Learning to Live With Google’s AI Overviews

Google has spent the past year lustily rolling out AI features across its platforms. But with each launch, it is becoming more clear that some of these so-called enhancements should have simmered a little longer. The latest update to stoke equal parts excitement and ridicule is AI Overviews, the new auto-generated summary boxes that appear at the top of some Google search results.

In theory, AI Overviews are meant to answer questions and neatly summarize key information about people's search queries, offering links to the sources the summaries were pulled from and making search more immediately useful. In reality, these AI Overviews have been kinda messy. The information the summary confidently displays can be simply, and sometimes comically, wrong. Even when the AI Overview is correct, it typically offers only a slim account of the topic without the added context—or attribution—contained in the web pages it’s pulling from. The resulting criticisms have forced Google to reportedly dial back the number of search queries that trigger AI Overviews, and they are now being seen less frequently than they were at launch.

This week, we talk with WIRED writers Kate Knibbs and Reece Rogers about the rollout, how Google has been managing it, and what it’s like to watch our journalism get gobbled up by these hungry, hungry infobots.

Show Notes

Read Kate’s story about Google trimming the frequency of its AI Overviews. Read Reece’s story about how Google’s AI Overviews copied his original work. Read Lauren’s story about the end of Google Search as we know it.

Recommendations

Kate recommends Token Supremacy by Zachary Small. Reece recommends the game Balatro. Lauren recommends the poetry book Technelegy by Sasha Stiles. Mike recommends the book Neu Klang: The Definitive History of Krautrock by Christoph Dallach.

Kate Knibbs can be found on social media @Knibbs (X) or @extremeknibbs (Threads/IG). Reece Rogers is @reece___rogers. Lauren Goode is @LaurenGoode. Michael Calore is @snackfight. Bling the main hotline at @GadgetLab. The show is produced by Boone Ashworth (@booneashworth). Our theme music is by Solar Keys.

How to Listen

You can always listen to this week's podcast through the audio player on this page, but if you want to subscribe for free to get every episode, here's how:

If you're on an iPhone or iPad, open the app called Podcasts, or just tap this link. You can also download an app like Overcast or Pocket Casts, and search for Gadget Lab. If you use Android, you can find us in the Google Podcasts app just by tapping here. We’re on Spotify too. And in case you really need it, here's the RSS feed.

Transcript

Note: This is an automated transcript, which may contain errors.

Michael Calore: Lauren.

Lauren Goode: Mike.

Michael Calore: Have you encountered any AI Overviews in your Google searches?

Lauren Goode: Oh yeah, sure. I mean, they're everywhere.

Michael Calore: Have you found them to be generally helpful? Distracting? Comically wrong?

Lauren Goode: Slow and mostly distracting. And we have seen some comically wrong examples. Yes, I think a lot of people have. There are also a lot of fake examples at this point. People are having fun with it. But yeah, it's a little concerning that this is Google Search now.

Michael Calore: Yeah, the rollout has not exactly been smooth. There have been bumps.

Lauren Goode: Have the bumps gotten worse? I have to say I've been out of the office for a week traveling in a place where GDPR is very strong.

Michael Calore: Humblebrag.

Lauren Goode: Yes. And my internet experience has been accepting or rejecting about 1,300 cookies. Haven't been paying as much attention to Overviews. What's been going on the past week?

Michael Calore: Oh, it has been a journey.

Lauren Goode: Oh really?

Michael Calore: Yes. It continues to be a journey, so we're going to talk about it this week.

Lauren Goode: All right, let's do it.

[Gadget Lab intro theme music plays]

Michael Calore: Hi, everyone. Welcome to Gadget Lab. I am Michael Calore, WIRED's director of consumer tech and culture.

Lauren Goode: And I'm Lauren Goode. I'm a senior writer at WIRED.

Michael Calore: We are also joined this week by two guests. In the studio we have WIRED staff writer, Reece Rogers. Welcome to the show for the first time.

Reece Rogers: Hello. It's an honor to be inside of the Gadget Lab.

Lauren Goode: It's so fun having you here, Reece.

