Not everyone believes that the technology we call artificial intelligence today lives up to the hype it's generated...
in the last year, and the gap between reality and hype could influence how the technology is ultimately used by enterprises.
"We still don't have real AI because we still don't know how the brain and mind work," MIT professor Josh Tenenbaum said in a panel discussion at MIT's Sloan CIO Symposium.
The panel discussed the differences between the type of applications we're calling AI today and true AI, programs that can think and learn for themselves. In general, the participants saw a wide gulf between the current state of the art and the ideal of true AI.
"The one caution I'd bring forward is to set expectations correctly," said Ryan Gariepy, co-founder and CTO at Clearpath Robotics Inc. "When you're working in your organization and exploring this technology, we've seen examples in the past where these expectations get so high and people starting buying the technology and then nothing happens."
AI has promise, but keep it in perspective
Gariepy said there is no doubt that the systems we call AI today are a vast improvement on the AI technology of just a few years ago. Clearpath makes autonomous vehicles for industrial applications like mining and warehousing. These drones couldn't function without the kind of machine learning and computer vision that today are lumped into the general AI category, according to Gariepy.
Ryan Gariepyco-founder and CTO, Clearpath
But despite this kind of progress, we're still a long way from truly autonomous robots that can think and function on their own without any kind of human supervision, Gariepy said.
"That's something we need to be careful about," he said. "There's a tremendous amount of potential in AI, but let's not say it's going to solve every problem without human intervention."
The point here is not just about defining terms. Whether the AI we see today is true intelligence or something short of that plays into how applications are used. After all, the existence of truly autonomous, intelligent systems could pave the way toward full job automation of everything from rote, routine tasks to higher-level knowledge work.
The notion that AI could automate all of our jobs has sparked debate about far-ranging topics such as political stability and the possible need for universal incomes. But panelists said people are ahead of themselves when they get into these topics because today's AI technology is not ready to put that many people out of work.
AI won't automate all jobs
Today's technology is far more likely to augment workers rather than automate their jobs.
"Since the very beginning of AI there's always been the debate about augmentation versus automation and that's very much still happening today," MIT professor Joi Ito said. "I don't think automation is an optimal answer."
Instead, he and other panelists said they believe AI will fill in for workers on the most routine tasks that demand simple pattern recognition and other basic skills. In their view, this will remove a lot of the drudge work from jobs and allow human workers to focus on the more creative and interesting aspects of their jobs.
But it's still early to even talk about this. The platforms we call AI today are finding the most success in fairly simple applications, like call centers and other customer service venues. Thinking about AI-assisted workflows for other types of jobs in areas like law, healthcare and journalism are hard because the technology is still so new and functionality is relatively limited, panelists said.
"The challenge any organization faces today is, how do you get there?" said Seth Earley, CEO at consulting group Earley Information Science Inc. "There's this vision of the future where everything is going to change. How do you get from here to there? You have to look for processes to automate but still keep people engaged."
Job losses from AI are unlikely, experts say
2017 will be the year companies wise up to AI
AI tools remain promising, but also controversial