It’s human nature to want to know the future. Dating back to the 1400th century BC, the Oracle of Delphi was the most important temple in all of Greece. People would come from all over the world to seek a glimpse of the future from the temple’s high priestess Pythia.
Only the priestess could see the future. Her answers would determine the course of human affairs, from what crops to plant next season to when a country should declare war.
Today, when we want to know the future we consult with a computer, not a priestess. Machine Learning and other forms of Artificial Intelligence provide predictions to a wide variety of fields people are interested in.
But we still have a “High Priestess” problem.
Making predictions with Machine Learning and AI requires a unique set of skills. Today’s High Priestess is the “Data Scientist” - someone who understands statistics, can write a computer program, can get data from a database, and knows how to create charts and graphs to explore data. They also know the inner workings of Machine Learning models and how to fine-tune them to produce the most accurate and reliable predictions.
Most of the recent advances in the fields of Machine Learning and AI have come from today’s Oracle of Delphi - Large technology companies (Google, Facebook, Uber) and research universities where the skill sets of the Data Scientist are in abundance. The focus of the Machine Learning research at these institutions has been on improving model accuracy on larger and more complex data sets. They have created image recognition and language processing AI that rivals human performance.
In other words, Machine Learning research and development in the past several years has gone Deep, not wide. As a result, we have a technology that produces incredible results but that only a limited number of organizations have access to.
It’s time for Machine Learning and AI to go Wide. It’s time to democratize the technology so that any organization, large or small, can have access to prediction models built with machine learning. This shift from Deep to Wide is not going to come from the organizations that got us here. It’s going to come from product designers and usability experts. From startups and entrepreneurs who see the potential of making machine learning models easy to build and use.
Hiring a data scientist is out of reach for most organizations. Average salaries for these positions as $120,000 and can reach as high as $300,000. If these organizations want to move from their current data analysis efforts to predictive analytics using machine learning, they’re going to need tools that automate much of the process.
That’s why we created the Prediction Laboratory. We want to give anyone who can use a spreadsheet the ability to build, run, and share prediction models. It’s also why we’ve designed the Prediction Laboratory to be accessible from within Excel. If you can use Excel, you should be able to do Machine Learning.