Notebooks And The Extended Mind

Philosophy is the new jam; NotebookLM gets a new name.

A few weeks ago, the Sunday NY Times business section published a long reported piece called “Revenge Of The Philosophy Majors” written by my old friend Ben Wallace, documenting the unlikely recent trend of AI companies adding academic philosophers to their payrolls:

“So, I think I’m going to major in philosophy” is the kind of undergraduate statement that for decades has terrorized tuition-burdened parents, inspiring dark visions of basement-dwelling offspring who fail to launch. Diogenes the Cynic lived in a clay jar. Baruch Spinoza ground lenses to pay the bills. Friedrich Nietzsche survived on the kindness of family and friends. The idea that a philosophy degree is a ticket to a lifetime of underemployment persists… But [recently] A.I. labs, and the related nonprofits around them, have been recruiting workers as versed in Consequentialism and John Stuart Mill as in neural networks and reinforcement learning.

While my background is technically more on the side of history and media theory, I suppose it’s fair to say that my involvement with Google is adjacent to the trend Ben describes in his (wonderful) piece. A few years ago, I described the trend using a slightly wider framing: as the revenge of the humanities, not just the philosophy majors. The specific peg back then was the surprising discovery that in the age of prompt engineering the writers had a superpower that many tech engineers lacked: the ability to communicate complex ideas in clear, persuasive prose. But even then it was clear that a broader set of skills had a newfound value on the market.

The core skills are not just about straight prompt engineering; they’re not just about figuring out the most efficient wording to get the model to do what you want. They also draw on deeper, more nuanced questions. What is the most responsible behavior to cultivate in the model, and how do we best deploy this technology in the real world to maximize its positive impact? What new forms of intelligence or creativity can we detect in these strange entities? How do we endow them with a moral compass, or steer them away from bias and inaccurate stereotypes? Can language alone generate a robust theory of how the world works, or do you need more explicit rules or additional sensory information?

All of those questions have been absolutely central to the discussion of AI for the past two years, but if you think about it, they were all questions that belonged to the humanities until the language models came along: ethics, philosophy of language, political theory, history of innovation, and so on.

One of the heroes of Ben’s piece is NYU philosopher David Chalmers, whose writing and teaching has had a defining influence over the nascent field of AI philosophy. But in a way, I think Ben undersells Chalmers’ impact. The piece focuses more on Chalmers’ famous provocation about the “hard problem of consciousness” and the debate over whether self-awareness is reproducible outside of biological systems. Those are both vital lines of inquiry, and beyond that just fun to think about, but they both involve speculation about developments that (almost certainly) have not happened yet. But the other big idea that Chalmers helped bring into the world—the extended mind theory—actually has enormous relevance to the here and now of AI circa 2026.

The core principle of the extended mind theory is one that will be familiar to Adjacent Possible readers: the external tools that we use to help us reason, remember, associate, calculate—from counting on our fingers to chatting with an advanced LLM—should be properly understood as extensions of our own cognitive process. “Thinking” is something that happens inside and outside our brains. In their original 1998 paper, Chalmers and co-author Andrew Clark argued for what they called an active externalism, based on the “active role of the environment in driving cognitive processes.” The paper features an oft-cited thought experiment involving an Alzheimer’s patient named Otto:

Like many Alzheimer’s patients, he relies on information in the environment to help structure his life. Otto carries a notebook around with him everywhere he goes. When he learns new information, he writes it down. When he needs some old information, he looks it up. For Otto, his notebook plays the role usually played by a biological memory. Today, Otto hears about the exhibition at the Museum of Modern Art, and decides to go see it. He consults the notebook, which says that the museum is on 53rd Street, so he walks to 53rd Street and goes into the museum.

Clearly, Otto walked to 53rd Street because he wanted to go to the museum and he believed the museum was on 53rd Street. The information in the notebook functions just like the information constituting an ordinary non-occurrent belief; it just happens that this information lies beyond the skin.

The alternative is to say that Otto has no belief about the matter until he consults his notebook; at best, he believes that the museum is located at the address in the notebook. But if we follow Otto around for a while, we will see how unnatural this way of speaking is. Otto is constantly using his notebook as a matter of course. It is central to his actions in all sorts of contexts, in the way that an ordinary memory is central in an ordinary life. The same information might come up again and again, perhaps being slightly modified on occasion, before retreating into the recesses of his artificial memory. To say that the beliefs disappear when the notebook is filed away seems to miss the big picture… the information is reliably there when needed, available to consciousness and available to guide action, in just the way that we expect a belief to be.

