In Searches: Selfhood in the Digital Age (Harper Collins India, June 2025), Vauhini Vara explores how big technology companies have shaped and exploited human language and communication for their own gain. We allow this to happen, she argues, because we also benefit from the convenience of these products. It’s an exchange that makes us complicit in technological capitalism: we are both victims and beneficiaries. Across the book, Vara enacts this tension through her own digital footprint – from Google searches and Amazon reviews to dialogues with ChatGPT.

In a conversation for the Southasia Review of Books podcast, recorded in June 2025, Vara reflects on AI, online selfhood, grief and the possibility of imagining different technological futures even in an increasingly algorithmic world. Searches is ultimately a meditation on how language can be reclaimed, how we use it to assert ourselves, resist and imagine alternatives beyond the interests of Big Tech power.

The episode is available on Youtube, Spotify and Apple Podcasts.

This is an edited excerpt of the full interview transcript. Please listen to the corresponding audio before quoting from it. 

Shwetha Srikanthan: Searchers opens with a conversation in which you ask ChatGPT to respond to your writing. As an experiment, you planned to get feedback on a series of chapters you’ve written to see what it might reveal about the politics of both your language use and the models. And in a recent piece for The Guardian, you write that after Searchers was published, much of the media framed the book as a collaboration with AI rather than the critique you had intended. Could you talk us through this misunderstanding and how you’ve come to reckon with complicity in these systems?

Vauhini Vara: From the start, when I decided to feed sections of my book to ChatGPT and get its feedback, I wanted to do it to see what would happen, and to see what it would show about the relationship between me as a user of this product and the product itself, and my use of language and its use of language. And I said to myself, I won’t include this in the book if it doesn’t end up saying anything interesting. I felt like the model showed fairly quickly and fairly explicitly the ways in which a product like this uses language in ways that one might describe as manipulative in order to get something out of the user: the ways in which there are really insidious biases embedded in these products that then get passed on to users in the form of things like advice on writing or ideas about how a book might be made better. I wanted to show that to readers. I wanted to put that in the book so that people could see it.

In addition to that, I wanted the book to be full of commentary and critique of how these models function. So in my mind, in doing that, I felt like it was pretty clear that the implicit message of this exchange wasn’t: look how interesting it is to get feedback from ChatGPT, ChatGPT has all these great insights into me and my work. Because in my reading, it felt fairly obvious that that was not what was happening, right? That it didn’t really have anything insightful to say about me or my work. And that wasn’t really the point of the exchange.

So what really surprised me, like you said, after the book was published, was the way in which a number of reviews or articles written about it, or interviews with me about the book framed that conversation as a collaboration, or as my use of ChatGPT to help me express myself or assist in my self-expression.

At first it really upset me, honestly. My first reaction was: Oh no, how could this happen? This is the opposite of what I had in mind. Am I going crazy? How could my reading of this have been so totally different from the reading of so many very smart readers? I think that was what got to me. I’m always so interested in the various readings that can come out of any book, and that’s what I find really rich. And yet this particular reading felt so out of left field to me, so surprising, and yet a number of people had it.

But then, as I reflected on it more, it made me think about the ways in which my book itself is largely about my complicity, and our complicity, in the success of these big technology companies anytime we use their products. That was the subject of the book. Over time, I started to realise that an element of the dynamic that occurred when I published the book is that the book itself, in some ways, turned out to be complicit in the rise in power and the wealth of these companies as well.

Maybe part of what happened is that some readers read the text that ChatGPT produced in my book and, rather than reading it sceptically or cynically as I had, they read it at face value. So when ChatGPT said things like “I'm here to help”, or “AI models can do a lot to assist users”, I imagine maybe some readers actually read that and thought, okay, I’m going to internalise that and take it at face value. That’s one way I have of understanding why a fair number of people who read the book read it that way.

