President Obama thinks that artificial intelligence will be one of the thorny issues awaiting his successor in the White House.

The implications of recent advances in AI were a major theme at a technology conference organized by the White House and held in Pittsburgh today. Speaking on stage, Obama said self-driving Ubers, which are being tested with passengers in Pittsburgh, were a good example of the coming complexities. Although welcome and useful, vehicles that drive themselves will have some downsides.

“A huge percentage of the American population makes its living, and often a pretty good living, driving,” Obama said. “So understandably people are concerned about what this is going to mean.”

A report on the future of AI published by the White House this week listed many areas of society and the economy that will be transformed by smarter software. Although positive about the technology’s benefits, the report also discusses potential drawbacks: robotics might displace workers, for example, or algorithms used in policing might exhibit bias. The report calls for more public discussion and engagement from policy makers on problems like those.

The Pittsburgh conference mostly highlighted ways in which AI is likely to have a positive impact. Researchers talked about using machine learning to build intelligent traffic systems, improve medical diagnoses, and protect wildlife.

But panel discussions, and hallway conversations between sessions, often turned to challenges that the government will face, including job losses. Given the prominence of debates about jobs in the current presidential race, the role of AI in shaping the labor market could become a major political issue.

The White House signaled that it believes future presidents should try to shape how AI technology evolves and is deployed, but it also conceded that it’s unclear how to do that. “This is an inflection point,” said White House chief of staff Denis McDonough, who moderated a panel on the challenges in AI. “[Current progress] will either keep going or crap out if we don’t handle it correctly.”

Academic and corporate researchers at the event who work on AI technologies generally agreed that recent developments suggested big, if unknown, changes for workplaces. Economists remain unsure about the extent to which AI and automation will eradicate jobs or increase inequality (see “How Technology Is Destroying Jobs” and “Who Will Own the Robots?”).

“Every industrial age has experienced disruption; this will be the same,” said Guruduth Banavar, vice president of cognitive computing at IBM. However, Banavar and others also argued that some AI systems will enter the workforce as coworkers to humans, not replacements. Many people will collaborate with machine-learning algorithms, said Banavar: “People have to learn from the get-go that learning machines are part of their everyday.” 

Jeannette Wing, a corporate vice president at Microsoft, said that the company is already changing in response to the increased power of machine learning. She predicted that all types of companies would have to adjust, requiring big changes in the skills needed in the workforce.  “Other fields are going to have to figure out what they should be training people in,” she said.

New technologies built on artificial intelligence will also require updates to regulatory regimes. Andrew Moore, dean of the computer science school at Carnegie Mellon University, pointed to self-driving cars as an important example. He said he is concerned that companies like Tesla are moving too aggressively to commercialize automated driving. “The thing that could really hold back the technology is making mistakes,” said Moore. The government has already begun issues guidelines for self-driving vehicles and related technology, although they remain vague (see “U.S. Issues Rules of the Road for Self-Driving Vehicles”).

Subtler problems may come from social biases embedded in machine-learning algorithms and the data they are fed. Fei-Fei Li, director of Stanford University’s AI Lab, pointed out that simply searching online for photos of grandmas provides an easy demonstration of that: the top results are all white.

Li expressed hope that increasing the diversity of people in AI and computer science could help fend off such problems.

On top of thinking about all those challenges, the next president will also have to consider how to enable continued progress in artificial intelligence. Yann LeCun, director of AI for Facebook and a key figure in developing the machine-learning techniques driving recent progress in AI, said that government-funded research played a significant role in the current boom. Continued investment is needed to allow for the next big breakthroughs, he said.