Democratizing AI

I recently met Barak Turovsky and was blown away to learn how his team used next gen Artificial Intelligence to dramatically improve the quality of Google Translate, one of my favorite apps. (If you haven’t read it yet the New York Times in-depth piece on the AI transformation behind the app proves why great journalism still matters.) 

The big consumer technology companies like Google, Facebook and Baidu are leading the charge to use AI to improve our lives. AI is helping us better drive cars, translate language, spot cancer, schedule meetings and even win games. Armed with their seemingly bottomless cash hordes, the tech giants are paying huge salaries for AI engineers. 

But what about the rest of us? The 99.9% of businesses that don’t have the resources or expertise to innovate and invest in AI? How can regular businesses benefit from AI or -- to ask a more pointed question -- realize a competitive advantage to grow their business?

Turns out, it’s all about the data. AI and the machine learning at its core requires three things: big data, big analytical models and big computing power. You can easily rent super computing power from cloud providers like AWS and Google. And the algorithms that power the analytical models behind AI are in essence free thanks to the open source community behind Apache SPARK Machine Learning Library. But the one thing that’s still hard to get is the big data required to train your machine learning to do something useful for you. And I’ve noticed that the one very conspicuous commodity that the tech giants aren’t renting us anytime soon is the vast troves of consumer behavior data that they’ve amassed on the down low. It’s the key to their kingdom. It’s their competitive advantage – the moat around their Walled Gardens.

Until now. Today, dataxu announced an initiative to democratize AI and the big data that powers it. We call it Open AI for Ads , and it enables advertisers and their agency partners to easily create and run programmatically their own proprietary AI. How? We’ve brought together some of the largest most valuable third-party data sets, including launch partner Oracle Data Cloud, to automatically turn your first party data into an internet scale model that controls the decision logic of our TouchPoint DSP software. The result is completely unique-to-you AI that has been shown to dramatically outperform the usual DSP performance.

At a time when advertisers are asking their agencies to somehow do more with less, I think the only way forward for our industry is greater use of smart software systems that reduce the time and cost of managing media budgets while also lifting performance. This is the promise of programmatic after all: managing media investments more efficiently through technology. Transacting through APIs is a nice if somewhat modest first step, but what’s truly innovative is bringing proprietary systems of intelligence to automatically optimize the investment. Google, Facebook and Amazon have shown us the way to build incredible enterprise value by doing so. Now it’s your turn!