Vikram Chandra on the search problem, twisting patterns, Lawrence of Arabia, and working at a higher level through AI

Vikram Chandra is a writer, startup co-founder, and professor at UC Berkeley. He can be found on his website and as @typingvanara on Twitter.

Could you tell me a little about your background?

That’s a big question! Well, I was born in Delhi. For my first five grades I went to Catholic schools. It was actually an interesting experience – St. Xavier’s and Rosary High School. A lot of my friends when I was young were young Indian Catholic kids with names like Benedicto Fernandes. That’s the India I grew up in, which was kind of naturally multicultural, at least in an Indian sense, because we have so many distinct cultures.

What we spoke at home was Hindi – we got The Times of India and my parents can speak English, but at home we always spoke Hindi. So I do remember learning English very consciously, the alphabet and then reading books. And now at home we are what you would call bilingual – we switch in and out without realising we’re switching in and out. It’s funny watching my mother when she’s talking to anybody who’s not a Hindi speaker: she’ll start speaking English and then she won’t realise that she’s going into Hindi.

My father was in a corporate job, so we got transferred all over the place. So I lived in several parts of North India and then because of all this travel I went to boarding school when I was eleven.

The school had a very distinct culture of its own. Originally it was set up by the Brits for the sons of Indian royalty to train them in being proper British subjects. They brought over Eton to India. It was very macho – Rajputs are the warrior caste of India. So it was Rajput machismo blended with British stiff upper lip and all of that muscular British stuff. Then I started writing fiction and publishing in the school magazine. I published a story in our student-edited school magazine, and that was my first publication. I was always a really nerdy little kid and I had a big world inside my head. I used to make up stories inside my head and then I started writing science fiction – I was a very avid reader of Asimov and Arthur C. Clarke. A friend of mine read these stories and he said, why don’t you give this to school magazine? And those got published and of course that was very validating. I wrote about this a little bit in Geek Sublime.

The big sports heroes of the school came up to me and said, we like that story. Which of course was very pleasing for a nerdy kid. So that’s when I started to write. I always knew I wanted to write. My mother is a writer – she’s a screenwriter and fiction writer. In our house I grew up watching her write on their kitchen table. The thing about the educational system in India was that I couldn’t take workshops. And I knew in the US you could take writing workshops as an undergrad. So I made it to the US as an undergrad and started taking fiction workshops. But then I was too scared of trying to make a living as a writer, you know?

I don’t even try.

My mother, as I said, was working in film. So after I got my undergraduate degree, I thought, well, at least that’s an industry, right? Nepotism will get me a job somewhere. (Laughs). So I went to Columbia Film School with the intention of doing screenwriting and directing. And I figured out very early I’m not built for the film industry. It’s too collaborative for me. I like it, I’ve worked in it since, and I’m doing some work on a screenplay right now, but I have to keep a distance from it.

While I was at Columbia, I was wandering around the library and by chance I found the translated autobiography of James Skinner, an early 19th century soldier, half-British, half-Indian. He’s a legend. When I found that book, I heard had his name. I had two uncles in the Indian Armed Forces, and Skinner had founded a cavalry regiment called Skinner’s Horse. So I knew about him, but very little. I took that book to my dorm – it hadn’t been checked out in, I think, 60 years. And then I just got obsessed with it. I knew it couldn’t be a film. It was too big, the way I was imagining it. And that became my first novel. I went off to and did two MFAs, two graduate degrees in writing, and used that time to write that first novel.

And it worked. Boy, did it work.

I’ve been very lucky. Sometimes look back at my life and think, God, you’ve been lucky.

I think it’s more than luck. But I’m so glad you got there. Geek Sublime, which you just mentioned, is one of my favourite essays. In it you talk about yourself as a writer and also as a coder, a programmer in relation to the Indian tradition of thinking about grammar and philosophy and aesthetics, particularly with regard to Pāṇini and Abhinavagupta. And now you have developed a software program for writers, Granthika.

I felt a need. I did a lot of research for my first novel and also for my contemporary cops and gangsters novel, Sacred Games. I get very obsessive about research. I do lots of interviews, I like to go and meet people in their environment. So, for Sacred Games, I would go and meet policemen in the police stations and I’d talk to them, I’d be writing in my notebook. I accumulate notes and photographs and academic papers and news clippings over the years. And then when I’m writing, I’ll suddenly realise, okay, I interviewed that guy seven years ago, and I know I’ve got this, I need to find that interview.

The cops and the gangsters and the spies, you can’t record them as we’re doing it right now. They would get nervous. A friend of mine, a senior cop in Bombay, got recorded by a journalist without him knowing it. The guy had hidden a small recorder in his bag. So everyone was getting paranoid. I would make sure that the people I interviewed knew I was only writing in a notebook. So over the ten years of researching Sacred Games I took my notes in all those notebooks..

