Data is the new oil. It’s what AI systems need to create detailed analyses of past events and, more importantly, forecasts of the future. A panel of experts at DWS 2018 discussed how they collect and process data today and what features they expect from data in the future.
The idea that data is “the new oil” is virtually an accepted truth in business. In fact, there are definite parallels to be drawn between petroleum extracted from the ground and information extracted from digital systems. Oil literally fueled the machines that re-shaped the physical and economic landscape of the 20th century. Data, meanwhile, is the fuel of today’s game-changing software which provides organizations with excruciatingly detailed insights into the past, and — perhaps even more importantly — the ability to predict events in the future.
Just as oil’s value is only realized when it’s converted into gas, plastic, and other useful chemicals, data only becomes useful when it’s broken down and analyzed. Processing massive amounts of data in a meaningful way is too big a job for humans to handle alone, which is why AI systems are increasingly vital in the enterprise space. The challenges and opportunities of this new data paradigm was the subject of a panel discussion at the recent Digital Workforce Summit.
The panelists agreed that leveraging machine learning and AI to extract data value can optimize business strategies. “I think companies need to take a look at data they collect from their users to find the things that are meaningful to their customers … [T]he trends and events that they need to be paying attention to, and project that forward,” said Mike O’Rourke, the Global Head of Machine Intelligence and Data Services at NASDAQ.
The ability to forecast future events isn’t simply a handy thing to have — it’s a vital business tool which all companies need in order to remain competitive. No matter the industry, if you aren’t using data to make more informed business decisions, be assured your competitors are, panelists agreed.
“The prediction capabilities are driving most of the new products. Whether it's better forecasting or better insight, or recommendation for a piece of content — those are all predictions,” said Philip Wiser, the Senior Vice President and Chief Technology Officer at Hearst. “So we're now really focusing in on data sets that will give us some sort of competitive edge, even if it's just a question of if that Cosmo piece of content is going to perform better to a certain demographic than others. All of those edges are really important.”
Beyond the low-hanging data fruit
Data is an unavoidable byproduct of modern life. We leave a digital trail in just about every action we perform. However, not all data is created equal. The most-readily useable data is structured — that is, highly organized data that can easily fit into existing algorithms. The far larger pool of information is unstructured data, which is not readily classifiable and therefore takes additional steps to quantify. The ability to mine useable data from the vast sea of digital noise is one the biggest challenges and opportunities in the AI field.
“If you think about it, 80% of the world's data is unstructured,” O’Rourke told the DWS audience. “Most financial data is structured — it's pricing data or something like that. But when it comes to unstructured sources, like for instance social sentiment or speeches that the Fed might give, you have to translate that down to give it structure, and by that I mean turn that into a positive or negative sentiment, and then translate that to a correlation to a particular tradable instrument.”
There is a wealth of largely untapped data in all written or spoken language that is digitally collected through various means (social media, blog posts, voice UI, etc.), but the trick is finding a way to distill human communication into a format useable by algorithms. This is no easy task.
“And there's plenty of people that will say, ‘All you have to do is a sentiment analysis and you're good to go.’ But the linguists will tell you that language is a constantly changing thing … and most information creation now is not happening in English, so there's plenty of opportunity,” said Anthony Scriffignano, SVP and Chief Data Scientist at Dun & Bradstreet. “This is absolutely a place where we need to pay more attention in the data science community.”
New technologies, new frontiers
Not only are there constantly new forms of data to process — both structured and unstructured — but there are new sources for data, which companies are only now beginning to integrate into their systems. “You think about how IoT [Internet of Things] is bringing all new data sets that never existed before,” said O’Rourke. “Sensory data [for instance,] — and really that's information about those clients and how they interact with those products.”
Beyond enabling adaptive business strategies, all this data is the fuel for the coming generation of increasingly capable autonomous hardware or software. “[These new technologies all] have AI, they can modify their goals, they have to introduce themselves to each other. They're going to engage in commerce with each other,” Scriffignano said. “We don't know how to do any of that yet, so we need to pay attention to this, or it's going to get in front of us.”
- Chetan Dube, IPsoft: The ROI on AI & Amelia Solutions
- Edwin van Bommel, IPsoft: Generating ROI for Your Business Through IPsoft’s Marketplace
- Max Tegmark, MIT: The Acceleration of AI Is Transforming Business Practices
- Christopher Manning, Stanford University: Inside Amelia’s Brain
- Anthony Abbattista, Deloitte: The True Impact of AI
- Panel Discussion: Cognitive AI in Healthcare Means Augmented Patient and Caregiver Experiences
- Panel Discussion: Lessons Learned from AI Pioneers
- Panel Discussion: How IPsoft’s Clients Innovate and Improve Business with Amelia
- Panel Discussion: AI Deployment Surprises and Strategy