Comb through headlines pertaining to artificial intelligence over the past 12 months and you'll see the pendulum of conversation around the machines swing from gushing optimism to doomsday scenarios and back again.
The machines will render hardship obsolete for humanity. The machines will take our jobs. The machines will extend human capabilities to their furthest reaches. The machines will enslave humans, or kill us off, or both. Elon Musk thinks they might kill us all within five years; Mark Zuckerberg wants one in his home, keeping watch over his infant daughter.
But the most extreme predictions unfold in a hypothetical, potentially distant, future when AI systems are smarter than humans. It's probably safe to say we won't be uploading our brains to the cloud or fighting off Terminators in 2016. So what can we expect?
The innovations we'll see in the year to come in AI will be largely incremental, but that doesn't mean they won't be important, say engineers and entrepreneurs. It's the difference between having Apple's Siri in your pocket and actually talking to her. We'll see tweaks to the technologies that comprise the systems being deemed artificially intelligent, and perhaps more significantly, we'll see shifts in the conversation about the technology. Here are five predictions.
1. Smarter robots.
Guru Banavar, vice president of cognitive computing at IBM Research, says to expect to see AI embedded in more robots and devices. IBM is using machine learning algorithms to train robots to better associate appropriate gestures and tones with phrases. The company's AI technology is already embedded into robots manufactured by other companies, such as SoftBank's concierge and sales associate robot, Pepper. Machine learning algorithms can help robots learn to better navigate space (think self driving cars), and can be incorporated into robotic devices such as bionic eyes.
2. Faster analysis.
A key application of machine learning algorithms is analysis of data. Advances in analysis of visual data, in addition to increasing speed of analysis, will have impacts across various fields. "We'll see artificial intelligence capabilities strengthen in the area of understanding imagery, including the context and meaning in specific elements such as objects, people, and places," writes Banavar in an email. A key area where enhanced visual analysis will have an impact is health care, a field in which, Banavar notes, human workers have to analyze high volumes of visual information. A radiologist has to interpret 16 medical images per minute, she says by way of example. Increasing speed of data analysis will lead to general improvements in business performance in 2016, according to research firm Forrester. "Machine learning will replace manual data wrangling and data governance dirty work. The freeing up of time will accelerate data strategies," writes analyst Brian Hopkins.
3. More natural interactions.
Improvements in machine learning algorithms used for language processing will make it easier to talk to computers. Virtual assistants like Cortana and Siri will become "genuinely helpful," Eric Horvitz, Microsoft Research Redmond Lab technical fellow and managing director, told Fast Company. "Really, what we really need to get over is we always kind of interact with machines on their terms using their language," Andrew Arruda, CEO of artificial intelligence attorney startup ROSS Intelligence, tells Inc. Advances in natural language processing, he says, will start reversing that relationship in the next year.
4. More nuanced fear.
Mention artificial intelligence in conversation, and it's not unlikely you'll hear someone make a reference to the Terminator film series. Tesla CEO Elon Musk referenced the action movies as a way of illustrating his fears about the worst possible outcomes of development of AI technology. "There have been movies about this, you know, like Terminator," The Guardian quoted Musk as saying in 2014. "There are some scary outcomes. And we should try to make sure the outcomes are good, not bad." Musk joined a group of tech titans in December in investing in the newly established nonprofit research organization OpenAI. With the shock value of the fears of Musk and others abating, Arruda thinks dialogue about so-called "bad AI" will start to have a little more subtlety in 2016. "I think it's moved away from 'Oh, my god, what is all of this?' and panic, to an actual conversation," he says. He also thinks it won't be too long until the presence of AI in electronics is an assumption rather than a surprise. "If it doesn't have an element of AI in it, it's going to be considered, like, kind of dumb technology."
5. Hotter competition.
In November, Google made its open-source machine learning framework Tensorflow open source. A couple weeks later, Facebook released the designs to Big Sur, the computer server that runs AI algorithms the company uses. Arruda thinks we're going to see more me-too scenarios along these lines as 2016 gets underway. Companies like Google, Facebook, Microsoft, and IBM (whose open source Watson tools Arruda's company ROSS uses) are competing with each other for reputations as leaders in artificial intelligence. "There's a battle for talent," comments Arruda. The release of open-source tools for entrepreneurs and developers isn't the only area where competition is likely to heat up. Google, Facebook, and Apple have all staked out claims in the sphere of AI virtual assistants--tools which, as Facebook's M has shown, could take a bite out of search and e-commerce.