ScarJo vs. ChatGPT + Neuralink’s First Patient Opens Up + Microsoft’s A.I. PCs
01:17:00 Share

This podcast is supported by Search Engine from PJ Vogt. Search Engine was named one of the very best new podcasts of 2023 by The Economist, Vogue, and New York Magazine. Search Engine answers fascinating questions about technology and everything else. Questions like, why are drug dealers putting fentanyl in everything? How do

How did ADHD medications get so popular so fast? What do trigger warnings actually do? Listen to and follow Search Engine with PJ Vogt, available on the free Odyssey app or wherever you get your podcasts. I had an interesting experience recently. I went in a float tank for the first time. Have you ever been in one of these? I have seen them in sci-fi movies. Was this like a bacta tank from Star Wars where you're healed from your injuries? Yeah.

No, this is like a trendy new thing in the Bay Area where you basically go into these pods. Imagine if Apple designed a coffin. It's like a shiny white pod the size of your body. And you go in and it's filled with a couple inches of very salty water. So you basically just lie there and you float for an hour inside this pod. And it's supposed to be relaxing. Yeah.

And was it relaxing? Sort of. Sort of. I mean, it's definitely relaxing to be like floating gently in some like warm, salty water. But it is a little claustrophobic because you can make it totally dark in there and do sort of sensory deprivation. But the thing that actually made it less relaxing for me was that you, they have music that you can choose from, but the selection is not good.

It's like cheesy yoga music. It's like pan flute. It's like, you know, chimes. It's not what I wanted to be listening to down there. Which was Espresso by Sabrina Carpenter. That's what I want to be listening to. Kevin, let me ask you this. Have you ever heard of taking a bath? Because you're able to get many of the benefits of driving somewhere to float in two inches of water. And you can choose your own music.

Might want to look into that. That's true. That's true. You know, you and I should go into a pod together and just start talking about tech news for an hour. You know what they call that? What's that? Podcast. Okay.

I'm Kevin Roos. I'm a tech columnist at The New York Times. I'm Casey Newton from Platformer. And this is Hard Fork. This week, OpenAI wanted Scarlett Johansson to be the voice of ChatGPT, but then something got lost in translation. Then, Nolan Arbaugh, the first person to get Elon Musk's Neuralink implanted in his brain, joins us to talk about how a brain-computer interface changed his life. And finally, The Times' Karen Weiss joins to tell us about Microsoft's new AI computers and his plans to record every single thing you do on your device.

Well, Kevin, just as it seemed like things were starting to settle after some wild AI demos last week, a shocking statement from one of the world's most popular actresses has made us reconsider everything OpenAI has been telling us about its voice assistant. Yeah, this is one of the craziest tech stories of the year. I've been totally obsessed with every twist and turn. I'm very excited to talk with you about it today. Now, did you ever think we would have a literal Avenger fighting back against the relentless march of AI? No.

Because that's sort of what this story is about. So last week, we talked here about the announcement from OpenAI about their new GPT-4.0 model, which was most striking for this very flirty voice assistant that they used in the demos they showed us. Kevin, remind us what was so striking about that demo. So,

The voice that they demoed, it was this sort of lilting female voice. It was a little flirty, as you said. It sort of varied its register. It kind of giggled at its own jokes. It was very lifelike and realistic. And basically, immediately as this demo is going out, people start making comparisons to the movie Her and to Scarlett Johansson's character in that movie, Samantha. And...

Like the company itself made that comparison, invited that comparison. Yes, Kevin. And people are actually calling this the greatest act of cultural appropriation since Scarlett Johansson was cast in Ghost in the Shell. You went there. That's right. So on Sunday, OpenAI posed to its website this mysterious blog post titled How the Voices for Chat GPT Were Chosen.

And in the blog post, it says, quote, we believe that AI voices should not deliberately mimic a celebrity's distinctive voice. Sky's voice is not an imitation of Scarlett Johansson, but belongs to a different professional actress using her own natural speaking voice. Kevin, when you saw that blog post go up, did you have any idea what was going on? No, but it was one of those things where it's like my Sky is not an imitation of Scarlett Johansson t-shirt. Yeah.

is raising a lot of questions already answered by my sky is not based on Scarlett Johansson t-shirt. It was like, if you were saying this, clearly something is happening in the background. You did not just decide to come out of this from nowhere. And it was just a sign that things were going to get a little weird. Yeah, absolutely. You know, this blog post went up in,

very late Pacific time on Sunday. And to me, it was a sign that this was going to be a rough night, which is the title of a 2017 film starring Scarlett Johansson. Okay, so on Monday morning, things start to become a little more clear when OpenAI pulls Sky's voice from the app

And Joanne Jang, who is the model behavior lead at OpenAI, talks to The Verge and says, quote, we've been in conversation with ScarJo's team. Pretty familiar there, Joanne. Because there seems to be some confusion. We want to take the feedback seriously and hear out the concerns. And she further suggested that maybe people hear similarities because there are so few convincing female voice assistants around. Does that seem convincing to you, Kevin?

No, of course not. Because what... Yeah, well, you know, to me it just seemed like Scarlett was really trying to get under the skin of OpenAI, which, you know, under the skin is a 2013 film starring Scarlett Johansson. Oh, boy. All right, so on Monday night, Scarlett Johansson herself releases a statement, and this is the doozy, right? Yes. This is sort of what makes the world stop, and she really lays out quite a narrative, which I think we should walk through. Yes. So...

According to Scarlett, which is what I call her, Sam Altman had approached her in September 2023 about hiring her to voice ChatGPT, saying that it would be good for everyone to see tech and creatives working together. And Kevin, you will remember that September was when they rolled out voices in ChatGPT in the app, right? So around that same time, it seems Sam has this idea. And, you know, to me, this seems...

I can tell you're already annoyed by this bit. No, it's good. Keep going. What Scarlett Johansson movie have you not mentioned so far? Well, I mean, to me, it just seems clear that when Sam approached her, he wanted the prestige of having her voice in the app. The prestige being a 2006 film starring Scarlett Johansson. All right. So did you do her interview?

Absolutely not. Absolutely not. So at the time, she declines for whatever reason. And then she writes in her statement, nine months later, presumably referring to last week, my friends, family and the general public all noted how much the newest system named Sky sounded like me.

And how did she feel about it? Well, she said, quote, when I heard the release demo, I was shocked, angered, and in disbelief that Mr. Altman would pursue a voice that sounded so eerily similar to mine that my closest friends and news outlets could not tell the difference. Mr. Altman even insinuated that the similarity was intentional, tweeting a single word, her. And this really threw me for a loop.

She says that two days before the demo, Altman had reached out to her agent asking her to reconsider, but then OpenAI rolled out the demos with Sky before she could respond. So isn't that a wild detail? It's a crazy detail. I mean, like, yeah, we can talk more about what it means later, but this is the part where I'm just like,

oh, they screwed this whole thing up so badly. Yeah, and also, like, if there was a chance that maybe you could work it out, why wouldn't you, you know, wait for that? And I think the answer to that, by the way, is that they wanted to sort of upstage Google before its own developer conference. But just to sort of finish out this statement, she says that she and her lawyer sent two letters to Altman and OpenAI asking for a detailed accounting of the

process that created the voice. And I think that is probably what led to the blog post that went up on Sunday night. And she closed with a call to action. And I don't know if we want to play the star-spangled banner underneath this. I do think it would sound nice. I'll just sort of read the quote. In a time when we are all grappling with deepfakes and the protection of our own likeness, our own work, our own identities, I believe these are questions that deserve absolute clarity. I

I look forward to resolution in the form of transparency and the passage of appropriate legislation to help ensure that individual rights are protected. So after that, Kevin, in the middle of the week, OpenAI puts out a statement which they attribute to Sam Altman, and it says...

So that was the statement in the middle of the week. And I have spent the last few days, Kevin, trying to figure out what in the Vicky Cristina Barcelona is going on here.

