Americas

  • United States

Asia

Contributor

Why this company will help change the future of artificial intelligence

Things are good in A.I. -- but there are growing concerns.

future tech
Credit: Thinkstock

Things are going insanely well for people in computer science. I mean, our work is everywhere. Nearly every process imaginable is powered by a machine at the middle. The computer has transformed communication, retail, how we access information and how we navigate around the world. Computers are winning.

Things are even better in A.I. We are at the beginning of a renaissance of interest and utilization of intelligent systems in an ever-widening sphere of influence. It started with the big tech companies (Google, Apple, IBM, Facebook). Now it is expanding every day as advancements in machine learning, voice and image recognition, and intelligent analytics move into the enterprise — at an increasingly faster pace, driven by over half of the Fortune 500 believing that A.I. is crucial to their future.

As I have conversations with many of these companies that are looking to adopt emerging A.I. technologies (most based on machine learning), there are two issues for them that keep popping up: “How do we integrate these technologies into our workflow and decision-making?” And even more crucial: “How do we integrate them into existing best practices, business rules and methodologies?”

That is, how do you integrate analytic systems that can learn to recognize entities in images, categorize activity as fraudulent or even predict the sales of different products based on weather information, with the business thinking on how to deal with these situations? How do you integrate systems that “think fast” with the deeper reasoning of the business strategies that require you to “think slow“? How are companies going to integrate the machine’s ability to recognize with the human ability to reason?

My concern is the potential of companies to become disappointed and disenchanted with A.I. — just as it is re-emerging — because these questions go unanswered. (Narrative Science has its own take on the solution, but that is a discussion for a different time.)

I am finally a little relieved about this issue because Elemental Cognition, the company founded by David Ferrucci (the man who drove the work on IBM Watson), has come out of stealth mode and revealed that its mission is to solve this problem.

On its website, the company defines the following challenge:

Today’s AI systems can help you locate the closest restaurants or find answers to simple questions. 

But, these systems cannot grasp the underlying meaning of language or provide rich explanations behind their answers.

The next grand challenge in AI is to build systems that truly understand.

Elemental Cognition’s technology seems to be aimed at the problem of going beyond the simple recognition/response models that dominate the A.I. landscape right now. I suspect Ferrucci is building out a technology that will combine outputs of deep learning and other machine-learning systems with the ability to draw inferences from, reason about and support decisions using the facts that they generate.  

I have faith in Ferrucci for a very specific reason: He is less a scientist and more an engineer. His history at IBM was one of building toward the solution rather than in service of a theory. He succeeded in building a system that leveraged existing technologies, but was built in such a way that it did the job at hand and did it exceedingly well.

There are other companies that claim to be building out general machine intelligence, but most of them are trying to prove a point. As far as I can tell, Ferrucci is trying to solve a problem.

There was an inflection point with machine learning a few years ago. We had the data, the machines and the raw compute power to learn at a level we had never seen before. We gave the machine the ability to look at the world and recognize what it saw. While the science related to this inflection point wasn’t new, the engineering was. 

In order for A.I. to make it to the next stage of development, we need a similar inflection point for the reasoning part of the equation. We need to empower the machine with a deeper ability to think about the words it hears and the situations it observes. I believe that the science is already in our hands. Now, we just need the right engineers to decide that they want to see it happen.

So join me in welcoming Elemental Cognition to the party. We’re all in this together.

As Chief Scientist and co-founder, Kris Hammond focuses on R&D at Narrative Science. His main priority is to define the future of Advanced NLG, the democratization of data rich information and how language will drive both interactive communications and access to the Internet of Things (IoT).

In addition to being Chief Scientist, Kris is a professor of Computer Science at Northwestern University. Prior to Northwestern, Kris founded the University of Chicago’s Artificial Intelligence Laboratory. His research has always been focused on artificial intelligence, machine-generated content and context-driven information systems.

Kris previously sat on a United Nations policy committee run by the United Nations Institute for Disarmament Research (UNIDIR). Kris received his PhD from Yale.

The opinions expressed in this blog are those of Kris Hammond and do not necessarily represent those of IDG Communications, Inc., its parent, subsidiary or affiliated companies.

More from this author