Why enterprise machine learning is struggling and how AutoML can help
There are some who warn us about machine learning (ML) taking over and replacing us in the future. As a person who has been working in the ML space, as both researcher and engineer, for years, I have come to a very different conclusion. Although irresponsible application of ML could indeed be dangerous to humanity, ML solutions are no where nearly capable of taking over our civilization at least any time soon. On the contrary, ML is well positioned to benefit humanity in two major ways.
First, ML can set humans free from certain boring and tedious tasks, e.g., data entry, by automating them. Second, ML can help expand our abilities by amplifying our cognitive abilities, e.g., generative design systems, powered by ML, are already helping designers take their creativity to the next level.
The demand for ML solutions has soared over the past few years. More and more enterprises are aiming to reap the benefits of this exciting technology. However, currently, most applications of ML at enterprise are not as successful as we would like them to be. In fact, the majority of enterprise ML projects are “pilots” and do not make it into production.