And what did you find? Well, I found, first of all, that is a Scarlett Johansson movie. Yeah, I got that. All right, all right. So on Thursday, I had a chance to ask OpenAI some questions. And my first question was, who exactly at this company knew what the heck was going on? Okay. And what I was told was this.

The voice team decided they wanted to record five voices for ChatGBT. But after that, they decided, hey, it would be cool if we could get Scarlett Johansson.

And as part of that, Sam Altman was sent out on a mission to get Scarlett Johansson. And according to them, that is when he reached out to her in September. And to sort of bolster this timeline, they did a couple things. They showed me a job posting from May of last year where they advertised for actors for these roles. And I saw the job posting. It did not mention Scarlett Johansson. It did not mention her or any other movies. But

They played... Did it not say only Black Widows may apply? No, it didn't say that. Okay, okay. They then played for me a clip from Skye's audition where she talks about, you know, walking around with her toddler and basically just gives you the impression of, no, this is a real voice. This is not a composite of other people's voices, which is like one conspiracy theory that was sort of floating around this week. And then finally, they showed me a video clip of

of the actor in the recording booth while they were doing this recording. Now, this video clip was very short. It was heavily pixelated and it was taken from so far away that I couldn't even tell where the human was supposed to be until the second time I watched it. So I wouldn't say that that clip alone is giving me a lot of confidence in the narrative here, but I have seen some sort of video suggesting that at some point a human being was saying something into a microphone.

Okay, so let me just repeat all this back to you and you tell me if I have the timeline and the version of events right. So OpenAI is saying that they did not initially plan to have a voice of Scarlett Johansson or even one inspired by Scarlett Johansson as part of this chat GPT voice release, but that

they later sort of came up with the idea, well, maybe we should have the sixth voice and maybe if Scarlett Johansson will say yes, then we can get her in as the sixth voice. That obviously never happened, but they are basically saying this is all, this was all never intended to mimic the voice of Scarlett Johansson, any resemblance to, uh,

people living or dead named Scarlett Johansson is purely coincidental. Is that basically what they are telling you? That is what they're telling me. And how do you feel about that narrative? So, yeah, I guess I buy the narrow version of events that OpenAI is

claiming happened here. And, you know, I also have listened to clips of Sky and listened to clips of Scarlett Johansson, and they don't sound totally identical to me. So it is totally plausible that they had this other voice actor play this role. But there are still two things that aren't quite adding up for me. One of them is like,

Okay, say you didn't, you know, cast a Scarlett Johansson sound alike. Why then spend so much time around the launch of this new voice feature sort of making people feel like they were listening to Samantha from her to sort of directly connect the release of this product to this movie and this actress? Why do that?

if you know it's going to get you in trouble. And then the second thing is, OpenAI itself has said in the past that they do not want their synthetic voices to sort of mimic public figures. In fact, there was actually a statement that they put out on March 29th earlier this year in a blog post that OpenAI wrote called "Navigating the Challenges and Opportunities of Synthetic Voices."

And one of the things they say in this blog post is that there should be a, quote, no-go voice list that detects and prevents the creation of voices that are too similar to prominent figures. So I'll be very interested if there is litigation around this issue. If in any of the discovery, they find evidence that OpenAI employees were sort of talking about how similar this voice sounded to Scarlett Johansson, whether or not that violated their own OpenAI policy about not creating synthetic voices that were too close to the voices of

prominent people. So let's talk about why this matters, because I can understand you might be listening and saying, this seems like kind of a small thing, right? It's just a voice. Hey, if super intelligence is coming soon, is a voice really what we should be worried about? But I think it's important for a couple reasons.

And the first one, Kevin, is that the creative community is already deeply skeptical of AI, right? Last year, we had the SAG-AFTRA strike, and this was a core plank of the fear there, which is that they worried that companies would steal an actor's voice and image, use it without their

permission and eventually either drive them out of a job or just drive their wages way down. So during that strike, actors were able to win concessions on this point. And now here you have a different case where Scarlett Johansson wakes up one day after saying no to this company, and now this sound-alike voice is the voice of ChachiBT. Yeah. So, Kevin, first of all, is what OpenAI did here legal? Yeah.

It may be, it may not be. And it's a little hard to tell, but I can imagine that there's going to be some litigation or maybe a settlement. I mean, Scarlett Johansson has said that she has lawyered up and they're actually two legal cases that people are sort of using to say about this case involving Scarlett Johansson and open AI, that actually the facts would be highly in her favor if she did decide to litigate. Do you want to tell us about one? So one of them is a, uh,

a case from 1988 called Waits v. Frito-Lay. This is a case, the Frito-Lay Corporation, makers of fine snacks, decided that they wanted to make a commercial for a new flavor of Doritos.

And they really wanted Tom Waits to sing in it. But Tom Waits is sort of famously anti-commercial. Like he just didn't want to have his songs used to endorse products. It was like against his values. So they went out and they paid $300 to a Tom Waits impersonator. Basically, a guy who's in a band who sounds exactly like Tom Waits. And they put the impersonator in the Doritos commercial.

They really said, Waits, Waits, don't tell me that you don't want to be in this commercial. Yes. So then the commercial comes out. Tom Waits' friends start calling him and saying, hey, I thought you were against commercials. Why are you all of a sudden endorsing Doritos? He gets mad. He sues. And he sues not for copyright violations, because you can't copyright a sort of way of singing or talking, but for false endorsement.

And a jury awards him $2.6 million. He's basically the heir to the Cool Ranch Dorito fortune at this point. Exactly. And congratulations to him. What's the other case? The other one is around the same time, Bette Midler had a very similar thing happen to her involving the Ford Motor Company, which went to her and said, hey, could we use your song, your voice in a commercial for our new Mercury Sable? Yeah, be the wind beneath our wings, Bette Midler. Exactly. So she says, no, I don't really want to endorse products.

So instead they go and they hire one of her former backup singers and basically instruct this backup singer to sound as close to Bette Midler as possible. She takes this to court. She wins $400,000 in damages. So I think Scarlett Johansson has not...

I would never say an airtight case here because there's no such thing, but I think she has a very strong case here. But even setting aside the legality of it, Kevin, I'm curious to get your thought on what this does for public perception, right? Where are we right now on how average people are thinking about AI, what role it might play in their life, and whether it might threaten them in some way? I mean, I think this is the most...

sort of damaging thing to come out of this particular episode. It's not actually, you know, I'm sure they will, you know, they will figure out a way to sort of make things right with Scarlett Johansson or they'll go to court. But I think the broader damage here is to the public trust in OpenAI. This is a company that has said, you know, we are building something that will eventually become

an artificial general intelligence. We are doing this for the good of humanity and we want you to trust us on that. And I think they got the benefit of the doubt for a while because they were releasing things that were cool and useful to people. ChatGPT, you know, was a moment where a lot of people said, okay, maybe they are, you know,

at the sort of head of the pack here and maybe we're okay with that. Then I think we saw things start to decay a little bit with each successive release and the sort of overall vibe being that this was actually not sort of a nonprofit research lab as it had been started, but was actually something more like a very traditional tech company

And so I just think we've seen a gradual erosion of that trust from the public in open AI. And I do ultimately think that hurts them long term. I agree with you. I think this has been a really bad month for the perception of tech amongst average people. I think this is a moment where we have seen tech companies get really greedy and greedy at the expense of working people. And so like,

As May is coming to a close, on one side of the ledger, you have Scarlett Johansson and an entire creative class of workers rallying around her. And on the other hand, you have OpenAI sound-alike voice, Google AI overviews eating the web, and the Apple hydraulic press from the commercial crushing everyone into a fine pulp. So I think the tech industry needs a better story to tell here than we're coming for your voice and there's nothing you can do about it.

Totally. So there's a second thing that I want to talk about, though, Kevin, which is the implications of this story for OpenAI, because I think it recontextualizes one of last year's biggest stories, which was Sam Altman temporarily getting bounced out of the company. So can you just remind us what happened in November to Sam?