There were about 22 of them, those fat little reporter’s notebooks. Remember those? Thousands of pages in all. So, how do I find an interview I did many years ago in all these pages? Information retrieval, it’s an ancient problem. That’s what the ancient libraries were intended to solve.

The search problem.

The search problem. So what would happen is I’m writing and I realise that I need to look something up, and then I can spend half a day just looking for something that I know is somewhere, and I would sometimes find it. The annoying thing with traditional word processors is that when you create or collect your data, there’s no way to connect it directly to the text. Unless you write a little note in the middle or add a comment. So when I finished writing Sacred Games I had downtime. So I thought, okay, now I’ll try and look around. Surely somebody’s written software that can deal with this. And nobody had. So then I got interested, why has nobody done this? And it turns out that attaching knowledge to text is a more difficult problem than one would think. As a non-professional tech person, I didn’t know this. So then I began researching the problem, thinking about it. And I eventually started to see how the problem could be solved, at least in the general area, how you could solve that as a technical problem.

My friend, Akash Kapur, is a nonfiction writer. He’s also one of these people who’s a writer but has a foot in the tech room, and he’s done a bunch of startups. One of his investors is Roy Bahat, who works at Bloomberg Beta. And Bloomberg Beta is the venture capital arm of the big Bloomberg, and their mission is to invest in anything that changes the nature of work. I had lunch with Roy and he said, I understand the problem and I’ll give you some money, but you need a co-founder. I was like, yes, I need a co-founder. I started to learn programming when I was in grad school, but I’m completely self-taught. I’m very like, what is the word? A drudge, somebody who can put together simple database apps. (Laughs) I’m not somebody who would get hired at Facebook or Google. So I was looking around for ideas, how to solve this problem. And there was an open source database, HypergraphDB, that I found on the Internet. I wrote to the guy who created it, Borislav Iordanov, asking him some technical questions, and he very nicely answered.

And then he said, what are you going to use the database for? After I had finished Geek Sublime, I was on book tour, and so I had used that time in hotel rooms and airport lounges to write up a software proposal.

As one does.

(Laughs) As one does, right? So I sent that to Boris, and Boris wrote back saying, if somebody gives you money, I’m in. I’ll do this with you. So that was very cool, and Roy gave us money and we started.

The idea then – if you look at our app, there’s an editor where you write your manuscript as you do in Word. But along the top you also see these other tabs, which are characters, events, objects, and locations. As you write, you can link your text to a character. If I’m on a character’s name, with a single click, I can jump to his or her page and then jump right back. It’s an attempt to make information management related to text easier to handle. And at least for me, it saves me a lot of time and energy. I’m now writing the current novel that I’m working on in it, and I’m trying to be as careful as I can. Anything that I can think of in relation to a person or an object or an event or whatever, I’ll make notes about it right in there.

Because, like I was saying, what’s frustrating is that when the time come for it to be published and you do a fact check, you can’t remember where your fact is. “Did I write it down? Where is it?”

So essentially, it’s semantic web technology. You’re attaching semantic meaning to things.

Exactly. The Semantic Web is based on XML, and XML has a tree structure. But human language, in terms of meaning, is not a tree structure at all. So you’ll run into the problem of overlapping ranges. There have been attempts to solve that within XML and it gets very complicated very quickly. So that’s what I was thinking about when I first started researching this. I was like, okay, here we have XML, you can attach metadata. Why doesn’t somebody build an editor in which one can do what I’m talking about? But once you start trying to get around overlapping ranges, you start building very complex structures. So what we are doing is with Granthika is that we store the text as a hypergraph and then we layer other hypergraphs on top of that. The overlapping ranges problem goes away. But the engineering to do this was very tough. Boris is a tech genius, so we are able to do it on a very small budget because Boris was able to execute it. So, as I was saying, it’s very useful for me.

There’s another thing which I’ve always had trouble with, and often, because I write against actual events, I write fictional events set in the real world. And for writers, keeping track of time is very, very hard.

So there’s these bomb blasts in Bombay. I need to know – where was my fictional character then? What was happening in her life? In Granthika, if you set up an event and you tell it happened on this date at this time, it automatically puts it on a visual timeline. And that becomes really useful because you can see relations of events to one another, what overlaps something else. Again, I think people who don’t write don’t realise how difficult a problem that is. Even if you’re writing a 200 page novel set within two weeks, everything has to line up, time management becomes quite tough. That’s one of the things of Granthika that’s really useful to me.

I wrote a novel set in 16th century Ireland. I had to do the research for that, and I wish I had that back then because as you say, it’s really, really hard to keep all this information. So Granthika is essentially a worldbuilding tool?

