How AI giants are losing Americans’ trust

Guests: Axios' Ryan Heath and Tina Reed.

Credits: Axios Today is produced by Niala Boodhoo, Alexandra Botti, Fonda Mwangi, Lydia McMullen-Laird and Alex Sugiura. Music is composed by Evan Viola. You can reach us at [email protected]. You can text questions, comments and story ideas to Niala as a text or voice memo to 202-918-4893.

Transcript

NIALA: Good morning! Welcome to Axios Today!

It's Thursday, August 10.

I'm Niala Boodhoo.

Here's what you need to know today: deadly wildfires rage in Hawaii.Plus, more health care workers face violence on the job.

But first, how AI giants are losing Americans' trust. That's today's One Big Thing.

NIALA: Zoom has had a lot of people worried this week over its use of consumer data, like meeting audio and video to train artificial intelligence. CEO Eric Yuan says the company's March update to its Terms of Service that started the backlash was a mistake. He said, "We had a process failure internally that we will fix."

As AI continues to expand and transform the world around us, distrust for the companies making this technology is also growing. 82% of American voters say they don't trust tech executives to regulate AI. That's according to a recent YouGov poll.

Axios' Global Technology Correspondent Ryan Heath is here with what the next stages of AI means for all of us. Hey, Ryan.

RYAN HEATH: Great to be with you again.

NIALA: Can you first just break down for us what's going on with Zoom?

RYAN: The very baseline is that Zoom had some fairly loose terms and conditions, which did in fact allow it to use your data and what's going on in the calls if it so chose to improve its own AI models. And as the company began to roll out more AI-related features in its products, that caused more and more people to wonder what was really going on with that data. And, Zoom was forced to clarify.

But of course, because we have no real way to check any of this, it leaves a lot of room for misinterpretation or confusion. And like many tech companies, when terms and conditions are set very loosely to give a company room to maneuver, that just gives the space for these debates and concerns to exist instead of there being clearer rights and clearer responsibilities being set out in the first place.

NIALA: And what would it mean to use consumer data to train AI?

RYAN: Well, it could work in any number of ways, one of the ways, for example, would be to learn from your speech patterns. To use your words, your gestures, your intonations, to improve the way transcription services or other pieces of feedback are created. And of course, we should be realistic about this.

If we want AI to work and to get better, it has to train on something. But it's a fairly standard principle now, if you want to have trust with your stakeholders and with your customers, that you should be upfront with them about how their data is being used. You should obtain consent. And in many jurisdictions, including the European Union, you're legally obliged to obtain consent when you want to use somebody's personal data. So it's not just a matter of touchy feely good customer relations, it's often a legal obligation now.

NIALA: Ryan, how do you think the situation with Zoom is illustrative of what's going on with AI, especially generative AI right now?

RYAN: Well, what we've seen in 2023 is there have been big advances in how AI operates and the ways it touches our lives. And so I think that's set off a real series of gut reactions in people where they start to worry about "what's happening with my job, is the world going to be turned upside down"? So all of that adds up to people having more and more questions about technology, which we probably most of us spent growing up trusting. We liked or adored the services that tech companies provided to us. And there's just been a slow gradual shift over a number of years in the other direction. And now that's starting to gather steam to the point where people are telling opinion posters that they actually trust the government more than tech companies now. And I don't think we've been in that situation before.

NIALA: And is this bipartisan, the distrust of tech companies? It occurs across party lines?

RYAN: That's the extraordinary thing in terms of levels of concern and levels of distrust, it's almost identical numbers between Democrats, Independents, and Republicans. And that doesn't mean that people at different ends of the political spectrum have the same concerns for the same reason, but they do have the same level of concern.

NIALA: There was a recent Swedish study that found AI-powered cancer screening could detect cancer at similar rates as two radiologists. Do you feel like this is the best-case scenario for AI, that it will help improve things like medical advances or treatment?

RYAN: I think where a lot of this is headed is where AI serves as a kind of assistant or a time saver to the medical professional. So there's still a human in the loop and if that happens the trust will be there. When you take the humans out, it's a different story.