Yeah, so he was fired in a surprise move by a number of members of the nonprofit board that governs OpenAI, along with Ilya Sutskovor, who was the chief scientist at the time. And by way of explaining why they were firing him, they made these kind of vague statements about how Sam had not been, quote, consistently candid.

and just basically implied that he was sort of a slippery person who was telling different things to different people and who they had sort of lost faith in. But they never sort of gave...

many concrete examples of that. And so I think it was tough for people to understand, like, why make such a sudden and important decision in sort of the dead of night without consulting anyone. And so Sam sort of had the trust and the faith of OpenAI's employees. And so they rallied around him. They all...

Remember, they were going to briefly go, I'll go work at Microsoft. And then the board members end up being sort of pushed off the board. And Sam is brought back as the CEO. It's, you know, it's a wild story. But I think a very interesting thought experiment for me over the past week has been, what if the board coup had happened now?

What if the board had waited to make its move on Sam until now when I think it's fair to say he would not get the same benefit of the doubt from the employees or the investors in OpenAI that he has today? I think that's true. Now, this is a company that is valued at, what, $80 to $90 billion. I think the employees who are working there want to see the equity that they have in that company realized. And I think that there's a very good chance that if what you just laid out happened, the

those employees would still support Sam. That said, you're right. We would have a good example of why this board might be a little bit concerned. I know, at least for me, ever since we did our own reporting, we talked to people involved in that situation, I have thought since November, there might be another shoe to drop here. We may eventually learn what the board was so concerned about. And I

feel like this week for the first time, we actually know now. Like it is this sort of thing. - Yeah, but I actually, as strange as it's gonna sound, I don't think this Scarlett Johansson voice thing is actually the worst thing that has happened to OpenAI over the past couple of weeks. - What do you think it was? - So there was this other story that came out just recently about these employee agreements at OpenAI.

And this came to light after Ilya Satskovor and Jan Leakey, who are the heads of the company's super alignment team, both announced they were leaving the company. And, you know, people started looking into the paperwork that OpenAI employees have to sign when they leave the company. And recently, Kelsey Piper at Vox reported that there was a really unusual provision in this exit paperwork that basically said that if...

OpenAI people left the company and then spilled the beans or said something or disclosed something about the company or disparaged it in any way publicly, they could not only break this contract, but they could have their vested equity clawed back, which we should explain why that's such a big deal. So normally, you go work at a tech company, you get stock options, some of those stock options vest over time, and this is traditionally how tech employees make a lot of money. Their stock options vest, they sell them, they get the money.

So it is not unusual when you leave a tech company to have your unvested equity forfeited. What is extremely unusual, and actually I've never heard of this happening before in the tech industry, is for a company to say, we can actually take back your vested equity if you've left the company and you disparage us publicly. So this was something that a lot of OpenAI former employees had been terrified of. It's a reason why we haven't seen a lot of former OpenAI employees speaking out.

And when it became public, a lot of people in the AI industry said, this is crazy. We have not seen this at any other companies. They are trying to silence former employees from speaking out. And then you saw actually Sam Altman make this statement about it, where he basically said, I didn't know about this. I didn't know this was part of our paperwork. I've been trying to get Scarlett Johansson to be the voice of ChatGPT. You think I have time for this? So he said, we've never clawed back anyone's vested equity.

Basically, he said, this is one of the few times I've been genuinely embarrassed running OpenAI. I did not know this was happening, and I should have, and we'll fix it. So the reason I think this is actually a bigger deal than the Scarlett Johansson thing, despite it not getting nearly as much attention, is because OpenAI is in a talent war.

They are constantly trying to pick off the best AI researchers from all of the biggest companies in Silicon Valley. It's a very talent-heavy business and talent-dependent business. And if they start losing people who say, this company seems like they're slippery and I can't trust them, I think that is an existential threat to them in the long term.

Well, look, if you're wondering why you've never heard me give an interview where I talk about what it's like to work with Kevin, I have signed something very similar. But you're right. It is very unusual, particularly for a company with Open in its name. And I agree, this was another black eye for the company this week. It also, of course, came on the heels of their entire super alignment team being dissolved, which we discussed last week. So there's just kind of a lot of swirling drama around that company.

Now, we should also say, while all that's going on, OpenAI's business is doing great. I don't want to pretend that it's not. Altman was on stage at a Microsoft Developers Conference this week, which we'll talk about in a little bit here. There has been some reporting that Apple and OpenAI are going to announce a big partnership next month at Apple's own developer conference. And finally, there was reporting this week that on the day that the GPT-4.0 model was announced,

OpenAI's revenue shot up more than 20% according to third-party estimates. So clearly OpenAI is doing great, but the sorts of things that we've seen this week have given me some pause, and I wonder if they've given you some pause as you think of what is the future of this company. Yes, I think the correct way to phrase what I've been feeling this week would be vibe shift. I think there's been a big vibe shift around

ai when it comes to the creative community but especially with open ai as it relates to the the sort of trustworthiness of what they're building um and you know i've talked to people who say you know i basically gave this company the benefit of the doubt i gave sam the benefit of the doubt um they seem to be saying a lot of the right things and now they're just kind of like i don't know man and i and i also think it's like i was thinking about this sort of idea

of like the sort of Silicon Valley builder's mindset of like, you know, ask for forgiveness, not permission, right? And I think that's been the way that a lot of successful companies have been built in Silicon Valley. Uber, you know, Facebook to some extent was a story of asking for forgiveness, not permission. And I think that that works with most technologies, but I think with AI, it's a little bit different. Yeah, absolutely. It's also like he asked for permission

not for forgiveness. He did it backwards. He asked for permission, didn't get permission, and then asked for forgiveness after he didn't get permission, which is not something that he wants to do. But no, he didn't even ask for forgiveness. He said, I would never ask for forgiveness because there was nothing to apologize for in the first place because this voice isn't based on Scarlett Johansson. It is the wildest sort of, you know, humbug.

hardest to pin down narrative. It is. I mean, look, the thing that this brings to mind for me, Kevin, is that you and I both covered the decline in public perception around Facebook, right? And Facebook once seemed like this silly little toy. Nobody paid too much attention. Then after the 2016 election, everybody is like, wait, is this secretly a mind control device that is, you know, making all of our teenagers insane and everything else? And

Obviously, it's way too soon to say that something like that is happening to OpenAI, but I'm telling you, this is how it starts, right? I think more and more people are becoming convinced every day that whatever AI is going to be in the short to medium term is not going to be a good bargain for them, and they're not going to give OpenAI the benefit of the doubt. And that means that OpenAI, I think, needs to be really careful in how it makes its next several decisions around this kind of stuff.

So look, I'm glad the Hollywood A-list is paying attention. If you are worried about what a company like OpenAI might do with your voice or your job, you're in good company, which is a 2004 movie starring Scarlett Johansson. Oh, good. I was worried there was one Scarlett Johansson movie we weren't going to mention. When we come back, Neuralink's first ever patient joins us to talk about how the technology is changing his life.

This podcast is supported by Search Engine from PJ Vogt. Search Engine was named one of the very best new podcasts of 2023 by The Economist, Vogue, and New York Magazine. Search Engine answers fascinating questions about technology and everything else.

Questions like, why are drug dealers putting fentanyl in everything? How did ADHD medications get so popular so fast? What do trigger warnings actually do? Listen to and follow Search Engine with PJ Vogt, available on the free Odyssey app or wherever you get your podcasts. I'm Julian Barnes. I'm an intelligence reporter at The New York Times. I try to find out what the U.S. government is keeping secret.

Governments keep secrets for all kinds of reasons. They might be embarrassed by the information. They might think the public can't understand it. But we at The New York Times think that democracy works best when the public is informed.

It takes a lot of time to find people willing to talk about those secrets. Many people with information have a certain agenda or have a certain angle, and that's why it requires talking to a lot of people to make sure that we're not misled and that we give a complete story to our readers. If The New York Times was not reporting these stories, some of them might never come to light. If you want to support this kind of work, you can do that by subscribing to The New York Times.

So Casey, as you know, one of the technologies that is fascinating to me right now is the brain-computer interface, or BCI. That's right. In fact, it's so fascinating to me that I forced you to try one a few weeks ago, and I think it's safe to say it was not all that impressive. No, I was willing to go through with it because I've always wanted to see if we could detect any brain activity for you, but...