Michael Calore: And beaming in from Chicago, we want to welcome back to the show WIRED senior writer Kate Knibbs. Hi, Kate.

Kate Knibbs: Hi. Thanks for having me.

Michael Calore: Always great to have you.

Lauren Goode: Isn't it funny how everyone says the same thing on the start of a podcast? Everyone's like, "Thanks for having me." No one says, "I've been really busy, but you're welcome, guys. I'm carving out time to do your podcast." We're very grateful you're all here.

Michael Calore: New paradigms. It has been a nutty three weeks since AI Overviews first arrived. If you use Google to search for things on the internet, you have seen them. They show up on certain searches where instead of seeing a list of links or curated information about the thing that you typed in, you will see an AI-generated summary of the topic that you're searching for. Most of the headlines about AI Overviews have focused on how weird they are or inaccurate, or how they just regurgitate whatever is in the top-ranking search result.

We have all the experts here this week to talk about it. Lauren, you covered the launch of AI Overviews when you interviewed Google's new head of search, Liz Reid, at the beginning of May. Reece, you've been digging into how AI Overviews work since they arrived, and you have some fun experiences to tell us about. And Kate, we're going to start with you because you have done some reporting around how Google has been managing the rollout. So please, if you could, tell us about the research report that you got your hands on this week.

Kate Knibbs: Yeah, of course. So I also had mainly been seeing AI Overviews in the news because of how wrong and bizarre they were. And it was interesting, the report is from an SEO firm called BrightEdge, and they were monitoring how often AI Overviews would appear when you searched for a variety of different keywords. And they saw when it was in beta testing, it was really frequent, up to 84 percent of the time that you typed in a keyword you would see an AI Overview. They tracked it dramatically decrease. And I had sort of assumed that the decrease would've occurred when there was all the backlash about AI Overviews telling us to make pizza with glue and eat rocks. And that Barack Obama was the first gay president and all of these extremely wrong things.

But what their data revealed was that the decline had been occurring since the launch. And it sort of suggests that Google was retooling the search right from the get-go, that they sort of knew when they launched that they didn't have a 100 percent clean winner on their hands and they were pulling it back from as early as the day of IO.

And the other interesting thing that this research brought out was that even though it's decreased dramatically across the board, there is still a few different categories where if you type in a search, you're really likely to get an AI Overview. And one of the categories … Some of them you won't see. If you type in something related to travel, it's extremely unlikely that you'll get an AI Overview. E-commerce, actually I thought it would be one of the leading categories, and it's not. The leading category is health care. So if you type something in asthma or do I have diabetes or basically if you're searching for anything involving diseases, you are very likely to this day to still see an AI Overview. I mean, it's not everyone, it's still like 63 percent. And I think it depends.

And also I will say Google, when I asked them about this, said that the numbers were off, but then wouldn't give me their own numbers. And that was interesting. And then I talked to some other SEO researchers who had also found that healthcare was still producing a lot of AI Overviews. So I independently verified the patterns existing, but if you talk to Google, they'll probably say something slightly different.

Yeah, so the SEO experts that I talked to, also a lot of them were actually more complimentary towards Google than I expected them to be considering how I feel like the general public has just seen this rollout as a total calamity. A lot of them were saying that they'd already seen big improvements and that they were optimistic. Jim Yu, who's the executive chairman of BrightEdge, called this a blip and said he really thought that AI was still the future of search. So I don't know, I was like, "I guess that's a good thing."

And one thing I will say also is I typed in … I was sort of trying to test some of the health queries and I watched one of the results improve in real time, which was interesting. The first time I typed in "is chocolate healthy," the AI Overview linked me to two different chocolate companies in addition to more reputable health resources. And then when I tested it again six hours later, it had removed the chocolate companies from its citations and was only citing reputable healthcare links. So I was like, "OK, they're trying. I don't know if they're doing a good enough job, but they're definitely trying." So it's definitely something that I'm going to continue to monitor because it seems like they're still definitely figuring things out.

Lauren Goode: What is the error rate actually at this point? How many AI Overviews being generated per a certain number of searches go against the content policies at Google or are actually just tragically wrong?