SS: There’s a popular argument that AI could never write like a human. But in the process of writing Ghosts, AI succeeded in moving you emotionally with a sentence about the death of your sister. You write that initially it seemed like GPT-3 was describing grief in a language that felt true to your experience, but the more you sat with it, you realised it was describing something that never happened and that any real significance was imagined. So over the course of, I think, nine attempts, you wrote a growing proportion of the text yourself by deleting the GPT words and replacing them with your own. You revised the piece and reclaimed the last lines for yourself. Tell us more about that process – what it revealed to you about grief, authorship and why AI ultimately couldn’t satisfy you as a writer.

VV: In writing this essay, I was using an OpenAI language model that came before the ones we’re really familiar with today – the ones that underlie ChatGPT. It was called GPT-3, as you said. I was interested in the way these companies promise that they can produce language when we humans find ourselves at a loss for language. And I thought of the thing that, for me, was really hard to find language for, which is the death of my sister when we were in college 20 years ago, and my grief over it, which is something I still struggle to find words for. So I decided to put that to the test – but not in a cynical way. I really was curious because this was pre-ChatGPT. This was early on in the AI conversation. I thought to myself, well, all right, if this company is saying its product can do this, let me see what happens.

I think ultimately part of what surprised me was the way in which GPT-3 produced language that I, as a reader, found – and still find – beautiful and meaningful and interesting and intelligent. I would call it those things if it had been written by a human, right? So I find it a little limiting when people say these kinds of products are never going to be successful in displacing writers because the writing is no good. I take issue with that because I think it’s a really limited argument. In writing this essay, using this earlier model, I found there were sentences it produced that one might evaluate as good sentences.

What then is the difference between an AI model and a human? And what I argue in the book is that the difference ultimately is really basic. It’s the fact that an AI model isn’t a human.

When we read a book or an article or an essay or a short story or a poem, I think that, at least for me as a reader, part of what I’m interested in is my relationship as a reader with the actual human writer who put the words on the page. There’s something significant about that relationship. The meaning that’s constructed on the page is partly constructed by me as a reader, but it’s also constructed by the writer who put those words on the page.

The difference with an AI model is that all the work is happening on the readers’ part. There’s no intention on the part of the machine that’s generating words, there’s no goal of having some kind of communicative experience with a reader because there’s no consciousness behind it. So the human on the other side – the reader, the listener, the user, as they call it – is the one doing all the work.

And then the question becomes: how much does that matter to us? For me as a reader, I think it matters a lot. But I think that’s something for all of us to decide together. It wouldn’t surprise me if we end up in a future where AI models proliferate, that can produce language that could be evaluated as high-quality language. And I think once we get there, we might want to have better arguments against using these products than simply – the language just isn’t that good.

SS: At one point, you discovered that Google had archived all of your search histories going back to 2005. It was this vast and intimate record that captured your everyday life – both superficial and deep thoughts – in a way that none of your diaries, texts or emails could. But Google still owns and benefits from that information, and you also, to some extent, give it permission to do so.

As you looked back at this archive, you felt unsettled but chose to reconstitute it as a piece of literature – as a way to reappropriate your own written language. How did it feel to confront and reclaim so much of your private self through this exercise?

VV: I felt conflicted because I think when I went into this exercise of trying to create an essay out of my own Google searches, my idea was that Google benefits so much from having this information about me, I let it collect this information about me, is there a way that I can reappropriate it for myself and define it according to my own set of artistic and ethical values? Literary values rather than the kind of financial value that Google as a company ascribes to it. 

That was my question. So I created this essay, and it was important to me to find language in my own Google searches for the essay that would read to a reader as moving or interesting or fun – that would have the qualities we want in literature, right? Because that, in my mind, is what would make it an essay and not just a collection of searches.

And I felt that I did that successfully in the end. I felt that I put together this collection of Google searches in a form that felt literary and existed on its own artistic terms, separate from the terms that Google uses to measure its value.