Afterwards, I thought, I can't do this anymore. So just to back up, brain-computer interfaces are a type of technology that allows you to basically control a computer directly with your brain. And this has been something that obviously science fiction has been talking about and that scientists have been working on for decades. The company that is perhaps best known in this space is Neuralink, Elon Musk's brain-computer interface company. And

I've just been fascinated by this whole area. I made you try one. It was what's called a non-invasive brain-computer interface, which means it's not inside your literal head. It's like a headband that you wear. And it doesn't work that well. The technology was not all that impressive. We didn't end up airing the segment where we tried this thing on because it just wasn't very good. But...

With Neuralink, the implant, the brain-computer interface, goes literally inside your skull, on your brain, and it allows you to control a cursor with your mind. So I think we should tell people a little bit kind of about what this thing is and what it looks like because there are these

that have electrodes on them that penetrate into the brain, and those electrodes read signals, which then get translated through the Neuralink device, and that's where I've lost the plot. So can you pick it up from there? Yeah, so it basically translates your electrical activity in your brain into commands to control something on the outside of your body, like a computer or something like that.

And for years now, tech companies have been looking at using BCIs to help people who have debilitating conditions like a spinal cord injury or a stroke or some limitation on their mobility. But also a lot of people in Silicon Valley just talk about this as a potential next step in computing altogether, like that in the future, you know,

some or all of us will have these kind of brain computer interfaces. And I've even heard people in the AI world say, this is the way we are going to stay ahead of the robots as the AIs get smarter, is that we are going to implant computers in our head that will basically increase our own cognitive capacity. Now, that technology does not exist.

It is mostly, you know, just an idea from science fiction. We have no idea whether that would work or not. But this is a big, important technology that a lot of people in tech are excited about. And just over the last few months, we've actually seen one of the clearest views yet into how this might work in a real human.

Yeah, and while you've just spun out some really fantastical sci-fi scenarios, Kevin, what appeals to me about this story is that it is a case of technology helping one person who had something really terrible happen to him. So I want to say up front, very few people have been as critical of Elon Musk as I think the two of us have been on this show in particular. And...

As hard for me as it is to set aside my personal feelings, particularly about what he did to Twitter, I truly am so inspired by how this technology is helping one person. And I think it is absolutely worth understanding what is this thing that the

that they built and how has it changed the life of at least one individual? Yeah. So today we have a really special opportunity to talk to the one person on earth who has actually gotten the Neuralink brain computer interface implanted in his skull. This man is named Nolan Arbaugh, and he now has a computer system about the size of a

coin and a bunch of threads with electrodes that connect to his brain that allow him to do things like move a cursor around on a computer with just his mind. And this is a big deal for Nolan because for the last eight years, he's been paralyzed from the shoulders down. He had a freak accident eight years ago where he suffered a severe spinal cord injury. And so he has been a quadriplegic for the last eight years.

And he volunteered to be patient number one in this Neuralink experience. So today we're going to talk to Nolan about what having this Neuralink device in his brain has been like, how it's changed his life, and why he volunteered to risk his body on this unproven new technology. Let's bring him in. Nolan Arbaugh, welcome to Hardfork. Nice to be here. Thanks for having me. Yeah, where are we catching you right now? Just describe where you are.

I'm in my house, in my bed. So if you hear any weird noises in the background, it's my bed. I have an air mattress, so it's kind of blowing air through the whole thing all the time. Got it. Where's home for you? Yeah, I'm in Yuma, Arizona. Cool.

So it's been a crazy last few weeks for you. Back in January, Elon Musk announced that the first human patient had been successfully implanted with a Neuralink device, but he didn't say the name. It's only really in the last two months that your name has become public, and just the last week or so that you've started to talk more broadly about your experience. What has it been like to be sort of the literal face of this technology? Yeah.

Yeah, it's been all right. I'm just here trying to get all the information out to as many people as possible. I think it's an amazing technology. I think what's going on in my life and what I foresee the future will hold is worth bringing the whole world along with. So it's been cool, man. I'm enjoying it so far.

Yeah, I just want to go back to sort of before you got this Neuralink device implanted in your head. What compelled you to register to participate in this extremely new and untested experiment?

Yeah, I mean, I didn't really know much about it. My buddy called me up one day and kind of gave me the five minute rundown. I wasn't expecting anything to come from it. So I like made a few jokes on my application. You know, I just I figured I would never hear back. And then once I did start hearing back, I had to think about it a little bit more seriously, had very serious, candid conversations with my parents, my friends, and everything.

Ultimately, when it came down to it and I was selected, I decided that I just wanted to help. I knew that I wanted to help make it safer for everyone after me. And I knew that I wanted to

you know, try to make a difference in the world and something I've been trying to do, something I've been looking for for eight years. And this seemed like the perfect opportunity. Do you remember any of the jokes that you put on your application? Yeah. Yeah. I said, I wanted an Ironman suit. I said that I wouldn't mind being uploaded into the matrix, you know, just things like that. Right on. Yeah.

One of the truisms that Casey and I have learned in reporting about tech over the past decade is that you never want to try version 1.0, right? It's risky to try the first version of anything because the bugs are still being worked out. Usually there's some rough edges.

But you were literally being asked to try version 1.0, not just of a new gadget, but of something that is going to go inside your skull. So was that part of your process of thinking through this? Like no one has ever had one of these put in their brains before. Maybe I want to let someone else be the first person. Yeah, it crossed my mind. Something that my buddy and I, the buddy that called me on the phone, we talked about at length was

you know, this is the worst version of it that's ever going to be in a human. Maybe someone else should go first and I'll get a better version later on down the road. Or maybe I don't do it at all and wait for it to be on the market to the public. And then I'd get an even better version. But ultimately I figured that if anyone's going to do it, then I should. I have a pretty solid foundation with my faith in God. And I just felt like

I've thought my entire accident. I'm glad that it happened to me and no one else I know because it's just a very hard thing to experience, being a quadriplegic. And I wouldn't ever want any of my friends to have to go through this. So it's just the mindset that I've had forever. And with the Neuralink, it was the same thing. If anything were to go wrong, I would feel terrible if I passed up to wait for a better version and something went wrong to someone else. So I knew that it had to be me. Yeah.

Your parents are your primary caregivers. What were your conversations with them about doing this?

Yeah, they were really hard for a while. I mean, as a quadriplegic, the only real thing that I have left is my mind, is my brain, is my personality. And it's hard to let someone go in and kind of rummage around up there, especially with something that's never been tested in a human. So one of the things that I mentioned to them was,

that if I had any sort of brain deficiency afterward, if I was mentally handicapped in any way, that I didn't want them to take care of me anymore, that I wanted them to put me in some sort of home. Because taking care of a quadriplegic is hard enough, but taking care of a quadriplegic with a traumatic brain injury is something that I would never want my parents to do. So I made them agree to that. Were you nervous the night before the operation to install your Neuralink device? What did you find yourself thinking about as you went to bed?

Yeah, I wasn't nervous at all. I was just excited. My buddy and I were sitting around making jokes. I don't know, hanging out the night before. I just wanted to get it over with, honestly. What kind of jokes were they making? We were planning on releasing some cyborg-related jokes, thinking of things that only I would be allowed to say. You know, just random turns of phrases, too. Like, oh, blew my mind, picking my brain. Things like that.

Yeah, just all sorts. I'm curious if you had a thought as you were heading into this surgery of what the first things you wanted to do would be once you had a working brain-computer interface.

