Google’s AI Platform Raises Question of A.I. Vendor Lock-In

Google has announced AI Platform, billed as an end-to-end service that will allow companies to carry out artificial intelligence (A.I.) projects from data preparation through distribution and management.

For example, a (hypothetical) data team would use Google’s BigQuery to store and “clean” the data, before shifting to a product such as AI Platform Training to build up the application’s capabilities; from there, the team could manage the results via AI Platform tooling, and then share any discoveries through the AI Hub. It’s a stitch-up of longstanding tools and new capabilities, designed to entice companies to jump aboard the A.I. bandwagon.

How much will that cost? Well, that’s a very good question. “Kubeflow, AI Hub, and notebooks can be used for no charge,” the company wrote on the AI Platform Website. “You can learn about the pricing of our managed services like AI Platform Training, AI Platform Predictions, Compute Engine, Google Kubernetes Engine, BigQuery, and Cloud Storage here. You can also use our pricing calculator to estimate the costs of running your workloads.”

Google has spent years assembling the various parts of its commercial A.I. infrastructure. It’s going head-to-head with companies such as Amazon and Microsoft, which are equally intent on securing the loyalty of A.I. researchers, machine learning experts, and data scientists. Google’s AI Platform is clearly an attempt to stitch many of those A.I. tools into a more cohesive whole—and it hints at a changing world for A.I. researchers and other tech pros.

If these companies begin uniting their A.I. options into comprehensive “packages,” we could begin to see an increased amount of vendor lock-in among A.I. players. Sure, for users, there will be convenience—just dump your data-sets into your handy platform and you’re off to the races. But with that convenience may come the unsettling possibility that you’ll have a very hard time extracting any of your data and findings if you want to switch tools.

Moreover, most companies are still very much in the “discovery” phase when it comes to A.I.; last year, several studies showed the majority of firms either researching the technology or taking their first, tentative steps into implementing a solution. Because it’s such a nascent field, executives just aren’t sure what the technology can do for them—and in some cases, they might be under the impression that current A.I. and machine-learning platforms are far more sophisticated than the reality.

Combine that with an increasing threat of vendor lock-in, and it’s clear that A.I. represents a lot of danger and opportunity for companies. They’ll need smart tech pros to see them through.