AI is far past the hype stage and is already making a measurable impact on the business landscape. The same can be said for big data. How significant an impact? By way of illustration, over 97% of executives recently surveyed confirmed that their organizations are investing in AI and big data initiatives of some sort. Moreover, over 70% believe that AI technology will have the most disruptive impact in 2018 – up from just 44% in 2017.
Yet impact is one thing, and value quite another. AI, with all its potential, is still hobbled by one crucial limitation vis-à-vis big data: it doesn’t know what it doesn’t know.
Because organizations still only use — liberally — at most 10% of the big data they accrue, the insights AI can produce are limited to what it knows. What about the other 90% of potential value waiting to be discovered?
What’s Broken with AI and Big Data?
The recent VentureBeat Transform conference highlighted a number of success stories of data-driven companies such as StitchFix and OpenTable. For many other companies, there’s a long way to go.
Richard Joyce, Senior Analyst at Forrester, notes that increasing data accessibility by 10% drives a surge of more than $65M additional net income for Fortune 1000 companies. Joyce warns, however, that nearly 2/3 of employees don’t trust company data. If data is inaccessible, of low quality, or incomplete, business analysts won’t put their trust in it when it comes to decision making. In addition, businesses run the risk of incorrect or misleading results.
Another problem faced at many companies occurs when employees confuse correlation with causation. When data is incomplete or dirty and, on top of that, employees attribute causation incorrectly, a chain of errors is unleashed.
In a recent interview, IBM CEO Ginni Rometty cited a study that found that on average, “A third of your decisions are really great decisions, a third are not optimal, and a third are just wrong.” In sales, for example, problems with data can cause situations where irrelevant offers are suggested to customers.
With so much hype around AI, it’s sometimes difficult to extract a realistic view of its true business potential. While much focus has revolved around chatbots and robots and singularity, it is the practical operations that simplify and improve on current systems that actually are the most useful.
What’s Working Well?
Even now, working on the limited datasets it can access, AI is effectively working with big data analytics to deliver value in several areas. Big data is the key to AI’s success, according to Jay Jacobs, Head of Research & Strategy, Global X. “Big Data is AI’s fuel. It is both what trains AI to become increasingly powerful and what AI systems are ultimately applied to in order to generate real-world insights. The more data AI systems can tap, the greater their intelligence and disruptive potential…But in order to reach these gains, there must be an easy way to ingest the data, analyze the data and prepare the fuel in a fast and cost-effective way for usage by AI and ML algorithms.”
Data Identification, Grouping, Clustering
For simple tasks like intelligent auto-classification of uploaded photos on social media and large-scale retail sites, AI-powered data identification tools are impacting efficiency. In more demanding business scenarios, AI’s ability to rapidly group events together is already in effective use for anomaly detection in cyber security, as well as fraud and money laundering detection. In retail, AI is helping online retailers take into account usage trends to more effectively monitor conversion and understand on-site behavior — helping keep engagement levels high and improve retention.
AI excels at identifying markers for better business decisions. In retail, for example, AI-based engines watch customer behavior closely. If predictive analytics decide that a customer is thinking of leaving, the system can offer discounts or other incentives to block the attrition. In logistics and delivery, AI is helping lower scratch in warehousing, and more effectively balance stock levels based on real-time consumption trends — markedly impacting numerous sectors, including manufacturing.
Autonomous Data-Driven Decision Making
In autonomous vehicles, drones, and other transportation, AI is revolutionizing decision-making. However, in its clear lack of emotional intelligence, AI still falls short. Consider AI-driven ad placement tech that decides to show ads for a law firm in an article mentioning that firm’s recent indictment, or ads praising a retailer’s commitment to social justice in an article delineating that company’s exploitation of labor in the developing world.
In an abundance of data, AI can help decide which data is relevant for which query – predicting the relevance of data to be processed. This is already enabling lower processing time and overhead, simply because less data needs to be processed.
The Bottom Line
AI and big data are making an impact. But it’s time to move from impact to value. Only once AI has access to the bigger picture – that is, the full scope of big data available for any given domain – its business value will exponentially increase.