Yet, when I finished it and published it – first in the New York Times and later in the book – I then had to sit back and say: but wait a minute. At the same time, Google continues to benefit from the fact that it collected all this information about me and still has it, right? My writing of the essay doesn’t change that fact.

In fact, one might even argue that readers might read my essay as a celebration of Google. Google has these advertisements on TV where there’s this succession of people’s poignant Google searches as a way of showing how useful Google is in our lives. There’s a way that essay could be read as propaganda for Google.

So what interests me, in the end, about my Google searches and the attempt to reconstitute them in a literary form is the way in which any time I try to do that – any time any of us tries to use the tools of these companies to create some value of our own – it’s always still going to be bound up in the wealth and power of the companies themselves. I find that really frustrating, but I also find it interesting because writing an essay like that is a way of enacting the problem of existing in a world dominated by these companies.

SS: Speaking of reconstituting experiences and memories, I love what you did with images and photographs in the book. Could you tell us more about that? 

VV: I was interested in the way we use language as humans. I was writing mostly about written and oral language and then realised after a while that the book was missing something, which was visual language – the way in which, even before we ever started using oral language probably, we were drawing pictures, depicting the world in visual ways. So I became interested in that history and started doing some research into the history of visual communication for us as a species, going back 100,000 years. I realised that nowadays, in contemporary society, we talk about visual communication and visual art as being something totally separate from oral or written language, but they’re all part of the same general human desire to understand others and be understood, and to understand our world and find a way to express what that’s like, which is something very uniquely human.

So what I set out to do in the first of a pair of chapters was to tell the story of human communication through images from the beginning, and the way in which images – like the way we use oral and written text – are always a reflection of our culture and time in which they were produced. Invariably, because our societies over the past couple thousand years have tended to be more patriarchal than matriarchal, you increasingly see images that reflect patriarchal norms: the objectification of women’s bodies and faces, the prioritisation of patriarchal ideals. And so I tried to use images from our culture and images from my own life to depict that over the course of the past many years.

Then in that same chapter, I talk about how the most recent manifestation of that has been AI image-generation technologies, which basically take images that already exist mostly on the internet and use them to make predictive determinations about what images should look like and then spit out images based on that. So because of this recent history of a certain kind of ideal or norm being reified in imagery, it then follows that AI imagery is more of the same.

I write about that in one chapter and then I thought, AI companies keep claiming that just like their language-generation products can produce language for us, that their image-generation products can provide us with images where we are at a loss for images. I happen to come from a family where my parents grew up in villages in South India in a time and place where cameras weren’t common. So there are a couple of pictures from my dad’s life when he was young, and one that I know of from my mom’s life. And my mom is a huge storyteller. She’s always told me stories about her ancestors’ lives and hers, and what it was like to grow up in the time and place that she did. So I wanted to imagine somebody trying to reconstitute these stories in imagery. I was trying to imagine what it would look like for me to try to get AI to generate images that might go along with my mom’s memories. What I did was, in the form of a kind of fake presentation to investors asking them to invest in a company I was starting, I told the origin story of myself as an entrepreneur, and as part of that origin story, I told little vignettes about my mom’s life and then asked AI products to generate images to go with them.

The funny thing for me about that experience is that invariably, every time I tried, the images had nothing to do with the stories my mom told me. I also talk about my own life growing up, my sister, the loss of my sister, the things that we did together that we never captured in photographs, and it failed at showing those too. For example, when I talk about my sister’s arm, it depicts this very pale, long arm – and we are not white. When I talk about my parents’ marriage – my mom is fairly light-skinned, my dad is fairly dark-skinned – the image that comes up is this cartoonish image of very, very light-skinned Indians with somewhat European features standing next to each other at a wedding. So you can see through those images the ways in which the promises these companies make fail.