Kate Knibbs: I would love to know that. I don't know if anyone has definitively pinned that down yet, especially because they are making so many changes. The SEO companies, honestly, they all had slightly different methodologies for even tracking the searches. And all of them were approximations. They don't have access to what Google has access to. They don't have access to what everyone is searching, so they're just sort of selecting popular keywords. So basically, I don't know what the error rate is. It still seems pretty high anecdotally, but we'll have to see what sort of research is done.

Lauren Goode: I actually just searched this as you were talking about that and did not get an AI Overview for it, but Google said that only one in every 7 million unique queries resulted in a Google policy violation of some sort. So that's one in every 7 million. OK.

Michael Calore: So that's like one a second.

Lauren Goode: I mean, yeah. How many billions of searches are done every day? Yes. OK. Well, that still seems fairly high now.

Reece Rogers: Yeah, that does seem quite high. And I don't know if this is a content violation, but when I was testing out part of the feature, I searched "do gays mainly get the Mpox or monkeypox?" And it was kind of funny because the AI Overview said yes, and then the featured snippet said no, and they cited the same source. So there's still a lot of things that they're trying to figure out related to health care queries. And I'm just really surprised to see Google still allowing AI Overviews to pop up for health-care-related results. That's something that definitely I feel like they would crack down on the guardrails a little more to try to fix hallucinations or like Kate mentioned referencing a chocolate website when you're asking about chocolate questions. There's definitely a conflict of interest there.

Lauren Goode: Right. Well, one of the things that Liz Reid said … So Google has been doing some PR control around this, and they put out the first blog post or news release that explained what the error rate was, and then Liz Reid, the head of search, put out another blog post where she was saying that the AI Overviews don't actually hallucinate. I'm guessing people like us, the press or others were just generally using that term. And she said that the way that it's working is it's not making things up in the same way that other large language model products might. It's wrong because the AI is misinterpreting queries or misinterpreting a nuance of language that exists on the web or it just doesn't have a lot of great information available, which are generally challenges that occur within just regular searches too.

What's surprising to me, I mean just to layer onto the "Wow, it's surprising that they're still doing this." It's also surprising it's health-care-related queries. But what's surprising to me is that Google has been dogfooding this or testing in beta for over a year now. They had to have seen these issues. They had to have, and yet Google IO came and they said, "Let's put it out to the world," or at least the English language-speaking world.

Michael Calore: Yeah. And there's a great amount of pressure in the industry to show your AI chops and to have AI products out there, and what better way to make them better than just to let millions and millions of people use them, I guess. But it's still kind of weird because we all rely on Google Search multiple times every day for very important things. And then it just feels like this sort of drunk uncle is coming in and crashing the party, making it difficult for you to get your work done. Right?

Kate Knibbs: And I will say I had last night an experience that made me realize how much I don't trust AI Overviews, especially with health care, which is my son has had a fever for the past few days because he has an ear infection, and we were trying to figure out whether we should wake him up to give him Tylenol. And it did trigger an AI Overview, and I found myself just ignoring the AI Overview and trying to scroll down to use Search as one would traditionally use Search. And I was getting frustrated when I saw the AI Overviews when I was doing different permutations of the search because I don't like them right now. And again, maybe it will improve or maybe I'll just get used to AI, but I would rather just go to the links and search in the traditional way. I really hope that they let us toggle this on and off going forward.

Michael Calore: I mean, there is a manual trick that you can do after a search. You go to the tabs at the top of the search where you can …

Lauren Goode: Go to the web.

Michael Calore: Yeah. Where you would click for news results, image results, video results. Sometimes you have to hit the little drop-down to see it, but there is an option for web, and if you click that you just get links.

Lauren Goode: And people are building Chrome plugins now for this too.

Michael Calore: Yep.

Lauren Goode: Yeah.

Michael Calore: Yep. And those Firefox extensions, there are different tricks you can do to make that the default experience. But yeah, Google is not going to turn it off. They're forcing everybody to use it and manually step around it.

Reece Rogers: Yeah. And I feel like it's not a sign of a good rollout for a feature if it's going viral to block it. If people are building Chrome plugins to block your new feature, that's not a great sign.

