A.I. is a powerful tool, but it can also bring negative consequences for those who work alongside it or are affected by it. It has already cost some people their jobs and inevitably will put many more out of work. It can make society worse in a variety of ways, by perpetuating bias, spreading misinformation, and increasing economic inequality. It could even worsen climate change.
Most of the public hand-wringing about A.I. has centered on one question: Will A.I. someday take control of humans, or even cause our extinction? That concern has been raised, in particular, by Geoffrey Hinton, widely considered the father of A.I., who laid out his concerns in a 60 Minutes interview. While that's certainly an important concern, it's also one that's difficult to address in the near term because there are simply too many unknowns about how A.I. will develop in the future.
Meanwhile, there are good reasons to worry about A.I.'s effects right now. At the recent GeekWire Summit in Seattle, devoted entirely to A.I., experts from different disciplines came together to discuss both A.I.'s future and how it's being used today. While most of the conversations centered on A.I.'s potential to increase productivity, draw more value from data, and help create successful startups, panelists also flagged some of the ways A.I. can potentially harm people, not in the future, but right now. Here are some of the questions all business leaders should ask themselves about artificial intelligence.
1. What happens to people who are replaced by A.I.?
Make no mistake, this is happening already and will happen more and more over time. At the GeekWire Summit, Bridget Frey, CTO at real estate company Redfin, was asked if A.I. had affected the company's hiring, and she said that it had. In the past, she explained, Redfin had integrated with a vendor company that provided written descriptions of what it was like to live in a specific area--things someone shopping for a home might want to know. These days, she explained, a large language model is doing that writing. "It shifts the work. You no longer need to hire people to do that kind of integration," she said.
Just a few minutes later, David Shim, co-founder of startup Read.ai, explained that fears that A.I. would take away people's jobs were unfounded--even though he was on the same panel as Frey. "What we've actually found is that it doesn't take away your job," he said. "This frees up your time to work on different things. Everyone here is probably saying, 'I wish I had more time in the day.' Imagine if A.I. can actually free up 25 percent of your day."
Think about that for a second. If A.I. can replace 25 percent of someone's work, and four people are doing that job, it seems likely that one of the four won't be needed anymore. So it's wise for every company leader who embraces A.I. to create greater efficiency to also have plans and policies in place for when that A.I. displaces employees.
2. How do we keep A.I. from making some existing problems worse?
What problems? Start with bias, something that gets inadvertently built into A.I. because it's pulling in data from a deeply imperfect world. If, for example, you're using A.I. to evaluate mortgage applications, and it observes that mortgages on homes in predominantly minority neighborhoods are more likely to be denied, it may build that assumption into its own figuring.
"One of our real challenges has been with fair housing laws," Frey said. "We can't simply re-encode years of bias into algorithms and call it a day." Instead, Redfin has been working with a large language model company to build rules and cases into its large language model that will keep it compliant with fair housing laws. She's optimistic that the company can make progress on this front, she said. "But it's something that's going to affect using large language models and A.I. in any heavily regulated or ethically sensitive business."
Beyond that, A.I. can worsen the spread of misinformation and even worsen climate change, given the huge amount of computing power it requires. Proponents will argue that A.I. can help with some of these problems as well, for instance by better monitoring forests to spot wildfires before they spread. Ultimately, any responsible leader who decides to use A.I. needs to consider its negative effects and be ready to mitigate those effects.
3. Has A.I. been overhyped?
While it's clear that A.I. will have far-reaching effects on every corner of our society, it's also true that this technology is still at an early stage, and the business case for using A.I. isn't always entirely clear. This is one reason that technology research firm Gartner says generative A.I. is at the peak of what Gartner calls its "hype cycle."
It's also why presenters on a panel devoted to A.I. and business emphasized the importance of thinking through exactly how A.I. can benefit your business before deploying it. "Is this something that is augmenting our population's experience? Are they being more effective at achieving Goal A, B, or C?" asked Diego Oppenheimer, entrepreneur and partner at VC firm Factory. "A.I. excels in a lot of cases--when you control for that."
Charlotte Yarkoni, president of commerce and ecosystems at Microsoft who works with startups, concurred. "The companies that we're seeing be successful are the ones that are solving problems for customers," she said. "There is a valuable customer use case, so it starts less with 'Hey, I'm in A.I.,' and more, 'What am I trying to solve for you and your business?'"
In other words, no matter how much A.I. changes things in the coming years, for entrepreneurs one thing won't change: Your customers should always be your first consideration.
BY MINDA ZETLIN, AUTHOR OF 'CAREER SELF-CARE: FIND YOUR HAPPINESS, SUCCESS, AND FULFILLMENT AT WORK'@MINDAZETLIN
No comments:
Post a Comment