Yeah, I mean, I'm a big gamer. I wanted to play games. That was one of the big things I wanted to do. I also wanted to be able to read. I mean, there's nothing like being able to hold a book again and the smell and the feel of a paperback book. It's one of my favorite things in the world and something that I've missed for a long time. I can't do that. Um,

And so the next best thing is just being able to read in general. I've had to listen to audiobooks for the last eight years because I haven't been able to sit in the same position to read a book. And

I had no way of like turning pages. I can do it sort of on a Kindle, but I was listening to audio books and I don't really like listening to audio books, to be honest. Sometimes the narration is terrible. I don't want to throw shade, but I remember reading like an Aragon audio book and the voice that the narrator had for the dragon made me turn it off immediately. I was like, this is, this is quite possibly the worst thing I've ever heard in my life. It was Gilbert Gottfried voicing the dragon. Yeah.

No, that would have been amazing. I would have listened to that on repeat, honestly. No, but just something as simple as that, being able to like, you know, lie in my bed and read a book. Like there's just, there's just something about it. So I was really looking forward to that kind of stuff.

So how was the actual surgery? Was it long? What was the recovery process like? Can you tell us just about the actual implant? Yeah, it was super, super quick. We got to the hospital at like 5 a.m. I think I was scheduled for surgery at 7. There was a lot of just getting me ready, getting me in bed. The surgery was supposed to last between like 4 to 6 hours.

They were expecting there to be hang-ups. They were expecting, you know, say like the needle on the robot to break. They brought, I think, 20 iterations of that needle in case it broke and then they had to stop and replace it. And the needle didn't break once. Just everything performed above and beyond what they expected. And so the surgery took under two hours. And then I was...

out of surgery and they prescribed some like pain pills. I didn't take a single one. I don't know. It was just so easy. The worst part was I wasn't able to shower for the first few days because my incision needed to heal. But outside of that, like the recovery process was so easy. Like I didn't feel any pain at all.

Wow. And when you woke up from your surgery, like what was actually different? Did, did you feel different? Um, and, and then sort of how long was it before you actually got to like turn on and use the Neuralink device itself?

Yeah, I mean, I had a gnarly scar with some staples in it. I mean, that was pretty freaking sick. Like I was, I was super happy about that. I got some cool pictures. But yeah, I think within, I don't know, like an hour or two after my surgery, they came in and connected me to like a little tablet they had.

What do you mean connected you? Like literally plug something into your head? No, no. Is it Bluetooth? It's a Bluetooth connection, yeah. So they just wake up the implant with a little coil, like a charging coil, almost like the same thing that you put your phone down on a mat to charge. It's very similar. You just hold something over my head and it wakes up. That's how you charge it as well. I put that in like a hat and I wear a hat and it charges. Okay.

And so they woke it up. They connected that to a tablet. And on the screen, they just showed a bunch of the channels. The channels are each electrode in my brain. And those electrodes are picking up like neuron firing and stuff.

So they showed me say like eight channels and I got to see like live, like real time, the neurons firing in my brain and everyone just kind of freaked out in the hospital room. Everyone started cheering. They were clapping, which was totally unnecessary. It was so awkward. But yeah, it was really, really cool. Right.

Right. I'm so curious, Nolan, about the actual experience of using this Neuralink device because one of the things that it allows you to do is to control a cursor on a screen as if you were, like, using a mouse just by thinking. But I've never known, like...

When I'm using a computer, I'm not thinking, I'm going to place my cursor here. I'm going to click this button. It's a much more fast-twitch, unconscious response. I'm curious, when you're trying to control a cursor on a screen, how intently do you have to think about it in order for the cursor to actually react? At first,

I wasn't very good at it. I was doing what we call like attempted movements. Attempted movements are basically, you know, like I said, all the signals in my brain are still firing. So the threads are implanted in my motor cortex. And so when I attempt movements,

to move my hand, those signals are firing, the implant's picking that up, and an algorithm is basically learning what I'm trying to do, and after doing it a certain amount of times, it'll translate that into cursor control in some way or another, and then it'll keep learning. As I was like a week in, maybe two weeks in, I just thought to move the cursor in one direction, and it moved. It...

blew my mind like it was it was so wild but then over time it just becomes second nature it's not like i'm thinking like cursor come over here and like i'm waiting for it to get there or anything it's just it's very very like i said intuitive and and how like fine grain is the control you have over a cursor now is it like roughly equivalent to like what it used to be like when you were using a mouse or is there still a gap there i would say

It's very similar. I'm not as quick with the cursor as a lot of other people, but I don't think that can't be made up. I think that just comes with...

a bit more practice, and also just a bit more tweaking on the software side. Like this is still very early days, a few months in, and we're already where we're at, which is amazing. I think by the end of my time in this study, whenever that will be, I'll be better than most people with a cursor. Wow. I mean, one of the details that I just love from

the initial reporting on your story is that after you got your Neuralink, you played eight straight hours of the video game Civilization VI. And I just love that because it's like, I imagine that, you know, you had doctors years ago thinking that if

In theory, if this ever, you know, this brain-computer interface thing ever worked, it would allow people to, you know, do more types of creative labor and be more productive at work. And you get it and just immediately start gaming, which I think shows that you just have the heart of a true gamer. Yeah.

Yeah, yeah. I keep telling them that. I keep saying, you know, I'm just so unproductive with this thing. Like, don't you guys give me more things to do? Like, I would much rather be doing work. And they're like, no, just do what you want to do. That's what we want is to make you able to play games, to go surf the web and do things that you want. It's not about, you know, what other people believe you should be doing or anything. It's just whatever makes your life better. Yeah, waste your life like the rest of us.

Exactly. That's what I'm saying. I'm no better than any of you guys. I mean, it makes so much sense, though, because it's like for what, eight years you had been deprived of being able to just use your hands to play games. I love to play video games. I play video games myself every week, and I guarantee you that if I had been in your shoes, I absolutely would have been playing Civilization when they went to. Yeah, yeah. Now, is it also true that you played Mario Kart? Yeah.

I did. I did. Very early on, maybe a couple weeks in, they hooked me up to a switch. And that was very hands-on. They were real-time tweaking things. Right now, we're working on giving me that capability on my own. So any day I want, I can just hop in. I think that's pretty close.

That's amazing. Now, I want to know how different this is from other assistive technologies that have come before because we've had things like eye tracking for computer control before for people who have lost mobility. So have you tried any other ways of controlling mobility?

computers before this? And sort of how does this implant stack up to other things that people have been using to do similar things in the past? Yeah, I've tried it all from, you know, the first few weeks, first few months I was in the hospital after my accident, they had me trying everything. And sure, things have gotten better since then, but they're just not even in the same league as Neuralink. Eye trackers, they're just not

as good. A lot of it has to do with being centered on the screen, making sure that your levels don't change. I have really bad spasms. I'm very spastic. So if I move at all, like my body spasms to the right and I'm off center, then the eye tracker doesn't really work anymore. I've tried other things like a quad stick. And there was a video of a guy, a quadriplegic who was using a quad stick to play things like Fortnite and stuff.

I tried, I tried playing, uh, like one of the call of duty games come out like after my accident and I hopped on and try the campaign. And I think they're storming like the beaches of Normandy and I didn't even make it off the beach. Well, unfortunately the case for so many people at that beach. It's a very real, real case. You know, that happens to me too. When I play a team shooters because I just get my, my butt kicked by 11 year olds. So I'm, I'm sympathetic. What's your excuse? Yeah. Um,

Nolan, I want to ask you about thread retraction because this is something that has happened since your surgery. A lot of the threads that connect the Neuralink device to your brain actually started retracting. I saw a figure of that 85% of them had retracted and this was potentially endangering your ability to use this device. So talk with us about this and sort of what the first signs you experienced were that something wasn't right.

Yeah, a few weeks in, I just started losing control of the cursor is what it comes down to. It would start drifting on me. I would want it to go right and it would go left. I could not get it to go down, things like that. And it just became impossible to use. And then, you know, about a week later, so this was about three weeks in, about a week later, they told me that, you know, they had seen some evidence of thread retraction, but I think they had only found out like a day before. They kept me in the loop the whole time.

So they actually took a scan of your brain and said it looks like we can see that the threads have retracted? No, no. So brain scans won't even show the threads. What they can do is look at the...

electrodes over time and see which electrodes on the threads are sending signals and which ones are sending strong signals or weak signals. And so they can really tell like which, which electrodes are still in my brain. And so right now it's about 15% that are still actively sending like strong signals in my brain.