Michael Calore: Yeah. But at least they allow them in the Chrome Store. OK. We need to take a quick break, but we'll be right back.

[Break]

Michael Calore: Now, Reece, I want to ask about your experiences with AI Overviews. You've been testing the feature rigorously since it arrived, and recently you noticed that there was some behavior that alarmed you. Would you please tell us about it?

Reece Rogers: I would love to. So as a service writer at WIRED, I love experimenting with new AI features, especially to help our readers really understand how they can use them and what are the limitations. And so while I was out on a camping trip with some friends recently, I was relaxing in the tent and seeing all these wild AI Overview answers go viral on social media. And now, I had tried out the feature before, but I decided to start searching questions on topics I've covered to see how the AI Overviews might reference my writing.

Lauren Goode: That's a good idea.

Reece Rogers: Yeah. I wasn't surprised to see my article linked as a footnote at the bottom of an AI Overview, but I was quite gagged to see the first paragraph of Google's answer pull so heavily from my writing. Reading the two paragraphs side to side, it really does feel like I'm back in the classroom and someone copied my homework.

Lauren Goode: Which article was this that it summarized?

Reece Rogers: Yeah, so it was an interview I did with one of Anthropic's Claude developers to learn more about how to use Claude. So it was kind of like the snake eating its own tail. It was my article about AI summarized by AI.

Michael Calore: Now, it wasn't plagiarized, right? There's a strict definition of plagiarism. But it was the same words put in the same context in the same order.

Reece Rogers: It was not a verbatim copy, but it did pull quite heavily for sure. So whether or not it was plagiarism or not, it was far too much of a copy for me to be thrilled about it.

Michael Calore: Yeah, I can imagine. So you called Google?

Reece Rogers: Yes. I had a tense call with a Google representative and they kind of pushed back against some of the things I was saying, but they did acknowledge that AI Overviews may copy part of what the websites are saying. But they said that A, that they link back to the original sources, the sources are highlighted, and that they only use a small portion if they do use any of the web result.

Michael Calore: Sure.

Reece Rogers: In my case, it seemed like it was almost a whole paragraph that had been lightly rephrased. There's a lot of the original words, a lot of the original stuff in there. So yeah, I was pretty bummed about it or just I would say that I wasn't thrilled. And I think it's also important to note that when presented with this specific example, they basically characterized it as a conceptual match and that it didn't replace the original writing.

Lauren Goode: So they're saying that by changing a word or a few that it's not actually just a carbon copy of your writing, but it's capturing the essence of it.

Reece Rogers: Yeah, it's a conceptual match in their words. And chatting with different copyright experts about this, I receive very nuanced opinions about whether or not I could win any hypothetical litigation against Google. One expert pointed out that instructional content is much harder to make a claim about than a copyright claim for something say more creative, like a poem or a screenplay. But another copyright expert suggested that it could still hypothetically be put in front of a jury for them to decide whether the AI Overview in question actually was substantially similar to what I wrote in the article. So there's a lot still up in the air. And with the litigations going on right now, not a lot of settled lawsuits are about AI and copyright currently.

Lauren Goode: So Google has been doing information cards for a long time, and not that I'm like, "Let me help bolster Google's argument for them here," but what is the difference between seeing one of those snippets or info cards versus, I guess putting a paragraph of Reece's article at the top of the results?

Michael Calore: I feel like the featured snippet is clickable, so you can …

Lauren Goode: Right. I'm going to look at that now.

Michael Calore: … click on the result and lead directly to the source that it's citing.

Lauren Goode: Like this, is chocolate healthy? That?

Michael Calore: Yeah.

Lauren Goode: Right. OK. And then it's John Hopkins Medicine.

Michael Calore: Yeah. So when you search for something and you get a featured snippet, which is the thing that we've been seeing for years and years, it's easier for you to understand that the text that's being cited is from an article and there is a clear link directly below the citation and sometimes the citation is accompanied by a photograph or a piece of art taken from the article as well. It makes it feel like, oh, here's the piece of the news article that you were looking for and if you want to read the whole thing, whereas AI Overviews, it's like Google just has the answer for you.

