Talking deep learning with AlchemyAPI CEO Elliot Turner

AlchemyAPI has a lofty but challenging goal: Democratize big data for the masses. A look at the emerging artificial intelligence stack.
Written by Larry Dignan, Contributor

AlchemyAPI is a deep learning company without a face---actually a user interface---with a lofty goal to democratize artificial intelligence.

The company, based in Denver, specializes in training deep neural networks to analyze information and carry out cognitive computing tasks. In some ways, AlchemyAPI could be considered David to IBM's Goliath. Or IBM just buys David at some point.


Deep learning, using algorithms to model data so machines can learn and adapt, is a hot space right now even though the so-called killer application or industry hasn't been found. For now, deep learning technology could mean anything from finding facial patterns on Facebook to combing through the human genome and medical literature to cure cancer.

AlchemyAPI's technology has been applied to vision and sorting through unstructured data. The company can process everything from SEC footnotes to images to video in context.

We caught up with Elliot Turner, CEO of AlchemyAPI, at GigaOm's Structure Data conference over a storage shed at Chelsea Piers in New York. Turner and company were good sports hanging out in 34 degree weather since GigaOm apparently doesn't do press rooms or briefing areas for anyone not on its research team these days.

Here's the recap:

The democratization of big data. Turner's main mission is to democratize big data and enable real-time analysis of unstructured information---Web pages, chats, video, text and SEC filings to name a few items---for both large companies and small. At the GigaOm Structure Data conference in New York, Turner was slated to be on a panel with Stephen Gold, vice president of worldwide marketing and sales for IBM's Watson business unit. The storyline is that deep learning should be available to all, not just large companies. "We're not solving just one problem and want to put our capabilities in the hands of everybody," said Taylor. "We want to do for big data what AWS did for infrastructure."

Where's the UI? AlchemyAPI's technology can be found at a bevy of companies ranging from Hearst to Jive Software to Outbrain to trading firms looking to combine news and regulatory filings with algorithms. In all of these cases, AlchemyAPI's technology serves as a base layer and customers put on the front-end experience. Should AlchemyAPI want to expand its wares to a broader market beyond developers, it may want to ponder a UI. Turner said AlchemyAPI would ponder a front end to make its algorithms and data more accessible, but wouldn't want to compete with customers. Nevertheless, AlchemyAPI's labs team has at least pondered a front end interface to target "non engineers." "The long-term vision is to make our technology available to a wide audience," said Turner. "It's such a huge space."

The artificial intelligence stack. Turner frequently returned to the concept of AI as a stack---much like a computing stack. That stack today is just being formed. AlchemyAPI is obviously at the base layer with its programming interfaces, but could plug into other levels over time. Today, there are a lot of companies that plug into various levels of the AI stack, but the problem is that customers have to comingle vendors and approaches. IBM has core language processing tools and moves up to Watson. AlchemyAPI also sounds like it would like to provide a full AI stack over time.

Do you need structured data too? Turner emphasized that AlchemyAPI's focus was unstructured data, but didn't rule out combining structured information too. "That's where the impact is: Correlating unstructured signals and structured data," he said.

Use cases. AlchemyAPI's technology is used across multiple industries---social media, finance, and advertising to name a few. According to Turner, AI used to be brittle and confined to one use case. Today, AI can adapt to many industries. That's why companies like Google, Nuance, and IBM are all chasing deep learning and artificial intelligence. The goal is to get to a point where data leads to insights instead of clever algorithms designed with human bias.

What's the end point? Artificial intelligence and deep learning is a disruptive technology, but I can't help but think about the end of the line and what the average bear actually gets. Will AI insights come via a Watson appliance, a smartphone or a wearable? Will there be a chip embedded in my head? "No one knows yet what's on the end," said Turner. "It's hard to see what's on the other end because it's a disruptive technology." Turner added that there will be more user facing components to AI, but questions and answers will be a part as will data visualization.

Editorial standards