One of the most provocative arguments in that original paper is that the human brain itself may have been specially adapted to work with these external “tools for thought”—starting with language itself. “Think of a group of people brainstorming around a table, or a philosopher who thinks best by writing, developing her ideas as she goes,” Chalmers and Clark wrote. “It may be that language evolved, in part, to enable such extensions of our cognitive resources within actively coupled systems.” In other words, embracing extended cognition is a “core cognitive process” for humans, not just a modern invention.

I mention this whole chapter in Chalmers’ career because it seems much more pertinent to the present day realities of AI. To me, the most pressing question is not whether the machines are on the verge of sentience, since the likelihood of that seems low. The more pressing question is how you build an extended mind architecture that deepens your understanding and expands your creativity, given that there are promising, if slightly chaotic, tools for those tasks on the market right now, at the fingertips of billions of human beings.

I’m not sure if this was by accident or design, but it turns out we don’t have a lot of capital-P Philosophers on the team at Google Labs. (Armchair we have by the dozens.) In my little corner of Labs, I almost never hear the consciousness question discussed, even over a casual work lunch. But we do spend an enormous amount of our time exploring the possibility space around Chalmers and Clark’s extended mind concept. It’s not an accident we called our first consumer-facing product NotebookLM: as in the Otto thought experiment, the long history of notebooks in all their manifestations—commonplace books, journals, diaries—has always supplied the most powerful examples of the extended mind at work.

Which brings me to some other news.

***

Almost exactly four years ago, I drove down to Mountain View for my first day as a Google employee. I was fifty-four years old, and somehow had managed to avoid ever having a proper job over that entire period, other than a few startups that I had founded in my twenties and thirties. At that point Google Labs, which had just been formed a few months earlier, was headquartered in a fairly anonymous looking suburban office park off 101, closer to Office Space than Silicon Valley on the Mike Judge spectrum. As it happened, my first day at Google was a Monday in the summer of 2022, which meant that COVID-era work from home policies were still the default, and the office was absolutely deserted. But I’d been set up with a badge and a computer and I had to schedule a phone call with tech support to get my account in order, and so I grabbed one of the empty cubicles and sat down to work through the problems I was having getting into the system. I remember feeling a definite sense of dread in that moment, sitting there in the silence, thinking: have I made a terrible mistake?

And then—and I swear to god this actually happened—the silence was interrupted by a lonely robot rolling its way across the carpet past my borrowed desk.

Before the end of the summer, though, an inspiring cast of characters began to populate the office, and by October a small group of us had built a prototype of an AI-first tool for thought that my colleague Adam codenamed Tailwind. About six months later, after Josh Woodward previewed Tailwind at I/O for the first time, we set about to give our little experiment a proper name. We had some branding consultants spend an afternoon with the team; we ideated and free associated and placed our post-it notes on the whiteboard, but all the names that they came back with were too impressionistic, too evocative. They didn’t sound like iconic Google names: Maps, Docs, Meet. Because of my long obsession with tools for thought and the extended mind, I had always been partial to simply calling it Notebook. There was something about that framing that just made sense: a notebook is the place where we capture information that we want to remember, where we gather our thoughts. It gives us a dedicated space for thinking, where new connections happen, where we can trace the evolution of our own ideas. And it had a simplicity that fit the Google house style.

But we were far too experimental and unproven to take on a name like Google Notebook—remember we had exactly zero actual users at this point—and so we found ourselves a bit stuck: too early-stage for a proper grownup name, but ill-inclined to adopt a name that didn’t sound like the classic Google brands. But at the time there happened to be another Labs experiment that had just launched called MusicLM, and so at some point in late May of 2023 it occurred to me that we could borrow that convention for our embryonic product, and call it NotebookLM. It wasn’t the most elegant name you could imagine, but it had a certain logic to it: we were building a notebook reinvented for the age of language models, after all. I wrote an email to a few folks proposing the idea and included the caveat that if the product ever took off, and Google wanted to elevate us into the canon, we could drop the LM and just be Notebook.

And so that’s the news. Three years and somewhere near a billion notebooks later, we are finally ditching the LM.