SS: Let’s go back to your early years on the internet in the 1990s. In the book, you recall trying on different identities and the act of searching itself. These days, we talk about the overwhelming presence of digital content – almost as if we have no real choice but to keep consuming these things, right? But back then, you write that it felt different, like early internet users weren’t so tied up in structures of power and profit. But as you point out, even before all this content existed, through the act of searching, people were communicating a desire for answers. Even your first use of the internet was bound up in the big tech and capitalism of the time. Were you surprised by that realisation, and did it change the way you look back on your early online experiences?

VV: I think it’s so easy – or it was easy for me – to think that, when I set out to write this book, I was going to write about how the internet changed from being this very innocent and somewhat democratic space into whatever the opposite is: a flawed and corrupted and undemocratic space ruled by big technology companies and their interests.

I started working as a tech reporter in 2004, and I started using the internet in the mid 1990s. When I sat down to research that period – the early to mid 1990s, when the internet emerged, when I was in middle school and high school and wasn’t working as a reporter covering it – I realised that what I had thought were these random occurrences, where all of a sudden one day at home we started using AOL and I would chat with people in chat rooms and encounter random strangers, and search emerged and I could use it to find what I needed – I had always thought of that as just a bunch of experimentation happening early on.

But I realised, upon researching it, that our first internet service was AOL, and what had actually happened was that in the early to mid 1990s, AOL embarked on this multi-hundred-million-dollar advertising campaign in the United States where they sent CDs and floppy disks to people’s houses all over the country and said: if you put this into your computer and download our software, we’ll give you a certain number of free hours of internet use, and after that you’ll have to pay a monthly fee, but you should try it out.

And I remember getting those CDs in the mail. Finally, I’m sure it was me and/or my sister who convinced our parents, “We should try this. It’s free for the first month. Come on. Everybody’s using AOL.” So even my first use of the internet was bound up in a specific strategic decision by a big technology company to get the internet into my house. And later, when I talked to experts on that period, they told me that the AOL campaign specifically was a really important formative moment in the explosion of the internet. So yeah, it surprised me.

SS: You write that social media isn’t meant to be a site of revelation – it’s meant to be a site of performance. For example, as a teenager, you found ways to craft your online persona. It was a way to position yourself how you wanted. That sense of self-styling still exists today and is so deeply tied to how companies profit from what we post, the content we consume and how we present ourselves online. You also describe this as a “subtle psychic violence” that we’re not even fully aware of. Do you think we’re doomed to be trapped in this loop of performing for the algorithm or collapsing into a single marketable digital identity? Or is there still hope for protecting selfhood in the digital age?

VV: I love that question. When I first started using the internet – there was no social media yet – there was this boy I had a crush on and I would send him these emails performing a version of myself that nobody would see if they just looked at me. It was the way I wanted to be perceived. And that guy, Ari Shapiro, was my audience. It was an actual human audience of one person. I was sending him emails.

The reason I bring that up specifically is that we went from that to now having the illusion of performing ourselves for actual human beings, but what we’re really doing on social media specifically is performing for the algorithm. So when we use text, we’re writing things in ways that we know will capture the algorithm’s attention. And this shows up in ways that seem as simple or innocent as when people have something interesting they want to share, or a request they want to make on social media to their friends, and then attach a picture of their dog because they know that’s what gets the algorithm’s attention. They’ll literally say in the post: “And here’s a picture of my dog for the algorithm.”

That’s not something we would organically do. I wouldn’t be trying to tell you something interesting that happened in my day-to-day life and then hold up a picture of my dog to catch your attention, right? That's not the way humans communicate. It is the way algorithms function.

You see this also on TikTok too – I’m not a big TikTok user – where people start talking in this certain way that the TikTok algorithm “likes” and amplifies. People see that and think, I guess that’s the way everybody talks now, and then they start imitating it. Pretty soon there’s a new way of speaking that’s catching on because TikTok’s algorithm has prioritised it. I think that already is happening, and there’s a danger of it happening even more in the age of AI. I’ve already heard of students turning in essays and teachers not being able to tell whether it was AI that wrote it or the student, because students have gotten so accustomed to the language of ChatGPT that they think that’s the way they’re supposed to write.