Lauren Goode: Right, it's in the language is more natural-sounding.

Michael Calore: Yeah.

Lauren Goode: Yeah. This right here is clearly not. It says, "Is chocolate healthy?" Here's the snippet, "Increases heart health:"—and then it writes this very clinical statement. It doesn't sound like the robots are taking over.

Kate Knibbs: That's what's so wild to me is that with featured snippets, Google was showing how to do this correctly, and then they came up with a way to do it worse, are saying, "This is our new feature." If you could just click on the AI Overview and go straight to Reece's words and it was obvious that it was culled from his writing specifically, we wouldn't be having this conversation. We would probably be saying, "Oh, it's a pretty good job." But yeah, I don't know. I know that there's technical definitions of plagiarism versus not plagiarism, but I've looked at how similar the AI Overviews are to your articles and for them to claim that it's not plagiarism is really giving Bill Ackman. I don't buy it. I don't know. I would be upset too.

Michael Calore: Well, Reece, let me ask you this. When you saw the AI Overview of your story, how easy/difficult was it for you to find the link to your story?

Reece Rogers: You had to get scrolling that was for sure. The link to my article was all the way at the bottom of a 10-part bulleted list. So it wasn't even referenced or linked directly below the paragraph that copied aspects of my writing. It was way at the bottom after you expanded it. And this is to say that for similar queries without the AI Overview, my article was the featured snippet. So it said WIRED real big at the top of the results. It had the article's title and it was easy to click through. But with this new AI feature, it was buried. The attribution was much harder to see. So even if there was still a link there, I can't imagine the average user actually clicking through anymore.

It's a feature that's built to summarize your articles, and they did the summary, why would you need to keep reading? Compared to the featured snippet where it's just a small part of the answer. So I could see people still clicking on that and going through, and even if they don't, they know where it came from. With these AI Overviews, a lot of the times you might not even know where your information is going if you're just looking for quick answers.

Michael Calore: Right. So Lauren, Kate, Reece, how does this feel for our future?

Lauren Goode: Is this therapy? This is Google Search therapy.

Michael Calore: OK. We're journalists. Our business relies on people coming to our website to learn things or going to our TikTok feed to learn things, and it seems like this tool is making it harder for people to find our website.

Lauren Goode: Yeah. I think that there have been a wave of stories that are about how the web is back baby, and I think some of that is projecting or wishful thinking in some ways. And I hope that to be the case, a more open web. But in general, this feels like this is bad for news and high-quality information.

Now, a lot of news websites are experiencing a decline in traffic for various reasons. It's not just this, which is a lot to get into for a singular podcast episode. We could probably do an entire episode on that if people were … It would be very inside baseball.

Michael Calore: Oh, I'm sorry, did I just drop the existential bomb on the show?

Lauren Goode: Right. Exactly. Is anyone of the Gadget Lab listener crowd that interested in hearing about our business? I don't know. Probably not. But we of course think about it a lot. So there are a lot of reasons why web traffic for a lot of news outlets is on the decline right now. This certainly doesn't seem like it's going to help things. Where Reece just described his service journalism, his how-to, and Reece is nodding right now, it's really valuable, it tends to perform really well, which means a lot of people are reading it and they're finding it helpful. It's also a driver of subscriptions for us.

This is not like … Reece, you put a lot of time and effort into doing this and reporting this out. If the AI Overview is generating something that gives someone a snippet of what they need in a short answer and changes the language a little bit and then buries the actual link 10 links below, one, I doubt that a lot of people would be curious enough to continue browsing. And two, then if they do, what's the likelihood that they're going to end up on the original source of the content? That's just one tiny example amongst billions and billions and billions of examples that we're talking about, but it doesn't seem like it's a good thing overall. And then there's the whole advertising piece of it too, which is... Mark that for the second podcast of existential crisis on Gadget Lab.

Reece Rogers: And people have been asking me why AI Overviews from Google. There are other companies working on similar search products. Perplexity is a smaller startup who's kind of exploring different ways people can search the web using AI. But I think it goes back to Google's consolidated power. It's one of the biggest referrers of traffic, and if you're one of the main referrers of traffic to these publications, you can make a tiny change and all of a sudden the traffic can drop off.

