Oct 12, 2019

Training real AI with fake data

Illustration: Aïda Amer/Axios

AI systems have an endless appetite for data. For an autonomous car's camera to identify pedestrians every time — not just nearly every time — its software needs to have studied countless examples of people standing, walking and running near roads.

Yes, but: Gathering and labeling those images is expensive and time consuming, and in some cases impossible. (Imagine staging a huge car crash.) So companies are teaching AI systems with fake photos and videos, sometimes also generated by AI, that stand in for the real thing.

The big picture: A few weeks ago, I wrote about the synthetic realities that surround us. Here, the machines that we now rely on — or may soon — are also learning inside their own simulated worlds.

How it works: Software that has been fed tons of human-labeled photos and videos can deduce the shapes, colors and movements that correspond, say, to a pedestrian.

  • But there's an ever-present danger that the car will come across a person in a setting unlike any it's seen before and, disastrously, fail to recognize them.
  • That's where synthetic data can fill the gap. Computers can generate millions of scenes that an actual car might not experience, even after a million driving hours.

What's happening: Startups like Landing.ai, AI.Reverie, CVEDIA and ANYVERSE can create super-realistic scenes and objects for AI systems to learn from.

  • Nvidia and others make synthetic worlds for digital versions of robots to play in, where they can test changes or learn new tricks to help them navigate the real world.
  • And autonomous vehicle makers like Waymo build their own simulations to train or test their driving software.

Synthetic data is useful for any AI system that interacts with the world — not just cars.

  • In health care, made-up data can substitute for sensitive information about patients, mirroring characteristics of the population without revealing private details.
  • In manufacturing, "if you're doing visual inspection on smartphones, you don't have a million pictures of scratched smartphones," says Andrew Ng, founder of Landing.ai and former AI head of Google and Baidu. "If you can get something to work with just 100 or 10 images, it breaks open a lot of new applications."
  • In robotics, it's helpful to imitate hard-to-find conditions. "It's very expensive to go out and vary the lighting in the real world, and you can't vary the lighting in an outdoor scene," says Mike Skolones, director of simulation technology at Nvidia. But you can in a simulator.

"We're still in the early days," says Evan Nisselson of LDV Capital, a venture firm that invests in visual technology.

  • But, he says, synthetic data keeps getting closer to reality.
  • Generative adversarial networks — the same AI technology that drives most deepfakes — have helped vault synthetic data to new heights of realism.

Go deeper

The hidden costs of AI

Illustration: Eniola Odetunde/Axios

In the most exclusive AI conferences and journals, AI systems are judged largely on their accuracy: How well do they stack up against human-level translation or vision or speech?

Yes, but: In the messy real world, even the most accurate programs can stumble and break. Considerations that matter little in the lab, like reliability or computing and environmental costs, are huge hurdles for businesses.

Go deeperArrowOct 26, 2019

Expert Voices Live: AI in 2050

Joshua New, Senior Policy Analyst at the Center for Data Innovation, on Thursday morning. Photo: Chuck Kennedy for Axios

The big picture: On Thursday morning, Axios' Cities Correspondent Kim Hart and Emerging Technology Reporter Kaveh Waddell hosted a roundtable conversation to discuss the future of AI, with a focus on policy and innovation.

The conversation touched on how to balance innovation with necessary regulation, create and maintain trust with users, and prepare for the future of work.

The relationship between the public and private sector

As AI continues to become more sophisticated and more widely used, how to provide regulatory guardrails while still encouraging innovation was a focal point of the discussion.

  • Rep. Jerry McNerney (D-CA) stressed the importance of regulators being more informed about new technology: "How can we best use resources? We need the expertise within the government to manage these developments as they come."
  • Dr. Mona Siddiqui, Chief Data Officer at HHS, on the existing gaps at the federal level: "Investment and infrastructure is lacking. A lot of departments need the support to build that."
  • Collin Sebastian, Head of Software Products and Engineering at SoftBank Robotics America, on how the government can serve as an effective partner to the private sector: "One of the best ways the government can help without stifling innovation is to provide direction...If you give me a specific problem to address, that’s going to guide my development without having to create new legislation."

Attendees discussed balancing regulation and innovation in the context of global competition, particularly with China.

  • Rob Strayer, Deputy Assistant Secretary of State for Cyber and International Communications Policy at the State Department, on the challenges of regulation in the context of international competition in AI development: "We need to not impede growth of AI technologies and...[be] aware of a competitive international environment. Other countries won’t put [these] guardrails in."
Preparing for the future of work

The conversation also highlighted who is most impacted by technological development in AI, and the importance of future-proofing employment across all industries. As AI is something that touches all industries, the importance of centering the human experience in creating solutions was stressed at multiple points in the conversation.

  • William Carter, Deputy Director and Fellow at the Technology Policy Program at the Center for Strategic & International Studies, highlighted the importance of future-proofing systems: "Creating trust is more than regulation and mediating algorithmic risk. [People want to feel that] AI can be a part of the world in which they can participate. [We should be] creating incentives for companies to retrain workers who are displaced."
  • Molly Kinder, David Rubenstein Fellow with the Metropolitan Policy Program at the Brookings Institution, on the importance of having a clear picture of who is most at risk to be adversely affected by AI job displacement:
    • "We’re finding that...the least resilient are the ones who are least likely to be retrained. Our insights suggest that we as a country are not equipped to help working adults."
    • "Latina women are the most at-risk group for AI [job displacement]...We need to make sure we’re human-centered in developing our solutions...[and that] we update our sense of who the workers are that are most being affected."
Creating trust with users

With the accelerating development of AI, creating and maintaining trust with users, consumers, and constituents alike was central to the discussion.

  • Kristin Sharp, Senior Fellow at New America and Partner at Entangled, on how keeping people informed can create trust: "People tend to be worried about their privacy when they don’t know what the end-use case is for the data that’s being collected."
  • Lindsey Sheppard, Associate Fellow at the Center for Strategic & International Studies, on the importance of seeing AI as part of social, economic, and educational systems that also need future-proofing: "You’re not let off the hook if you’re not using AI. You need that infrastructure whether or not you’re using AI. You still need skilled workers that have those software and data skills."

Thank you SoftBank Group for sponsoring this event.

Keep ReadingArrowOct 25, 2019

Automating humans with AI

Illustration: Eniola Odetunde/Axios

Most jobs are still out of reach of robots, which lack the dexterity required on an assembly line or the social grace needed on a customer service call. But in some cases, the humans doing this work are themselves being automated as if they were machines.

What's happening: Even the most vigilant supervisor can only watch over a few workers at one time. But now, increasingly cheap AI systems can monitor every employee in a store, at a call center or on a factory floor, flagging their failures in real time and learning from their triumphs to optimize an entire workforce.

Go deeperArrowOct 12, 2019