That’s all a preamble to say that I would not have written this book if it weren’t for feeling strongly that there are other options available to us. As a journalist, I’m always really interested in describing the world we live in. But part of the goal, I think, for any journalist in doing that is to make clear that there are other possible worlds we could live in. Part of the reason I’m so interested in my responsibility and complicity – our responsibility and complicity – isn’t to say that we should all feel badly about ourselves. The specific goal is to say: listen, we are using these products, we do have a choice. We’re using them in part because we’ve decided to, but then it follows that we could make other choices.  We could try to figure out what it is about these products that we genuinely find useful or entertaining or moving or interesting, and see if there are other ways to build technologies that retain all of those things while leaving out the parts we don’t like. Are there different models for building companies and technologies in the first place? Could governments use their resources to build technologies that we would publicly own? Can nonprofits and foundations put together their resources to build them?

Things like this already exist. I live in a town in the United States – Fort Collins, Colorado — that has built its own municipal broadband network underground. It’s more affordable than the private offerings available to a lot of us. And when you call for help, somebody actually picks up the phone. It’s a service the city is providing rather than a private company. There are also foundations like the foundations behind things like Wikipedia or Signal, which is an alternative to WhatsApp or Facebook Messenger.  So all these other models actually already exist, but we sometimes don’t notice them. And so I’m interested in all of us getting together and thinking: if we like some things about these products, what can we do in the future to keep those things and throw out the things we don’t like?

SS: Reading the conversation with your close friend Sanam Emami about Amazon reminded me of similar conversations I’ve had, for example, around the BDS movement to end international support for Israel’s oppression of Palestinians, and the question of how personal choices are tied up in larger systems of exploitation and resistance. Sanam’s argument for boycotting Amazon was about productisation and exploitation. You argued that real change would need to come from government policy – that individual boycotts can’t make a dent in Amazon or take down Jeff Bezos’s empire. But later you came to see that Sanam wasn’t claiming they could. She was making a case for living according to one’s own values. Looking back, how do you feel about that exchange, and how has your approach to individual ethical choices changed since? 

VV: In some ways, I think that conversation encapsulates everything the book is about. Some people have read the book – and I guess because I’m the narrator – and people have read it as a kind of permission to use these products and not feel too bad, because here’s somebody writing about how she knows the problems with it but is still using these products. But that wasn’t my goal. My goal was, in some ways, to be the villain in the book, right? And Sanam’s position in that discussion is meant to be the more heroic one. I think maybe the reason that can sometimes be hard to see is that one doesn’t necessarily expect the narrator of a book to position themselves as the villain. But I very much see myself as playing that role.

The reason I use such a strong word – villain – is that I think it’s very easy to fall into the complacency of saying: my individual act can’t change anything; somebody else needs to do something. Because that leaves us in a place where we can feel faultless and blameless and just go on with our lives without changing anything. And it feels important to me to put that pressure on ourselves to change things.

I am in the process of trying to divest from some of these big technology companies. I no longer use any social media platforms except Bluesky, for example. I’m trying to pull away from using Amazon. 

It’s interesting because going through this exercise and writing the book really put me in a place where I became interested in figuring out how I could do this. But it’s also interesting to see how quickly you start finding the ways in which one’s individual choices impact others. For example, I keep saying to myself: I need to get off Amazon. But then I think – my mom always uses my account, what’s she going to do? Or for example, I wanted to cancel my X account a while back, and when I talked to somebody from my publisher about it, they said: “You can’t cancel your X account – not now, with your book coming out. You have 15,000 followers.” So I waited a couple more months after the book came out so they wouldn’t get upset about it, and then I cancelled it.