And do they have any theories about why these threads had come loose? Yeah, it has to do with how much the human brain moves. Apparently, they had thought that everything they had read, all the surgeons they had talked to, said that the brain moves about one millimeter. And then when they implanted everything in my brain, they found that my brain moves actually three millimeters. So it's on a scale of 3x what they were expecting. So...

Obviously, this is not like fixing a computer or an iPhone or something where you can just like open it up and fix it. Like the opening up would involve opening up your head and your skull and doing brain surgery on you again. So how did they go about trying to fix this?

Yeah, I offered them to go in, take out the implant and put in a new one. I was like, if it's going to get me back to peak performance, then that's what I want. If it's going to help me stay in the study, I offered that and they said, no, we're going to stop, take a step back and stop.

see if we can fix this on the software side, which is ultimately what ended up happening. They just tweaked the way that they were recording signals from the threads and from the electrodes. And that ended up working there. There were a couple of different ways they were recording the neuron spikes in my brain. I mean, there's a ton of information coming from the neurons at

all times. And they were trying to interpret those spikes in a certain way. And they had found that the way that we started was the most efficient. But then once all the threads started retracting, they needed to rethink that. And so they switched to a different way of recording those signals. And they found that that was actually much better.

And so, like, where are you at now? Like, Neuralink has said that your performance with the device is now better than it was before all of this happened. Does that continue to be the case? Yeah, yeah, absolutely. I'm still getting better, too. I'm curious, like, some of the people that I've talked to in the tech community believe that BCIs are going to be just a major mainstream technology in the future, not just for people with disabilities, but for basically anyone. Yeah.

And that, you know, the next big platform shift may not be people putting computers on their head, like with VR. It may be people putting computers in their head. And eventually we will all be walking around with these brain implants. Based on your experience, do you think that is a plausible future here? Yeah, I don't see why not. I think they're safe. I think the possibilities are endless with this technology. I mean, we're just scratching the surface. I don't know what kind of things people are going to

be able to do with this in 10 years. I don't think anyone really knows. There are applications that we see can be useful, like helping cure paralysis or different motor diseases, helping cure blindness. But once we start getting all of that, then...

you know, that begs a lot of other questions. Like if we can do this, why can't we do even more? And, you know, with the AI revolution that we're in right now, like how can this be applied to all of these pieces of hardware in our brain? I just think we're in for an explosion, like exponential growth in this field, especially now that Neuralink has come out with this. It's going to bring all the other BCIs up and it's going to push Neuralink to get even better. It's going to be like a new space race, but in the brain.

Hmm. You know, I'm curious to get a sense of what this has just been like for you. You know, you were talking earlier, I was really moved when you were talking about just craving the experience of holding a book in your hands again, something that I take for granted. And now, presumably, you've been able to read books, you've been able to play games. What is this done for you emotionally to kind of get access to some of those things that you had been missing out on?

I mean, it's hard to even put into words. Just this amount of independence that I've been given, it...

It changed my life for the last few months. It's changed my parents' lives. Little things, I mean, very, very little things have made huge differences. Like when I was able to get a drink of water on my own in the middle of the night because I got like a little bottle that stretched across my bed and allowed me to have drinks in the night. That

relieved about 90% of sleepless nights that my parents had. And now with Neuralink,

it's even more than that. I'm able to do a lot more on my own than I was never able to do in the last eight years. I don't have to wake anyone up in my family to come help me in the middle of the night. I don't have to feel guilty if at 2 AM I want to connect and read or listen to an audio book or play a game or just go on and check my social media or text someone back. I don't have to feel guilty about trying to wake someone up. Um,

There's just so much that I'm grateful for being able to do this and ultimately I want to use it to help people Find some way to help people and I'm on the path to doing that and that's what I've wanted since I was a kid just to find some way to help people and after my accident I didn't think that was ever gonna happen I knew that I could still speak but I mean who would want to listen to me speak about you know nothing I had no like life experience to give them I guess now it's a bit different but um, I

Yeah, I don't know. It's just it's been such a huge blessing to me, honestly. Yeah. I'm curious, Nolan, what have your conversations with Elon Musk been like? I haven't had many. I talked to him on FaceTime right before surgery. I was like, hey, thanks for choosing me to do this. I'm really excited, really thankful, really blessed. And he was like, yeah, you know, this has been great. We're looking forward to it, making a huge step. And I said, let's rock and roll. He's like, let's do it.

And that was it. And then after surgery, I spoke to him in person. He came to the hospital. I was still pretty drugged up on anesthesia and I couldn't get his sweet bomber jacket out of my mind. I was just lying in my bed thinking the whole time, don't mention his bomber jacket. Don't mention the bomber jacket. But it was cool. I think we have

very similar ideals about what this can do for humanity and where we can go from here and just our drive to help people in that way. I think it's amazing that someone of his caliber has stepped up and stepped into this role for helping people like me. I mean, I never thought

anything like this would ever happen to me or to people like me and to have such a high profile figure say, you know, I'll, I'll take that on and I'll fight that. It's just amazing.

I'm curious what you make of the promises that people like Elon have made for how BCIs could improve in our lifetimes. He did get a little criticism a few years ago for some statements he made at a Neuralink presentation where he suggested that these BCIs could eventually allow blind people to see or give people with spinal cord injuries, like the use of their full body's back.

a lot of health experts were very skeptical and they basically said, it's irresponsible to say this stuff given that the science is just nowhere near there yet. And I'm curious how you feel when you hear the kinds of lofty promises about what this technology may be able to do someday. Do you get excited or do you say like, hey, wait a minute, let's like stick to what the science is capable of now?

No, I'm super excited about it. It gives me and people like me something to hope for. Once you take away hope, that's the end for most people. And for him to promise something like that, even if it never comes about, it's just the fact that he's trying and he sees it as a possibility. I take that kind of passion to heart. I don't agree with people who say that it's irresponsible. I think

It's a reality from my perspective that it's going to happen probably in my lifetime. And if I'm irresponsible for saying that, then like, I'm sorry, but it gives me something to look forward to and gives me something to strive for and to work towards. And maybe I fall short of that, but I'll be damned if I don't give it my all.

Yeah, I mean, what's very clear to me about you, Nolan, is that you just have like a much higher risk tolerance than I do. Like, I get nervous to go to the doctor to get like some little, you know, thing. And you here are saying, I will volunteer, I will step up to be patient number one for this potentially very severe use of technology. So yeah, my hat's off to you for just being willing to put your hand up for it. Yeah, for real. Yeah. Thanks, man.

All right, Nolan. Well, great to talk to you. Thank you so much for your time. Dude, you're awesome. Thanks so much for coming on. Thanks for having me, guys. I really appreciate it. After our interview, we reached out to Neuralink to confirm some of what Nolan shared with us about his surgery, but we didn't hear back from them. You can read more about his experience on their website. When we come back, we'll talk to my colleague Karen Weiss about Microsoft's big AI announcements.

This podcast is supported by Search Engine from PJ Vogt. Search Engine was named one of the very best new podcasts of 2023 by The Economist, Vogue, and New York Magazine. Search Engine answers fascinating questions about technology and everything else. Questions like, why are drug dealers putting fentanyl in everything? How do

How did ADHD medications get so popular so fast? What do trigger warnings actually do? Listen to and follow Search Engine with PJ Vogt, available on the free Odyssey app or wherever you get your podcasts. Well, Casey, it is the most exciting time of the year in the tech industry, which is developer conference season. That's right. For a lot of people, Kevin, this time of year is about dads and grads. For us, it's about APIs. So last week we talked about Google's iOS.

developer conference and everything they showed off. And this week, Microsoft had its big annual developer conference called Build. You did not go in person, did you? I didn't. And candidly, while I read some coverage of this, I want to learn so much more because, you know, I only had so much time left over after I finished researching the filmography of Scarlett Johansson.

Right. So Microsoft obviously is also very excited about AI. They have been building out a lot of their own AI tools and products and services. And this week at Build, they actually demoed some new hardware that they are making that is sort of built around AI.