NIALA: Axios' global technology correspondent. Thanks, Ryan.

RYAN: Thank you.

NIALA: After the break, the escalating violence in America's doctors offices and hospitals.

NIALA: Welcome back to Axios Today. I'm Niala Boodhoo. It's becoming more dangerous to work in health care in America. Axios Tina Reid reports health care workers are increasingly being assaulted or shot on the job. And U. S. health care workers suffer more non-fatal injuries from workplace violence than any other field. That's according to federal data. And now healthcare workers are speaking out. Tina's with me now. Tina, why has this issue gotten so bad?

TINA REED: This is not a new problem. I remember covering workplace violence in hospitals at least a decade ago. However, it seemed to hit a boiling point during the pandemic. People couldn't go in and see their loved ones when their loved ones were dying, there were a lot of mask mandates, there were a lot of political divisions. And what we're hearing is that, even though some of those stressors have gone away, there's still a lot of stress, and there's still violence occurring in these settings and it hasn't necessarily gone away or, or reduced since the pandemic began.

NIALA: Tina, when we say violent incidents, can you tell us what kind of things we're talking about?

TINA: The things that make headlines such as shootings are obviously a big concern, but I think when you start thinking about the day-to-day violence that happens, everything from pushing, kicking, you know, we hear about spitting, which is actually caught up in the data as well, it kind of runs the gamut, and it often happens in emergency departments because that's where people come in, they're in crisis. It also often happens in psychiatric settings and in maternal wings of hospitals as well.

NIALA: How much are worker shortages in health care factoring into this?

TINA: Speaking with the nurses union, they would say that's a big part of this. So you think about if a nurse is overworked and they can't seem to get to a patient's bedside very quickly, that patient has to wait on their pain medication, that's going to be something that causes discomfort and agitation, not only for that person, but for their loved ones.

Niala: Obviously just because someone's waiting for a long time for health care doesn't mean that they should resort to violence. So what are health care workers saying about what they're experiencing?

TINA: So the nurses union says that essentially they just need more help, because when you have enough staff. When there is a situation that is escalating, you have more help bringing it under control, making sure they have security where they need to be.

NIALA: And what are hospitals saying about this in terms of solutions that they're considering about how to try to keep their workers safe?

TINA: They're training staff on violence prevention, calling in de-escalation teams. Some are actually investing in artificial intelligence technology that can better help detect guns that are entering the buildings. One nurse that I spoke with actually talked about several years ago, metal detectors were a bargaining point when they were negotiating with their hospital because they wanted to have a metal detector at the emergency room door. And she said in the first year they actually were able to prevent 800 guns from entering the hospital. So, there are a number of different tactics that both the hospitals and the nurses unions are really trying to deploy to try and make it a little bit safer.

NIALA: Tina Reed is a healthcare editor for Axios and author of the daily Axios Vitals newsletter. Thanks, Tina.

TINA: Thank you Niala.

NIALA: Finally today – a few headlines we have our eyes on:

Wildfires fed by the winds of Hurricane Dora are raging on Hawaii's big island and Maui, and last night Maui officials said the fires had killed at least 36 people. Thousands of others are in shelters. Homes, businesses and communities have been destroyed, and tourists are being warned away. Hawaii's Lt. Gov. Sylvia Luke said officials did not anticipate that the hurricane, which did not make landfall on the islands, would cause these types of wildfires.

The Biden administration yesterday launched a new federal system for tracking heat-related illness across the country. The Department of Health and Human Services says the tracker will help state and local governments prevent heat-related illness and death by doing outreach to at-risk people, and putting cooling centers in places they're needed most. Southern states, meanwhile, continue to suffer through a long heatwave.

And in other news from the White House, President Biden has signed an executive order restricting U.S. investments in high-tech sectors in China that are deemed national security threats. Axios' Maria Curi writes that private equity and venture capital firms won't be able to invest in the riskiest transactions in advanced-tech sectors like some artificial intelligence systems, quantum computing and advanced chips. You can find a link to her story in our show notes.

That's it for us today!

I'm Niala Boodhoo - thanks for listening - stay safe and we'll see you back here tomorrow morning.