So I think I’m interested in putting pressure on myself to see – how can I do these things while still being respectful of the people around me, who have their own needs?  It’s not easy. It’s a difficult balancing act, but I think it’s a very worthwhile one to attempt.

SS: In the book there’s the project of writing and translating an essay in Spanish about communication. You reflect on your relationship with different languages and the fear of not being understood. You note that even if we were to invent a universal language, full understanding would still be out of reach. And interestingly enough, there are people working to solve this using technology. Can you also tell us more about how your first novel, The Immortal King Rao, dealt with this, and the way you connected the limits of communication in both human and machine form in Searches?

VV: As a writer – but maybe also just as a person – something I find myself always coming back to is the very natural human desire to understand other people and be understood by them, and the barrier you come across given the fact that our brains aren’t connected to one another.

So in my novel, I created this device, this technology through which the narrator of the book is able to access the consciousness of her father. But it turns out that there’s something limited about that access itself. Even with a technology like that, she doesn’t ever attain the kind of full understanding that her father might have hoped she would. In fact, she ends up going in a direction that’s probably quite counter to what he might have hoped for from his daughter.

In this book, and in that chapter about language in particular, I was interested in delving into that in a non-fictional way, into the difficulty of communication, but then also the paradoxical way in which attempting to communicate, even knowing that that barrier or friction exists, is what makes human communication so beautiful and meaningful.

The fact that we keep trying and trying and trying, even though we know communication can’t be perfect.

So writing an essay in Spanish – at a time when my Spanish was relatively limited, and it’s still limited now, though better than before – was a way for me to enact that effort. To show a person, that person being me, trying to express herself, fumbling and failing, and yet hopefully conveying something meaningful in the attempt itself, even though the use of language is quite limited. And then I used Google Translate to translate it into English and showed that on the page as well. I wanted to show how a tool like that can, on a superficial level, aid understanding, but not on any deep level, and in fact can impede understanding once you go beyond the basics.

SS: Searches ends with a chapter based on responses from a survey you put out asking what it’s like to be alive in the world, specifically through the voices of those who identify as women. The outcome is so powerful in its specificity. It highlights the magic of language and the human ability – especially among women – to tell their own stories. It also comes across almost as a resistance to the flattening effect of LLMs. And even though the future can look bleak, readers have told you that the book made them feel both complicit in Big Tech’s rise and moved to act because of that awareness. 

So where do we go from here? Do you think there’s a real chance at pushing back or reclaiming some power in this struggle against Big Tech?

VV: I absolutely do. Yeah. I think one thing that’s difficult about that feeling of hopelessness, or the sense that the future is inevitably a Big Tech-dominated future, is that the incentives are all lined up for us to think that. What I mean is that big technology companies benefit when we feel that way, right? They would like for us to imagine an inevitable future dominated by their power. And so there aren’t really big institutions telling us: hey, let’s not have that future, let’s imagine other possibilities.

So it’s incumbent upon us, people and our communities to come together and say, this is a real possibility, let’s make it so. I quote in the book the writer Ursula K Le Guin, who compares technological capitalism today to the era of kings, when monarchies were the dominant form of social organisation. And she points out that at that time people probably also thought, this is the way things will be forever. This power is so entrenched that it can’t be otherwise. But then people came together and decided to make a change. They came up with new systems of government, whether democracy or socialism or communism or anarchism or all kinds of other forms. That happens in history all the time. And so I think it would be a shame if we were to sit together where we are now and say: okay, I guess this is where we’re destined to be forever. Because that actually wouldn’t be in keeping with how history has worked in the past.

There’s often a kind of pendulum effect where we move too far in one direction and then there’s a correction. Human imagination often does prevail. We come up with ways of doing things that nobody had thought of before. And the reason we can do that, I think, in large part, is because of our use of language – to communicate with one another and come up together with new ideas that we haven’t thought of before. So I think there’s very much a possibility of that in the future.