Google’s Cloud AutoML is the future. Bringing easy machine learning to the masses.

At Google Cloud, Next ’19 Revisited event today for Google Cloud for Startups. We think that Google is doing it right in the Cloud AutoML direction. It will be the future.

The time-consuming part of machine learning lies in the pre-processing stage and often the ambiguous step of deciding which models and parameters to use. AutoML eliminates all of these worries. It does these in a simple, fast, elegant manner that greatly outperforms manually created models, believe it or not.

Instead of hiring and tasking a machine learning engineer to read through the ml papers to decide which models to run, and then setting up the environment for a machine learning job, we now only need to spend $20 to run to train a 1 hour Cloud AutoML model and have comparatively better results.

Imagine a world when the best machine learning models are offered to you as a cheap, easy to use commodity. The burden of entrepreneurs will shift from the technical responsibility of running machine learning jobs properly, to how we can create real-world uses out of machine learning to make things better.

This is more fun for sure.

A reminder of real world use cases for ml: Optimize inventory (retail), manage default risks (finance), understand customer conversion (marketing), optimize resources (transport), improve patient outcomes (medical).

Downside of Cloud AutoML: It still remains a blackbox and the trained model can’t be easily exported offline.

Soon, running a machine learning job will as easy as running an arithmetic function in excel. Mark these words.

For more posts like this, visit Latent App’s blog.