Yes, that’s exactly right. When Sacred Games was made into a Netflix series, before it was made, I visited the production company in Bombay that were doing it. Sacred Games was published in 2000 and the events in it end in around 2002. When we first started talking about the TV series, the big question was, do you move it forward to the mid-2010s or do you do the entire thing as a period piece? Even for just purely logistical reasons, moving it back in time even by a decade would have been really tough to shoot in India. Because everything is changing. You would have to be careful about the ads that show up on an outdoor shoot, what kind of phones people were using.

Right, yes, I was thinking of the technology.

It would have been a nightmare. Especially in India because liberalisation started in the mid to late 1990s and then the landscape completely changed. It felt like every month you were seeing new things. So the production team said, okay, let’s move it up. So I walk into the office and there’s a big conference table in the writers’ room, and they’ve got a big whiteboard where they have every character’s birthdays attached. Because now we’re going to move everything forward by 10 years or 12 years. If the cop becomes 12 years older, then he couldn’t have been a junior cop who showed up as a first responder on the bomb blast site. They were shifting times, all these chronological questions. It was an interesting problem to have.

You are the executive producer of the series. Did you have to do this yourself or did someone else do it?

They all did it. I was asked early on whether I wanted to be in the writers room, but this was all pre-pandemic. Nobody had gotten used to the idea of daily collaboration on Zoom. I have two young kids, they’re now 13 and 15. I would have loved to move to Bombay for a year and then work on it, but obviously my life is now built around the kids’ lives, and I definitely wasn’t going to camp in Bombay for a couple of years, far away from them. What ended up happening is that the writers would send me drafts of each episode and then I would send them hundreds of notes, literally, and they would ignore 99% of them, as they should! (Laughs) It was really fun, watching them go translate a book into a visual medium, they did an amazing job. It’s a very hard book to adapt. Just after the book was published in 2005, my agent told me that a company in LA wanted to buy the option to make a movie. I thought, how are you going to make a 900-page novel into a movie even if it’s two and a half hours long? I had lunch with the guys from that company in LA, because I love the movies that this company made.

I said, okay, if you want it, you can have it, but I had doubts about whether an adaptation could really work. They hired this really talented British writer to work on it, and then, after a couple of years, he’s done a screenplay in which two timelines go in and out of each other, you have this large universe of characters. But my impression is that the company got cold feet and let the option go. Then the streaming revolution happened, and prestige television had access to a worldwide audience. Netflix had proven to itself with Narcos that they could do a show set in one culture that could sell and be watched elsewhere. For us, all these pieces came together at exactly the right time, and they made a series. So it all just fell together.

And it was great because the writers and directors made a brilliant adaptation, they found things in the book that I hadn’t seen. There’s a character, a transsexual character, Kukoo, in the series, who became hugely popular. In the book, she never appears on stage. Sartaj and Katekar, the cop, the protagonist, and his partner, are on stakeout and they’re very bored, so they’re telling each other stories. One of the stories they tell is about this woman who dances in a bar, and all these men, including this very macho, tough policeman, fall in love with her.

Kukoo doesn’t even appear on-stage in the novel, she’s a person who a man tells a story about. And then somehow the writers found her and they made her a major character. They made her the main gangster’s girlfriend. I’m so happy that they did that. I couldn’t have done that, because I had already imagined it one way once. Kubbra Sait, the actress who played that part, became an overnight star. I’m so happy for her. She recently was in that Apple adaptation of Foundation – she had a major part.

It was interesting watching the show travel abroad because we all wanted to believe that we were making something good, but to see it go abroad, especially, and even in India, to have that kind of impact was amazing.

One interesting thing for me is that Sacred Games is, in a way, a Tolstoyan novel. It’s huge. It’s seething with characters, like a classic 19th century novel. But also, it’s full of Indian narrative devices. For example, nestled stories within the story. And also there’s a circular dynamic to it, constant recurrence.

One of my frustrations as a writer is that I can never plan architecture before I start writing. So I discover it along the way. And then it’s terrifying because you can be hundreds of pages into manuscripts and you’re like, what is this thing doing? I discovered that form as I wrote.

(Shocked) Are you telling me that you didn’t plot Sacred Games?

No, not at all. Not remotely.

Sartaj, the cop, first appeared in my short story book, Love and Longing in Bombay, so I knew him. But at the beginning all I had was an image of this gangster who has barricaded himself in this strange bunker-like house. I didn’t know who the gangster was or what was going to happen.

(Still in shock) And you figured it out as you wrote?

Yeah. What happens is I start thinking about the characters and imagining them, and then I start learning things about their past. Who the person is, where they come from, what they want. The frustrating thing is that it’s exciting because I’m learning things that I didn’t know. But, especially in that book, there were branches that went on for 70 pages that I finally realized were going nowhere. So in later drafts I ended snipping them out, but I wrote all those pages first. Somewhere along the book, I finally figured out that the conceptual architecture was a mandala. In the mandala, often you have these juxtapositions – like there’ll be one segment that will feature one character, or a part of a narrative? And next to that segment there will be another narrative. And there seems to be no obvious connection, but by putting them together in the same frame, the artist suggests that all these narratives belong together, the artist makes a cosmology around the entire thing and constructs meaning. Then I thought, oh, that’s what it is. There are all these intertwining stories, these many lives that are connected somehow, and that’s what the book is about.

