Facebook, Pinterest and other tech firms discuss AI’s future

  • Queenie Wong
  • The Mercury News
12:00 a.m. Thursday, March 8, 2018 Nation & World

Artificial intelligence might conjure up fears about a doomsday future where machines destroy humans.

But Silicon Valley engineers who work with AI have a more optimistic outlook.

“I would hope over the next few years that we see a lot more of those use cases where it’s man and machine versus man versus machine,” said Andrew Bolwell, who heads the venture capital arm of HP, at a roundtable discussion recently about AI.

Those who compete in a chess or Go game against a computer, for example, could learn new strategies and become better players.

And while there are concerns that machines will make jobs obsolete, can AI also help create new roles?

“We’ve seen through a lot of instances of automation in the past, it doesn’t necessarily take away economic opportunity,” said Airbnb VP of Engineering Mike Curtis. “It might replace a class of jobs, but it just pushes people to work in professions that are fundamentally more human.”

Pinterest Senior Vice President of Engineering Li Fan said she hopes AI will add more emotional value, helping people to become happier.

As artificial intelligence continues to play a larger role in our everyday lives, questions about its dangers continue to loom.

Tesla and SpaceX CEO Elon Musk has repeatedly sounded the alarm about the threat AI could pose to humans, while Facebook CEO Mark Zuckerberg begs to differ. This week, Google CEO Sundar Pichai called AI “more profound than electricity or fire.”

At Facebook, AI is being used to automatically translate posts to other languages, describe photos to the blind or to keep terrorist content off the platform. HP printers have vibration sensors that can help the company predict whether the product will break down. Payments company Stripe and others are also harnessing the power of AI to combat fraud.

“It’s hard to find anything that doesn’t have some element of AI as part of the experience,” Bolwell said.

There have also been concerns about human biases making their way into algorithms — an issue that’s tricky to tackle.

“The optimization space becomes very vast and as a person you’re likely to shoot yourself in the foot,” said Facebook Director of Applied Machine Learning Joaquin Candela.

People’s actions also don’t always reflect what they want, Candela noted.

Curtis said AI can help people become more aware of their own biases. On Airbnb, if a host doesn’t accept a guest then he or she is automatically prompted to block out those dates because the system assumes they are no longer available. If the host doesn’t, then he or she is asked why.

And Curtis isn’t buying into the idea that AI will destroy the world.

“I think that the (idea that) doomsday robots are going to take humanoid form is just totally ridiculous. They would definitely take a different form than that,” he quipped.