From the Inside September/October 2019
Artificial Intelligence is a hot topic for businesses. As with most big things, it hasn't come to the forefront because of rigorous research. It has come because of the success of other projects like it. It's already transforming several industries. It has been all over the news.
Despite what the growing excitement implies, AI is not a new concept. This hot new thing has been with us for quite a while.
There's a Spectrum Magazine article, written back in the early '90s, that shows you how to create an "Expert System," a form of AI decision making, inside the PICK Database. To be fair, the introduction of much more powerful computers and specialize chips has also driven the uptick in attention. AI software tends to be very calculation heavy. Faster processing, disk, and RAM have all extended the reach of AI.
While most CEOs or CIOs are aware that it exists, many are still unsure how to employ it to their benefit. Knowing AI makes robotics, drones, image recognition, and driverless cars more viable is only useful if you are in those specific markets. When innovation comes from executives seeing it on the news, it tends to lack clear links to the business they are operating.
Additionally, a lot of AI software is outside the reach of most companies' resources and budgets. Without clear plans and serious benefit analysis, attempts are likely to be money pits, not successes.
It comes down to a fundamental confusion over exactly what AI software is and what it can do.
I like to explain it by dividing AI onto three separate components: Data Analytics, Predictive Analysis, and Machine Learning. And yes, I'm lumping high-speed data mining in for simplicity's sake.
True AI is all about Machine Learning. But this is the most sophisticated form of AI and likely not going to be used in business software. Those edge cases are growing and becoming less edge-like, but they aren't center-stage today. Data Analytics and Predictive Analytics are really what businesses should be focusing on.
Data Analytics is all about finding trends in your data. Most enterprise software has so much data. No one know where to start to get answers without someone asking smart, focused questions. While this sounds a lot like Data Mining, the difference is that Data mining returns the specific results, already formatted for processes like dashboards and reports. While Data Analytics is designed to take raw data, and look for any patterns outside of the already understood Dashboards. One pushes data into assumed relationships while the other derives relationships from the data.
Data Analytics is also used to support or debunk existing assumptions.
Predictive Analytics, on the other hand, is all about taking the data you have and projecting into the future. If you have an inventory control and purchasing system, then you are likely already using an older form of Inventory Forecasting. These forecasting reports are using a Predictive Analytics model.
The main difference between these existing reports and the newer Predicative Analytics AI processes is the complexity. Modern Predicative Analytics take into account far more data complexity. They address What If questions, rather than just estimate future needs based on past performance.
AI Machine Learning is the Holy Grail; and much more complex. It combines the concepts of Data Analytics and Predictive Analytics, but instead of the person asking a specific question, the computer returns its own assumptions based on the data provided.
This is why Artificial Intelligence and Machine Learning are so important for business. You provide your data to an AI, and instead of have a person who knows how to ask a precise question, you have the computer providing these assumptions for review. As we all know, people are better at picking apart an answer than at providing a complete one.
Make sure you join us at the International Spectrum 2020 Conference in Florida to talk more about how Artificial Intelligence will be used with your MultiValue database.