The marketing for Sacred Games often describes it as a cat and mouse game between cop and gangster. But wait, the gangster is already dead and they only meet once in the book. No, twice – one is at the bunker and one other time, but that other time the cop doesn’t know that he’s met the gangster, because this man’s in disguise. And yet they profoundly change each other’s lives. So all these lives that are intertwined, that ricochet against each other, that affect each other – the mandala became the structuring visual figure and then also the entire book moves in a circle as well.

It’s not that I do it consciously, but this idea of circling repetition has always been with me. I don’t quite understand why. And in the fiction that I’m writing right now, it’s again there, although in a very different form, this idea of echoes. It’s kind of obvious, if you think about it, that history echoes forever. Things don’t go away. We think things have died in our own lives, but what happens to you stays with you forever, and sometimes in buried ways. But there it is.

And the interesting thing is that the repetition is constant, but it’s never quite the same. It shifts every time.

Yes, exactly. It’s like a spiral. Like I was saying, I don’t know why it’s been with me, but this shape has always been with me right from my first novel.

In Geek Sublime you talk about dhvani, the resonance that a text has for a reader, through rasa. I found the concept of rasa really interesting – the meta-experience of experiencing yourself experiencing a stable emotion. A circular structure like a mandala lends itself so well to that because you are in it and at the same time you’re experiencing it on a sort of meta level.

Especially in my first novel, Red Earth and Pouring Rain, there’s kind of meta commentary which is very common in pre-modern Indian texts. There are stories within stories, and often people in an outer story, a character will break into the telling of a story, and they’ll have commentary on the inner story, and then you go back into this inner story in the frame, and so on.

Pāṇini, actually, wrote this incredibly concise grammar of Sanskrit, which has meta rules. Which is also, in a way, what you’re doing with Granthika. You’re establishing meta rules for your universe, which you can then transfer into other mediums.

Right. With hypergraphs.

To go back to the adaptation of Sacred Games, a TV series is a linear medium – one scene after another. Do you think this circularity aspect of the book has translated well?

Well, they got the idea. Before I saw it, I thought that it would be incredibly difficult to do the intertwining of the two narrative tracks. Which they did, brilliantly. Not just in terms of breaking up and picking what to put next to each other, but even in terms of what shot are we going to cut from one thing to the other. That was one problem they solved very well. And if you watch the entire thing, again, they get that sense of ending where you begin.

Without giving out any spoilers, at the very end, they cut to a shot of the young gangster standing in the boat setting out on his journey. I thought that was brilliant. And then the song that’s playing in the bunker in the beginning comes and the lyrics of song in Hindi are literally saying “leave your story behind in the world”.

And again, that song has a particular cultural resonance because it came out at a particular time in Indian history. Indian film songs are kind of like our cultural equivalent of rock music. When you hear a song, it brings back the smell and taste of a particular era. So all of those choices that they made, I think, were really great.

Because, of course, in filmmaking there’s also a grammar, a visual grammar, which, as a writer, I find fascinating.

It is very different. When I worked on scripts, it was hard thing to go from one to the other. My brother-in-law, Vinod Chopra, is a big-time producer and director, very successful. And the way he met my sister is that I’d worked on a film with him. I met him before my sister did. I was working on a script with him, and I would write scenes and bring them in to him. My chunks of dialogue would stretch across the page, and he would just redline them. And he was right, because a minute or two minutes in screen time is an enormously long length of time. In novels, I have the luxury to float for 30 pages in one scene, but in film, you can’t do that. You’re right – a script is just a suggestion for a medium that has a grammar, a language in which the semantics are constructed through depth of field and sound and camera frames and the length of shots.

I’m delighted that the Sacred Games series works so well because it’s a really hard novel to adapt.

Like I was saying, we had some incredibly good luck in the casting, both the two leads and everyone around them. That’s the strange thing about a collaborative medium like film. Sometimes you put together what seems like a perfect project and it just doesn’t work. And it’s not because people are fighting on set or anything, but somehow the chemistry doesn’t happen and it just falls flat. And sometimes, like now, all the elements just fall together.

And you’re back to novel writing now?

Yeah. I’m working on new fiction. Grinding away at it. I’m very slow. Very, very slow. Like I was saying, with kids, life gets more scattered anyway, and then there was the pandemic and all kinds of other stuff happening in my life. But it’s interesting also how one adapts to these changes. Earlier, I was only able to write by myself, in a study that I had always imagined and was finally able to build in the mid-2000s. But now, after the pandemic, I was unable to write in that room. It felt too lonely, too isolated. Now, I like writing in coffee shops. I have actually become one of those people who take up a table for hours. (Laughs). I take my laptop to my round-the-corner coffee shop and write happily and all the kids behind the bar know who I am. We say hi.