AI. Now, isn't putting AI directly into the computers how Skynet began with the Terminator films? I'm not sure. It's been a while since I watched those movies. But I think Microsoft gets less coverage by tech journalists like you and me than it deserves, in part because a lot of what they do is like boring enterprise software stuff. But, you know, they are the biggest company in the world, and they have been investing in AI significantly over the past few years. And I would say between their stake in OpenAI and

all of their own AI projects. They're just a major, major player in this world. Yeah, and it's true. I probably don't pay as much attention to Microsoft as I should, and it is for a somewhat selfish reason, which is I just use Macs. And so sometimes it feels like this stuff just is not as relevant to my life. But let's face it, for most of the working world, they are doing their work on a PC. And so if Microsoft says we're putting AI in it, then we should be paying attention. Yeah. So

to talk about this, we're going to bring on my colleague Karen Weiss, who covers Microsoft for the New York Times. She went to Build and she's going to tell us all about what Microsoft announced. Karen Weiss, Falcon Hard Fork. Happy to join you guys. So Karen, you actually went up to Microsoft's headquarters earlier this week for their annual Build conference. So just set the scene for us a little bit. Like, what was it like? How did it compare to previous experiences? Well, I think what we

I think what was unique was on Monday, they tried to really hype this AIPC announcement. And they did something on the campus, like the big headquarters in Redmond, just outside of Seattle. And you had to be in person there. They weren't live streaming it. And Satya was going to give a keynote at it. So it was trying to definitely build that.

And so, you know, they tried. It was like a lot of hoopla. There's a lot of like music in the background and stuff like that. And they tried to recreate the magic, if you will, of when they launched the Bing Sydney chatbot that shall not be spoken of about a year and whatever ago. Yeah, that all turned out great. So, Karen.

One of the things that Microsoft announced are these things called Copilot Plus PCs, which, as I understand it, are basically a personal computer, a Windows PC that is basically built to run AI and that runs it very fast and that is sort of all wrapped around the capabilities of these AI models. So what did they actually announce and sort of how is it different from the Windows PC that people use today? Yeah.

Yeah, the main thing is that these PCs have a bunch of AI models, AI systems locally on the computer. And they can run different AI systems.

or models because they have this new type of processor, essentially. It's called a, bear with me, an NPU, a Neuro Processing Unit. And it is very quick with really little drain on your battery because the problem you have is when you run AI, it's like super intensive, right? It's running all these calculations constantly. And so this is a whole new generation of chip. Right. So I guess I'm just struggling to understand what it means for AI to be able to run locally because I also have run...

large language models locally on my hard drive before. And is that what they're saying? Is that if you want to run something like a chatbot, it'll just be much faster to do? Or are there actually capabilities that these will have? Is it going to be so deeply woven into the operating system that...

people's experience of their computer will actually change somehow. Right. So they are hoping the latter. You'll be shocked to hear. But basically the idea is because there are these things that are run locally, they can have access to information that is only stored locally.

locally and they can be faster and more interactive in a way where, you know, even right now when you go into a chat bot, you ask your question, right? And it like takes a second and it goes boop, boop, boop, boop, boop. And each little word kind of comes out one at a time. This would speed that all up as well. And so they're hoping that by putting these different, they have said there are more than 40 models that can run, can kind of come pre-installed, so to speak, and

on the laptops. And their hope is that developers now start playing with those and build tools off of it. And they liken it to the iPhone when the iPhone said, oh, wait, here's a, you know, GPS. Here's a altitude meter. Like whatever kind of tools are in the hardware, people start building it. And then you get...

Uber or whatever type of system that then made the iPhone this kind of like key platform for people to build off of. And so I think there's a reason why they announced this before their big event for developers is because they're trying to say to developers, build for this. There's a future here.

Now, I think speed matters a lot. I do think that that can really change the way that this stuff gets used and I can absolutely see this becoming a developer platform. But at the same time, Karen, I am reminded of the last time that Kevin and I went up to Redmond for an event and we were told Bing has AI now and AI is going to change everything. I think we were optimistic that maybe that would be the case and then you fast forward to today and Google is still by far the most dominant search engine in the market.

So I wonder, as you're hearing this presentation telling us that AI chips inside PCs are going to be a next generation kind of PC that is truly going to change everything, I wonder, what did you think of that?

I think the difference between this and Bing is Windows. So Windows is this ubiquitous operating system that Microsoft controls. So when they announced this, it wasn't just them trying to move from a tiny market share. They have this enormous market share, and they had all of the biggest laptop makers in the world there showing off versions of these devices. So I think they have a power or an influence on the PC in a way that they don't necessarily in search or with Bing. That's

That said, they have to prove the utility of these things. And part of why running the models locally is important is it's cheap, right? You pay for the processor up front when you buy the laptop. But every time you ping it, you're not spending a penny. Whereas if you're inferencing things in the cloud, if you're running things off the cloud, it becomes really expensive to

for a developer to think, oh, can I even afford to offer this product to people? Now, if they can just do it locally, it might not be quite as good. Like the language models that can run on a laptop now are not as good as GPT-4, but they say about as good as GPT-3.5, which was what ChachiDBT was initially launched with. So you can get some like good enough uses is the theory.

Got it. But again, they have to find the kind of key use cases that will show it. And I was a little surprised. The examples they released with were not – they were clueless.

were clearly trying to demonstrate certain capabilities. Like what? What did they demonstrate? Some were a little more speculative and futuristic, saying this is the kind of thing you would be able to do. So like one was a dad speaking with a voice chatbot, essentially asking for help on how to solve the particular Minecraft situation. And the voice thingy can come in and then say...

oh, you need these materials to build this new thing in Minecraft and therefore, oh, hurry, this thing just popped up. Go run over there and hide in this basement thing. And because gaming is real time, you can't do that if you're pinging the cloud. And so this was an example of a way where you could bring in generative AI and assistant and

into like a real live moment, essentially. But that was one of the more futuristic products. It's like not launching with that. Right. I want to ask about this other feature that they announced that I think got a lot of attention, which is something called recall. What is recall? Yeah.

They liken it to a photographic memory. And it's actually kind of makes sense because what it does is it basically builds a history of everything you have looked at on your laptop. And it literally takes constant screenshots of your screen, stores those locally. And then you can ask it, oh, where was that thing that Kevin sent me that was like kind of weird and he laughed when he told me about it?

And since you and I are, of course, chatting on Teams, Kevin, on Microsoft Teams, it was like, oh, we can pull that up here because the transcript says that he laughed in this moment. And you can kind of scrub back and forth in time trying to visually look for what you were searching for. So basically, if you've always wondered what would it like to have an FBI agent living inside your computer, you can now have that. Perfect. So...

There's an app called Rewind that has been doing something that sounds very similar to this. But just so I'm clear, this is taking screenshots, it is storing them on your machine, and then it is allowing you to use generative AI to sort of search back through your processes.

previous encounters with your computer and say, what was that restaurant menu I was looking at last Tuesday or whatever? But like, I'm curious, Karen, what you think the target audience for this, who is actually going to use this feature? And like, what examples did Microsoft give of how it might be useful?

They gave some of kind of personal uses. They had this very funny example of this woman trying to get a dress for her grandma. And they talked about how she had searched things online. She had chatted in Discord with her abuela, which I thought was very funny because like my grandma used to AOL instant message me, but I'm not sure how many grandmas are on Discord. But the idea was that there's this like digital record and you can go back and be like, oh, what was that sparkly one?

