knowledge about Go is limited to collecting and analyzing numerous moves in numerous recorded games.
If one were to analyze AlphaGo’s inner workings to try and discover why a particular move was made, it’s likely that one would find nothing but a set of extremely complex weighted relationships between its data. AlphaGo simply may be incapable of interpreting the moves in a manner that humans can understand.
Machine learning’s algorithms function because it has superhuman powers of capturing, given the complexity, fluidity, and the beauty of this comprehensive universe where everything affects and is affected by everything else all at the same time.
As humans, we are beginning to realize and submit ourselves to accepting that our universe is far more complex than the principles that we devise to find meaning
Machine learning isn’t the one and only tool and strategy that has enabled an encounter for humans with the unfathomable complexity of the universe. This privilege though has a price tag humans ought to be unaggressive about their pursuit to understand our universe and try and find meaning. We can easily conclude the biggest challenge for AI researchers, Machine Learning: Raising the contradiction between Knowledge & Understanding!Related Posts