I recently moved to a new house, which is about 20 minutes away from this coffee shop, as opposed to three minutes away. I was telling my younger kid, there’s another coffee shop near my new house, I should go there. And they said, no, you have to keep going to the other one, they love you there. So that’s where I work now.

Now you get to exercise before you write.

Yeah, forty minutes of walking, at least, every day. And it’s fun. I think we forget how malleable the human mind and your body is. You get used to stuff and then you think that’s your life. But then you can change very fast.

In Geek Sublime, you said that writing is not exactly pleasurable for you, that it’s a bit of a purgatory – not quite hell, but a purgatory. Is it getting any better?

(Laughs and shakes his head ruefully). No. Despite my coffee shop friends. Like I wrote in the book, there’s lots of writers who feel like this, that writing is hell. I think it’s because there’s the effort of trying to write a good sentence, but also – I don’t know if it’s particularly fiction or poetry – you have to be in a kind of meta-awareness so you can shape the sentences, the language. I’m trying to be in a scene and at the same time I’m trying to be outside of the scene, living it and constructing it simultaneously. So there’s a kind of split, I think, that’s hard to maintain for any extended length of time. It wears you down.

I also see it in actors. When I was at film school at Columbia, it was a requirement that even if you wanted to be a writer, you had to take an acting class. It was an amazing experience. The idea was that by being on the actor’s side of the camera, you would learn what the actor’s problems are and be able to be a better screenwriter and director. My teacher was Brad Dourif, who won an Oscar for his role in One Flew Over the Cuckoo’s Nest, who was wonderful in Deadwood and many other films and series. Trying to act certainly taught me a lot, both about the craft and what it costs to be an actor. And I think there again, there’s the same swing between trying to live another life while watching yourself, to be aware of what you need to do to be that other self.

And then actors famously have to go down deep and find their own trauma to work with. I say this with all love, but actors are as crazy as writers.

Absolutely. Why would anyone do that to themselves?

It’s not pleasant while you’re doing it, acting or writing. But there’s a lot of satisfaction that comes from it.

Well, you went to a Catholic school. So you know about the Catholic tradition of suffering – passion as ecstatic suffering.

(Laughs) I do!

So, as a writer and as someone who has developed a piece of writing software, what are your views on this whole AI thing that’s going on right now?

Being in the trade, as it were, it’s right on top of my mind. It’s interesting, we’ve been talking about this for years at Granthika, about how do you help the writer apart from the worldbuilding aspect of it. I’d always wanted to tackle the problem of creating fresh similes and metaphors.

So what we were talking about is that you do two or three similes and then you tell the system to generate 100 more. And the AI will come up with some really crazy stuff. But maybe in those 100 similes you’ll find two that are interesting. What the guys have done at OpenAI is that they’ve solved that problem on a massive scale, essentially.

Everyone thinks of AI as somehow magical, but it’s next-word prediction. What it’s doing is very sophisticated guessing based on an incredible amount of data. So what does that do to the writers’ trade or craft? I’m sure you heard that there’s a sci-fi magazine that had to close down their submissions because they were flooded with AI-generated stories.


Right. I think AI is going to be able to produce facsimiles of human texts in whatever genre. These imitations will get better. And so the question everyone is fixated right now is, will you be able to tell the difference finally, between writing by humans and writing by AI.

There was this brouhaha on Twitter because a romance writer revealed that she had been using generative systems to write her last three or four books, and her readers had been happily reading them. Her argument is, if my readers are happy, why aren’t you? So in that context, I’m thinking, okay, we’re going to have art which has to reveal that it was made by a human hand. It’s going to be like certified artisan cheese – this organic cheese was made locally with local cows, no big corporations involved, it’s not processed food. Is a novel going to have a certification? “No AI was used in the making of this.” So I think there’ll be some amount of that.

What we crave, I think, from poetry or fiction or drama is a replication or an echo of our existence. And pleasure that comes out of that, entertainment that comes out of that, but also a kind of human connection. Knowing that there is another human back there who is writing about pain or joy and you’re communicating with a human through the text. I think that’s inevitably going to stay.

I’m reading right now this amazing book called You Could Make This Beautiful by the poet Maggie Smith. She writes about the coming apart of her marriage, her divorce. And one thing about it is the form that she’s using, which is not linear story-telling, it works like a series of images, which are not necessarily connected in terms of causal plot. When I wrote Geek Sublime, that was the first time I hadn’t used a linear form. It was terrifying for me because I didn’t know what I was doing. But this idea of breaking a novel into small fragments and doing this kind of mosaic structure, she does that so well. I really admire what she’s done, it’s absolutely beautiful writing. But there’s also the confessional aspect of it – she’s revealing to you her own pain and confusion and her attempts at recovery. That’s a really important part for me as I’m reading the book. If somebody told me that AI had produced that, I would be very angry because I would then feel I was taken advantage of. Not in a commercial way – I don’t care that I paid whatever I did for the book, but you seduced me into an emotional connection and you didn’t tell me the truth. Her book is about her divorce, her husband cheating on her. What I’m connecting with is that feeling, that pain, which she embodies in that book. AI will be able to imitate that text, to produce some version of it, but I’d feel like I had been ripped off. 