And you can use language like that. And because it has visual intelligence, it can go back and look for the sparkly blue dress you had looked for. And then, oh, didn't Abuela say that she actually really liked a pantsuit instead? And then you can kind of find the pantsuit. So that's kind of a personal one. Like, I was recently shopping for jeans, and part of me is like, oh, I could see that. Like, where was the jeans that I liked? But also, I'm scared if I were to type in, like,

wide-leg jeans I was just looking at, how many wide-leg jeans would show up? It's kind of mortifying to think about how many wide-leg jeans you need to find to find a good pair. Totally. So, you know, I can see all those use cases. At the same time, all of us are journalists. We often talk to people off the record confidentially. We have sensitive information on our laptops. We're not a

And that I think most people working jobs have some sort of confidential information that is on their computer. So I hear everything that you've just described, Karen, and I think absolutely not. I tried Rewind for a while. I found it terrifying and deleted it from my phone. What is your thinking right now about whether a lot of people are going to be willing to invite this level of surveillance onto their devices? Yeah, I mean, the default is definitely Rewind.

to take all of it. You can go back in and manually delete certain days. You can have it opt out of certain applications. But like if you're going to opt out of the web search, like that's one of the main uses for it. So it's a big question. That said, it's defaulting to being there. And one of the things when I was kind of researching before the event was,

A lot of people haven't interacted with chatbots. They haven't used chat GPT, but they get a basic Windows laptop from their work. And there's a big old button that says Copilot on it. And these tools will be there. And for many people, it might be the first time that they really have exposure to it. Also, I'm not sure people will understand the technology behind it and that it's literally taking a picture of everything that you're doing.

Right. And we should say, like, you know, Microsoft has said that all of these screenshots stay on the device itself. They are not being sent to Microsoft. Microsoft has also said, you know, they're not doing any kind of like content moderation on them. So like if you've been looking at your bank account information on your PC, they're not going to like scrub that from the screenshots, but that it will all stay locally on the device. And so only the person whose device that is can access.

access that. But I think there are a lot of questions that people will have, especially, you know, if this is maybe a corporate issued computer, does my employer then have the ability to go back and look at screenshots of every time I've ever used this computer? Of course they will. And as this stuff gets normalized, I can imagine employers handing you your shiny new Copilot Plus PC and saying,

you have to leave this recall feature on. And then if we ever have a disciplinary issue with you or we are just suspicious about you for whatever reason, we can review everything you have ever done on your corporate-issued laptop. That seems like a nightmare dystopia to me. And they can search for it easily. Yeah. Because they now have this search tool. Right. Your only chance to survive that is just that they're using Bing search, which doesn't work most of the time. Right.

Well, Karen, that brings me to one of my other questions about all of this stuff that Microsoft announced this week is like, who is it for? Because I think traditionally Microsoft, a big part of their business is selling to businesses. It's enterprise customers. It's, you know, large companies that already run a lot of Windows PCs that already use Outlook and Teams and all the other Microsoft products. And they can just kind of keep adding to that population.

So do you see these AI PCs and all the features that are on them as being aimed at businesses or are they really making a consumer play here? You know, I initially thought beforehand that it would be more of the businesses. Businesses buy literally the majority of laptops now. When was the last time you guys refreshed your PC? Mine was in 2011, I believe.

You know, you're eligible every three years at the New York Times. Oh, no, no, my personal one. When was the last time you bought a personal one? Wait, your PC is 13 years old, Karen. What are you using it for? Are you running Windows 98? XP? It's a MacBook, and it's lasted for the Internet, right? Everything's on the Internet that I do now. I'm not ripping my CDs anymore and taking my music from one thing to another. You're not mining Bitcoin? I'm not.

I was surprised, though. They definitely sort of benchmarked it and kind of compared it to the MacBook Airs. So they're smaller. They are lighter. They say they start at $1,000. So they're not like Chromebook prices, but they're also not a MacBook Pro super heavy thing.

I'm curious how this positions Microsoft against Apple. Apple's having its own developer conference next month, and there's a lot of speculation that Microsoft's announcements at this event this week were designed to take some attention away from anything Apple might announce in just a few weeks. Apple is also expected to do a lot around generative AI in its own hardware products. What is the competition between Microsoft and Apple like right now?

These chips, these new NPUs, do like propel the performance of these laptops basically in the realm of Apple now. So it's in the same class because they've changed over this chip generation.

architecture or style essentially now to a model that actually Apple had been pursuing. So the question is, will people make the jump from one ecosystem to another? Will their employers make a jump from one ecosystem to another? All of that stuff is, I think, I don't know the answer to that. But they are clearly trying to show that they are building even their hardware with this kind of AI-first mentality. And again, if people want that, if they can demonstrate the utility of that, that's like the kind of question about all of this.

Yeah, I just think so much of this is going to depend on how Microsoft implements all this stuff and honestly, like how annoying they are about it. I mean, one of the, so I, you know, I have a Windows PC. It's my daily sort of PC that I use. And I would say most of the time I really like it. But then there are just these times when it's very clear that Microsoft is just getting a little bit greedy and they just start popping things up or

putting things in weird places. Like the other day, I was using my computer and I got a Skype news alert. Have you seen these yet? I've seen somebody posting these on threads and it's so funny. It's so funny. I was just minding my own business, doing email. Up comes this little notification. It says Skype says the U.S. economy added this many jobs last month. And I'm like,

Why is Skype talking to me? Did I ask Skype to talk to me? No, I did not. But that is just a classic case of Microsoft sort of trying to juice engagement by doing something that I think a lot of people would feel is very annoying. So I can see these AI PCs being very useful and I want to try one, but I think that if they can't resist sort of

I don't know, just trying to nudge you into using it more and more or in different ways, I think that is going to turn a lot of people off. I mean, I think a lot of AI things are going to be nudging you more and more because they want discoverability is that phrase that you hear a lot about, about the features that these things can do. You know, this is a huge Alexa problem, right? People know Alexa can do the timer or whatever, but she can...

she can do more and they want you to do more, but like they got to push it to you. Otherwise, how do you know? So I think that's a very hard urge to resist, even though I completely understand and agree. Yeah, I mean, Kevin, I'm like you. I would like to give one of these things a try, see what it can do, see how fast this AI is once you get it on the device. But I do continue to have trust issues with Microsoft. This is a company that just last month started testing everything

ads in the Windows 11 start menu. So every time you go to like look at the programs on your computer now, you might just have to see an ad. And I don't know, the more AI is on my computer, the less I'm excited about a company that is looking to shove ads into different parts of the interface. You know, imagine you're going to, you know, shop for jeans for Abuela and then, oh, you're looking for that? Well, here's an ad for that now, Karen, right? So there's just a lot of stuff in here that sort of has my eyebrows arched.

Karen, did you actually get to try any of this stuff? Did you get to get your hands on one of their new AI PCs? Yeah, they had like a demo station set up. And so like it's populated with like all the demo data. And then so the recall process

example was hard to know what it would be like when you had a real body of your own data. Would it feel super creepy? Would it be really useful? Would it be not useful because it returns so much information that you can't actually scrub through it all? I understand the problem they're trying to solve with that. Like, I think we all have the zillion tabs open that you keep open just so you don't forget about it. So I understand that, like, impetus behind it, but it was hard to get a sense for that for me of, like, would I personally use this and

like it essentially because it was just filled with all this dummy content essentially. Right. You know, Karen, dummy content was actually the original title for this podcast. Some people think it still should be. All right. Karen Weiss, thanks for coming. Thanks, guys. Thanks, Karen.

This podcast is supported by Search Engine from PJ Vogt. Search Engine was named one of the very best new podcasts of 2023 by The Economist, Vogue, and New York Magazine. Search Engine answers fascinating questions about technology and everything else.

Questions like, why are drug dealers putting fentanyl in everything? How did ADHD medications get so popular so fast? What do trigger warnings actually do? Listen to and follow Search Engine with PJ Vogt, available on the free Odyssey app or wherever you get your podcasts.

Thank you.

Special thanks to Paula Schumann, Pui Wing Tam, Kate Lopressi, and Jeffrey Miranda. As always, you can email us at heartforkatnytimes.com. Earning your degree online doesn't mean you have to go about it alone. At Capella University, we're here to support you when you're ready. From enrollment counselors who get to know you and your goals, to academic coaches who can help you form a plan to stay on track. We care about your success and are dedicated to helping you pursue your goals.

Going back to school is a big step, but having support at every step of your academic journey can make a big difference. Imagine your future differently at capella.edu.