Some major industries could soon be shaken up by a new development in artificial intelligence-- the technology's increasing portability. Chip manufacturers are making processors specifically for machine learning and related AI features, noted Deloitte analysts David Schatsky, Jonathan Camhi, and Aniket Dongre in a new report. In many cases, those chips are being designed to consume minimal power, making it possible for them to be used in small, portable devices. Meanwhile, AI software developers are designing new algorithms that can be run directly on such chips without ever needing to connect to computers in the cloud. The result is that we are about to enter an era that the analysts dubbed "pervasive intelligence," filled with devices and sensors that have AI technologies embedded within them.

If there's one big thing that might thwart companies' headlong rush to adopt artificial intelligence for their businesses, it's data. AI generally requires lots of data. But it needs to be the right kind of data, in very particular kinds of formats. And it often needs it to be "clean," including only the kind of information it needs and none of what it doesn't. "The biggest challenge most organizations face when they start thinking about AI is their data," said Paul Daugherty, the chief technology and innovation officer of consulting firm Accenture, in an interview earlier this month.