I think that it’s going to be interesting for filmmakers and film audiences. When I’m watching a film nowadays, and a lot of it is consists of CGI, I don’t really care about that except where I can see that it’s really bad CGI. But I know that films are made on sets, and I don’t really care about that except where I can easily tell that it’s a set. We’ve already got AI actors, we’ve had AI rock stars for a while. I have to admit that I don’t really feel a connection with these artificial beings. In animation, I know that the voices come from humans. If I know that an entire film is written by an AI, and made by this AI, will I engage with it? I really don’t know. Maybe my kids, or their kids, won’t care. We’re at an event horizon, and it’s very hard to know what’s coming.

At least for now, I need to engage with the humans behind the curtain even as I engage with the fiction. I know that the living dream I enter is made by humans. And I understand that effort and all that work.

One of the most significant experiences in my life was watching Lawrence of Arabia. I was about six and I still didn’t really understand English. I fell down at home before going to school, I hurt my knee and I made such a big fuss I was allowed to stay home. My mother and her friend had booked matinee tickets for Lawrence of Arabia. So I went with them and saw the film. I didn’t understand English then, but it was an amazing aesthetic experience. I’ve rewatched it many, many times. I’ve recently watched a documentary about Peter O’Toole. So I was thinking about what an amazing film it is, for all its Orientalism and white saviour tropes and the lack of women in it. It’s such an amazing feat of filmmaking and acting and music. Every element falls into place. Miraculous.

And maybe I appreciate the film more because I’ve been on the filmmaking side of it. It’s so hard – it’s a physically strenuous process. Shooting in that environment must have been crazy. You’re out in the middle of a desert in the heat and you’re doing retake after retake. Peter O’Toole is being directed by David Lean, who was not a nice man, at least on set and location. Like many directors, he was ruthless about getting it right. (Laughs) I wonder if somebody made the current equivalent of Lawrence of Arabia and they did it entirely on their laptop, and it was equally aesthetically interesting, what would my response to it be? Because certainly my internal mythology about that film includes the fact that these guys were out in the middle of the desert doing this insane thing. So if they were all sitting at home on computers doing this, with AI-generated actors and landscapes and music, would my aesthetic pleasure be lessened if they produced something brilliant? Perhaps not. But my emotional attachment to the story of the making of that art would be different.

I think in the craft of film and our response to it, we are already used to the idea of technology as an intervention. So what is that going to do in terms of books, of writing? Will I really care that somebody used AI-generated characters? Somebody’s already done this a while back, in a sci-fi book titled The Salvage Crew. The writer, Yudhanjaya Wijeratne, is a programmer. So he wrote up a Python script to generate aspects of his characters, and went on from there. He sent his manuscript out to editors without telling them that he had done this, and the book got published and got reviewed. Well, so do I care? I don’t.

What I thought to me when I first heard about this, and this was before ChatGPT, was that, oh, that’s a really cool use of text. Now we’re in this stage where we’re saying that any use of AI is lessening the value of writing or fiction or literature. I don’t think that’s a useful response, even though I’m as confused as everyone else.

I do think that people tend to underestimate the extent to which AI requires human input.

I think that’s exactly right.

Starting with data input, you need people to tag the information, and so on. You gave the example of getting lots of similes, but there needs to be a human being picking the right ones, because the machine doesn’t understand language. It doesn’t know what it’s doing. There’s no cognition there.

Yes, exactly so. I was just at breakfast talking to a bunch of writers who have come to the festival [the Jaipur Literary Festival in Spain]. So in Silicon Valley right now, there are already people teaching classes in prompt engineering. I used that phrase, and none of the writers knew what that word meant. In the context of what you’re saying, new skills will allow humans to work at higher level. Programmers already see this happening in their work. You can get AI to generate code, ask it how you can make the code better. You can give AI a picture of a user interface and it’ll translate that into code. So as a programmer, I don’t have to do that grunt work. I get the machine to generate the UI, and I check the correctness of what it’s done.

In the domain of computers, we’ve already been doing this for a long time. A high-level programming language like Java or Dart is translated by the machine into an intermediate form and then finally into binary code. The programming language that a programmer uses every day allows her to work at a higher level, to not worry about the lower levels. Now AI will give me the ability to work at a higher level still.

Our comfort with these higher levels is going to be naturally based on age, on prior experience. My kids are so naturally at home in these waters that they will see the future. I don’t think we oldies can actually see what’s coming up completely at this point.