Lauren Goode: Right.

Reece Rogers: So that's I think what people are especially sensitive to the way Google is approaching this. They have so much power, they control so much, so people are freaking out. And maybe, maybe they're overreacting, but maybe they're not. Google won't share the data with us about click-through rates or anything like that, so it's impossible for us to tell and I guess we'll just have to wait.

Lauren Goode: Right. Yeah, it feels like an incredibly rapid shift in media.

Michael Calore: Oh, yeah.

Lauren Goode: Because the transition of print to web took a really long time. And then some of the traffic peaks I'm referring to from the past four or five years, during Covid people wanted a lot of online news information, high-quality news information. During the Trump administration, people who wanted information about politics were being driven to hopefully high-quality news sites. Those things have shifted a little bit over the past three to four years, and so traffic has ebbed and flowed. This feels like something that with just the flick of a light switch could have a really, really big impact on how people access high-quality news.

Michael Calore: I agree.

Kate Knibbs: I think it will be really interesting for us to especially look at how sites like news media outlets that focus on healthcare topics are faring right now because at the moment, it's less than 10 percent of Google queries are triggering the AI Overviews. So we're kind of safe for now I think unless it switches to gadgets. I mean, I would also … We could even look at how our biotech coverage is affected because it seems like there will be certain canaries in the coal mine for how this is going to impact the media industry and it will be outlets that are falling under the categories that are really triggering AI Overviews quicker.

I will also say though, the one reason I'm not freaking out is because as a consumer, as just someone who searches, I hate AI Overviews so much that I'm like, "Maybe they'll just not catch on." Maybe Google will give up and give us the search experience that we want that doesn't have machine-summarizing news articles. I don't know about that.

Michael Calore: I mean …

Kate Knibbs: Yeah.

Lauren Goode: I think we're purists and maybe not representative of everyone. I don't know. What do you think, Mike?

Michael Calore: I think a lot of people really like them, but …

Kate Knibbs: Who likes them? Who's like, "I love AI Overviews."

Lauren Goode: People in the Mission Mike is talking to at independent music venues.

Michael Calore: Yeah, write in.

Kate Knibbs: They like AI Overviews?

Michael Calore: I have talked to a couple of people who think that they're helpful and friendly and they welcome them into their lives. And then I say, "Great, subscribe to WIRED."

Well, look, we do have to take another break because we are running a bit long, but this has been a wonderful and invigorating and depressing discussion, so thank you all. I should let everybody know that you can read Kate's story about the research report showing how frequently AI Overviews have been showing up, and you can read Reece's story about how AI Overviews copied his work. Both of them are on WIRED.com.

Lauren Goode: Go to WIRED.com.

Michael Calore: You can also find Lauren's story, an interview with Google's head of search, Liz Reid. All of those links will be in the show notes.

Lauren Goode: That interview feels like it was seven years ago.

Michael Calore: But it was really, it was three weeks ago.

Lauren Goode: A lot has changed. Lot's changed.

Michael Calore: All right, we'll be right back.

[Break]

Michael Calore: OK. Welcome back. Now we have our recommendation segment where we go around the room and we ask everybody to recommend a thing they're enjoying that our listeners might also enjoy. Kate, we'd like to ask you to go first.

Kate Knibbs: Yes, of course. So I just finished a book that I think listeners will really enjoy. It's called Token Supremacy by Zachary Small. They're a New York Times art reporter and they wrote a really rollicking, fun account of how the crypto boom totally morphed the art industry into a freak show version of itself and I highly recommend it.

Michael Calore: Nice. Is it out now? Is it a new book?

Kate Knibbs: Yeah.

Michael Calore: Nice.

Kate Knibbs: Yeah, I think it came out earlier this spring.

Michael Calore: Great. That's a good one. Reece, what is your recommendation?

Reece Rogers: Balatro. It's this inexpensive indie game published by Playstack that I've been obsessed with for a couple months. The overall vibe can be described as a psychedelic poker game.

Michael Calore: OK.

