Why is AI / ML in vogue?
Why is AI / ML in vogue? … An explanation that is easy (but not easier) that a layman could understand ..
I was asked yesterday by a curious (and bright) student, “Everyone and everybody wants to implement ‘AI and ML’ to their data. What exactly is AI and ML, why is it so fashionable and trending now?”
I had to pause and realized that the answer is not simple. It has many folds to it .. A look back at application of Statistics, pattern recognition, costs and advancement of computation (processors) and cost of chip memory is necessary to understand the significance and trending of AI (Artificial Intelligence) and ML (Machine Learning)
In any research group after analyzing data, previously a statistician’s standard set of redundant answers would “Hence the null hypothesis is rejected”, “Two populations are not similar” or something on the similar lines .. with almost no understanding of the pattern of the data and trends in data.
Now, such answers are simply not acceptable. In fact, the role of statistician with the fancy names of their “tests” and “corrections” (Chi square test, Bon Ferroni’s correction etc etc ) might be going in oblivion as simple to complex statistical programs using Excel sheet, some available free on open source, would be taking over the work of what the previous statisticians were doing for their living.
Performing statistical test on any data, big or small, is now just a click away. Statistical analysis packages can run a data through almost all possible tests to give “Insights” in to data. Data visualization programs can give good understanding of the data and population cohort that is under study. These programs/applications/packages are now even available in open source or could be bought as packages (e.g. Tableau, Azure, Power BI to name a few). These applications require minimal coding or programming)
The falling cost of memory and processors has made these applications affordable to even a common man. This has improved the efficiency of data analysis and has boosted “Analytics” as routine part of every business.
Dealing with large data sets and big data (data flowing in, in real time) is now easier with such inexpensive computational power. Data fed to neural networks (Artificial Neural Networks) to train them for variable to output correspondence going through a black box is critical to train the ANN systems. The more and accurate the data, the more likely it would model a system.
Similarly learning algorithms for Machine Learning and algorithms for interpreting the patterns and associations (artificial intelligence) become “smarter” as they “digest” more data and learn! They connect variables to outcomes with precision as they learn with every data point they consume.
These algorithms and application can then be used easily for Generative or Predictive analysis. This has huge significance especially in big and continuous data feeds or for genetic data where genetic associations of disease to genes or Single Nucleotide polymorphism (SNP) could be predicted for prognosis or occurrence of a disease.
So, in short, the cheaper hardware and storage with faster and powerful processing at affordable cost, computing has gone to next level in application. It would be common in next decade to have a ML and AI algorithm working in background for every process that a human requires, be it data generation, acquisition or processing and management.
I must clarify here that AI and ML are in vogue now because of cheaper hardware, but they were in practice for very long. Previously patterns in data were determined by complex statistical methods, requiring high end mathematics and computing that required specialized resources (statisticians) and expensive computing tools. Now we call it AI and ML in short.
(In the next article in the series, we will see how “wearables” with embedded sensors, iOT (Internet of Things), Data Analytics, Artificial Neural Networks (ANN), Automation, Robotics, AI (Artificial Intelligence) and ML (Machine Learning) is causing a huge paradigm to shift in diagnostics and prognostic Clinical Medicine & Surgery)