And then I don’t know why this is reminding me… I’ve been thinking a lot about – do you know Ian M. Banks, the sci-fi writer?

Oh, yes.

I love his work. Every two years I read the entire Culture series in sequence. I’m about to start now.

Which one is your favourite?

I think it’s The Player of Games, because that’s the first one I read. And also I’m very interested in games, in the rules of play.

It’s a wonderful book.

He deals centrally with this idea: when machines get better than humans at a lot of things, what do humans do? The player of games in The Player of Games plays games, literally. And he’s famed for his skill at playing games. In another of his book there’s a composer who has galactic fame. So why are people going to hear his symphonies if one of the sentient machines could write a quote-unquote better symphony? Again, I think it’s for the sense of connection, and this idea that the meaning of human experience is embodied in art made by humans.

Jaron Lanier says in You’re Not a Gadget that we don’t realise the extent to which we lower our standards so we can be amazed at what a machine can do. We go “oh, this is great”. But if we knew that it came from a human being, we’d say it’s ridiculous. I think that the human element, the human input is required at a certain level because otherwise it may be formally perfect but by itself it really doesn’t work for us.

Right. What an AI machine is doing is pattern recognition and reproduction of the pattern. Before Love and Longing in Bombay, I’d written a big novel [Red Earth and Pouring Rain], and I wanted to exercise different muscles. Love and Longing is a collection of short stories, in which every short story is in a genre that I love: ghost stories, story about social warfare among ladies, that kind of thing. But the fun is, you take this pattern and then you twist it. Sacred Games is a big riff on cop stories, on gangster stories, on spy sagas. How can you take the patterns of the venerable detective story and distort them? That distortion – that is human intervention. You can ask ChatGPT to write a detective story, but can you ask it to make a beautiful distortion, a subversion? That, I think, it probably cannot do.

I do translations to pay the bills, I translate legal and corporate documents and things like that. And a lot of the repetitive stuff goes away with machine translation, a lot of what you call grinding. But you always still need a human being to look it over because when the machine gets it wrong, it gets it really wrong.

Yeah. And in translation also, what is the taste of the work in the language you are translating into? You can’t just do a one to one translation, word to word.

I don’t think you get to use AI to take any hard work off from the novel you need to write now, though. Sorry about that.

 (Laughs) No, you’re right. I’ve been going on, on Twitter, about “can somebody write me a version of ChatGPT that fits on my laptop so I can ask it questions about the thousands of PDFs and .doc files and website clippings?” Someone will do that soon. I’ll pay for that immediately.

I recently sold my house, and I’m very much in the middle of all this. In the United States, they charge you a huge amount of capital gain tax when you sell anything. But if you’ve invested money in repairing stuff in the house, you can deduct that.


Yeah. We’ve lived in that house since 2007. And I can remember stuff that I paid money for, but of course I don’t remember the name of the contractor, and I want to find that invoice.

I can try to find the documents with dumb keyword searches – like “floor repair” as a search phrase. But then if that’s not a phrase in a document, I won’t find anything. If I could ask an AI, find me all the contractors I’ve hired from this time to this time, give me the details about my floor repair from 2006, and it gave me some sort of sensible answer, that would be very helpful. I think that would be a great boon to all of us. I know people are working on this, because it’s obvious, I think, that the small world applications to human lives would also be very significant.

Lawrence of Arabia, don’t think so, but perhaps finding your invoice, yes.

Right. But think about it. Maybe you’d want to make a small indie film, and you’d be able to do it. People are already working towards this and the results they are showing are very interesting. These are two-minute, three-minute long short clips, but you can see the potential there.

And then the other thing, the democratisation of programming that everybody’s been looking for since like, 1950. You can describe what an app should do and then the AI makes the app for you – which might be very useful, but perhaps or probably as buggy as hell. You wouldn’t know where the subterranean frailties were until it destroyed your life (laughs).

Yes. And I think for an obsessive like you that would be very triggering.

Right. But I’m thinking some version of Visual Basic, where you could ask the AI to help you put together smaller pieces of code that you could test thoroughly and then put together something that would be fairly reliable. I think that’s actually a big thing right now, low-code or no-code programming. It’s huge.

What was weird was that I started programming, using Visual Basic or similar tools, I could make very useful programs. I made a living doing that. I always thought that programming would get easier. But as the years went on, it’s gotten harder to make a simple program because making the front end, the UI, has become insanely hard.  There’s a really steep learning curve before you can make the front end of a website, for instance. And there are a million frameworks that people use, and every year the hot fashionable frameworks fall out of fashion, and you start all over again. That’s where the low-code and no-code apps come in. SquareSpace and Wix and similar apps allow ordinary users to build their own websites, and this niche is where they make money.

As a hacky programmer, I just want a swift and efficient way to put together a useful app. If AI can help me do this, blessed it be.