Reece Rogers: During each run, you win by collecting jokers with bizarre power-ups that stack up, boost your score. It's kind of weird, kind of wild. It's a kind of game where I throw on my headphones to play a couple rounds before bed and next thing I know it's 3 am and I haven't been sleeping. Yeah, so check that one out. The game is called Balatro. It's only 15 bucks. It's available for PC and all the major consoles.

Michael Calore: So like Switch?

Reece Rogers: Yeah, definitely.

Michael Calore: OK.

Reece Rogers: Get that for your Switch.

Michael Calore: OK. Balatro. Thank you. Very good recommendation. Lauren, what is yours?

Lauren Goode: The way you said that was funny.

Michael Calore: Well, because I don't know what's coming. I never know what's coming.

Lauren Goode: I'm going to keep on my poetry kick, I guess.

Michael Calore: OK.

Lauren Goode: This is a book called Technelegy. It's like technology but with elegy. It's a new book by Sasha Stiles. She's a language artist and AI researcher who incidentally Kate here introduced me to a few weeks ago at an event, at the Internet Archive that Kate and I were both at. And I happened to be interviewing Sasha this week at another event in San Francisco. The book Technelegy is a collection of poetry and prose and some of it uses generative AI in either a supportive role or to generate language. And Sasha's work is interesting because it both embraces technology and I think spells out a future that can feel a little dark and surreal. And yeah, it's an easy read. I always say that and it doesn't sound like it's describing it accurately because it's a pretty profound read, but easy in the sense that you want to keep reading it.

Michael Calore: Nice.

Lauren Goode: Yeah. So I recommend Technelegy.

Michael Calore: Nice. Both embracing technology and painting a dark future, just like this show.

Lauren Goode: And this could be our new tagline I think. I like this. Let's workshop this. Mike, what's your rec?

Michael Calore: I'm also recommending a book. It's a brand new book. It's called Neu Klang: The Definitive History of Krautrock by Cristoph Dallach. Krautrock, terrible name, terrible name, but over the years has evolved into being sort of a term of affection for music made by German bands from the mid-60s until the mid-80s, maybe beyond, but that's generous I would say.

Bands like Can, I'm never sure how to pronounce that, but C-A-N, Can, Kraftwerk, Faust, Neu!, bands that you've probably heard of if you're into weird rock. Anyway, it's one of my favorite styles of music, and this is an oral history of the style of music. So the author spoke to all of the players who are still alive, the people who were adjacent to that world, people like Iggy Pop and Brian Eno, and really brought it home in a big tone. This is the first big definitive oral history of the Krautrock movement.

So if you're into it, the English language version is available in the UK right now. So I ordered it from Rough Trade and it got a little early, but if you want to wait for the US version, it's coming out at the beginning of August. So it's coming out this summer. So if you're a super fan, you can get your grubby hands on it quick and have it shipped internationally, or if you're an American and you just want to wait and be lazy, then you can wait.

Lauren Goode: Are your hands grubby from the Kraut?

Michael Calore: From the sauerkraut?

Lauren Goode: Yeah, sauerkraut.

Michael Calore: No, it's just from all the ink on all the books that I'm reading.

Lauren Goode: Sweet.

Michael Calore: Yeah.

Lauren Goode: Appreciate you.

Michael Calore: Thanks. All right, well, thanks everybody. That was a lovely discussion. Thank you both for joining us this week.

Reece Rogers: Thank you.

Kate Knibbs: Thanks for having us.

Lauren Goode: Yes, standard podcast answer. I love it.

Kate Knibbs: Should I start being rude?

Lauren Goode: No, that was great.

Kate Knibbs: Yeah. Fuck you guys, actually. I don't know if I'm allowed to swear.

Lauren Goode: That's staying in.

Kate Knibbs: Just kidding.

Lauren Goode: Yes, you are.

Kate Knibbs: OK.

Michael Calore: And thank you all for listening. If you have feedback, you can find all of us on AI Overviews. Just search for our names on Google. Our producer is the mighty Boone Ashworth. We'll be back next week with a new show. It is Apple's WWDC next week, so come back to hear us talk about all things Apple. And until then, goodbye.

[Gadget Lab outro music plays]