But of course no-code programming is sort of a misnomer because there is code. It just a user interface that means you don’t have to deal with it directly.

That’s what Visual Basic and other tools like Paradox and Clipper used to do. You drop some fields on a form and then you write code. You hook it all up together behind the UI and it works. I made lots of money doing that for clients.

I think that’s a very good thing. Programming became harder and harder. It’s so bizarre because I think it should have got easier and easier.

You’re optimistic about that then?

I haven’t had time recently, but there’s a couple of really great projects that were pre-AI, but now they’re starting to attach AI to their tools.

People should look at FlutterFlow. It’s done by a couple of young ex-Google engineers, and it’s amazing. You can use their tech to construct an app using drag-and-drop, and this app will work on iOS, on Android, on the desktop, everywhere. I’ve only played with it, I haven’t built anything serious with it yet, but it’s great. And then the other thing is, – and again, I’m getting really nerdy – is that building a program and then getting it into an app store, especially on the Apple site, is an insanely annoying process involving lots of your energy and time. We were starting to talk about putting Granthika on mobile platforms, and that would have cost us a lot of time and effort. FlutterFlow will do is help you along that process, which alone is worth paying for. If I were building something, I would pay for it just to help me get it into the store. Those kinds of things I think, are tremendously useful.

I agree, if AI does that right.

So beyond dragging and dropping fields and writing code, there’s also the problem of making your AI look good. So if I can tell the AI, okay, I put these six fields on there, give it this colour scheme and maintain it across all the screens, give all the fields a drop shadow, three pixels, and if it does it for me, that’s really worth something. And then I can tell the AI to change some aspect of this theme, and so on.

I do some UX design, and there’s lots of talk about that, about whether UX designs are going to get replaced. And people aren’t too worried because in UX you really need to look at how human beings behave. They are very excited though about its potential to streamline lots of things that are really, really boring and really hard.

The early reactions against even using word processors were pretty intense. People were very attached to the feel of pen on paper. Iris Murdoch said, “The word processor is… a glass square which separates one from one’s thoughts and gives them a premature air of completeness.” She also asked how one could possibly write with “a machine between you and the page?” People are always very attached to the sensory experience of what they’ve been used to. I still hear people saying that they like writing their first draft with pencil on paper, but they are very much in the minority. Everyone uses the convenience of computers.

So UX is crucial. How a tool feels in the hand is central to one’s use of it, to how useful the tool is. This is true of software as anything else, and that’s why UI is so important. Aesthetics matter, the feeling of a tool matters. I obsess about keyboards now as much as I used to obsess about fountain pens and nibs when I was a kid.

When you build software that is streamlined and beautiful as well as being functional, it eventually becomes invisible to you even as you enjoy using it. Many people, including me, believe that version 5.1 of WordPerfect for DOS was the best traditional word processor ever made. I’ll die on this hill. 5.1 was lean, and you could use it without taking your hands off your keyboard. Some of the keyboard shortcuts are still baked into my muscle memory—F6 for bold, F8 for underline, and F11 for reveal codes (which let you see exactly why your text looked like it did). I wrote most of my first novel using various versions of WordPerfect. The company made some mistakes, and Microsoft used its juggernaut strategies to crush them, so now we all use the bloated mess that is Word. But WordPerfect’s UX was beautiful, and that was why it was beloved. And, yes, get off my lawn.

I’m unable to write by hand. I need to do it on the computer because I edit as I go.

I can’t even read on paper anymore because I annotate a lot, especially when I’m doing research. With the convenience of being able to annotate on my tablet and have it show up on my laptop, it would be impossible for me to go back to paper.

Again, technology allows you to work at a higher level to solve your problems. Early on during the Pandemic, the Austrian Embassy in Washington, DC, and the consulate in San Francisco, put together a program connecting American artists and writers with artists in Austria and having them making pieces together. I collaborated with Adele Razkövi, an artist and animator, on a short three-minute film, Sequence of Tenses ( She works with real analogue material, with clay objects. She uses computers to animate the analogue materials that she moves and shapes by hand. And she said the same thing, that if she had to worry about photographing each frame by hand, if she had to animate all this material manually, it would drive her nuts.

What tech allows me to do is to delegate tasks to the machine and then work at a higher level, work at tasks that actually address my creative process and make it easier. With Granthika, we tried to address the problem of information retrieval and management. And now the question is, what else can the machine do that will free me up? I want it to save me from having to do these word searches which are frustrating because they often don’t work, especially if I’m looking for a common phrase. If you look up “Buddha’s view on death” on the web, you’ll find thousands of results. It even takes a lot of effort for me to find things in my own saved research materials. I want to ask the AI assistant questions about my research materials. A more semantic search, I think, can do that. And that’s something I’m very excited about.

Leave a Reply

Your email address will not be published. Required fields are marked *