I did not imagine the problems I was having was due to a lack of temples and worshipping the right gods. Being from India, this should have been obvious. I had figured out long ago that a steady supply of beer, dedication to craft, good means, and romance were the were critical to happiness. Spirituality had not been considered. Hewing a temple out of granite improved focus.
These days Dwarf Fortress gets lumped into the colony management category of games. It is a pioneer in the genre but it is also so much more. It is for good reason that it is one of the few games thought worthy of collecting by the Museum of Modern Art (MoMA). Even there DF changed the way MoMA preserves art. DF one the most complex game ever created, starting from simple experiments, coded by a single person over 20 years, available for free, through all the normal human hardships. To play Dwarf Fortress is to experience audacity.
DF is a complete simulation. From the growth rate of trees and grass, to simulating individual body parts of creature that allows cats to get drunk. The point of Dwarf Fortress is not to win. There is no way to win. As they say, losing is FUN! The gams starts normally enough. You set jobs for seven dwarfs to that helps them create a home in an unkind wilderness. Sometimes unexpected things happen like a giant farting bird attacks or the elves are cross with you because you used wood to make beds, which is fine in a new game. Eventually though something happens that tells you that there are more layers to this. Like my spirituality problems.
There is a lot of well known lore surrounding DF. From the famous story of Boatmurdered, Oilfurnace, Webcomics, to the drunk cats bug. DF in its small ways also reminds us of life’s important truths like, cats adopt the person not the other way around.
Tantrum spirals, goblin sieges, chairs of different qualities and the happiness they impart, dwarf fortress is deep. Like any effort by a single person, it started simple. Zach and Tarn Adams are brothers who created many many games as kids. DF was not even created for any kind of commercial aim. They simply wanted to simulate as many things as they could so that the game had the ability to tell great stories. Bit by bit, Tarn Adams coded DF without any external help, while finishing his PhD, and eventually getting enough in donations that he could dedicate his time to just building the game.
Recently the brothers worked with a publisher to bring their game to Steam. It made them “overnight” millionaires. That night was 20 years long. Along the way they build up a dedicated fan following, some contributed art, and music to the game, others hacked into the software to provide utilities to improve quality of life. Many of them are now part of the team working on DF full time. A few years ago Tarn estimated that the game was about 44% complete. I have a suspicion that the number hasn’t changed much because despite the regular updates, the brothers keep adding new ideas to build on.
You are unlikely to ever play Dwarf Fortress, but that doesn’t mean it’s not worth knowing about this bittersweet human story. No Clip has made a four part documentary, you should watch it.
In a time when games were simple, computing power limited, with no funding, and life’s challenges, DF was created. To play Dwarf Fortress is to experience audacity.
I was all of 15 when defenestration was forever implanted in my mind. It means to throw someone out the window. It happened in Prague, 1618. Some important people were defenestrated, fell 70 feet, landed in dung. This led to the thirty years war and the coining of the word ‘defenestration’. Defenestrating happened to important, visible, people held responsible for mismanagement leading to widespread discontent. While the defenestrated may represent the idea, surely we can’t imagine that it was that specific person who was going around causing the suffering. No, they had minions. Here we explore a bit of their story.
Horned owl (Hoornuil) (1915) print in high resolution by Samuel Jessurun de Mesquita. Original from The Rijksmuseum. Digitally enhanced by rawpixel.
Royalty is meant to be seen. They were either chosen by or were the local gods to lead the people. They were the head of everything and if something were to go wrong it was their responsibility. Royalty also means creating good memes. Whether the Alhambra, Taj Mahal, or Beijing projecting power through architectural memes was the standard.
Administration and bureaucratic structures is the silent clockwork that powers the projection. These guys, are antimemetic. The antimeme is a recent invention and denotes ideas that have high impact but are hard to spread. This is important because when the tax burden gets too high you want the peasants to go for the king not the local tax collector.
The Mughal emperors were the head of the administrative machinery with final say over all important matters. The administration itself was antimemetic in nature. The provincial officials such as the bakhshi, sadr as-sudr, and finance minister reported directly to the central government rather than the subahdar (provincial governor). Matrix organization, I hear you thinking. This complex, multi-layered reporting structure, while designed for central control, also diffused responsibility and made the precise locus of decision-making less transparent to external observers and even to other officials.
In the Ming dynasty, the Hongwu Emperor abolished the Central Secretariat to assume personal control. However, the volume of letters got so high that he soon appointed a few grand secretaries. They never held a high rank and always merely “recorded imperial decisions”. If merely were a boxer he would be a heavyweight. Can’t blame that guy with the pen if he’s just doing what the king asks him to.
From the al-Andalus through the Ottomans, Safavids, and Mughals the the ulama shaped legal systems and molded public morality. Of course the monarchs decrees but the ulama interpreted them and applied them as law into daily life. This interpretive authority, operating subtly within the legal and religious bureaucracy, allowed for continuous adaptation and influence without the visible, attributable acts of formal legislation, making it profoundly antimemetic.
Let me end with the quote from the wonderful, and joyfully mimetic, Yes, Minister:
“Hacker: Humphrey, did you know that 20% of all honours go to civil servants?
Sir Humphrey: A fitting tribute to their devotion to duty, Minister.
Hacker: No, their duty is what they get paid for. The rest of the population has to do something extra to get an honour. Something special. They work for 27 years with mentally handicapped children six nights a week to get an MBE. Your knighthoods simply come up with the rations.
Sir Humphrey: Minister, her Majesty’s civil servants spend their lives working for a modest wage and at the end, they retire into obscurity. Honours are a small reward for a lifetime of loyal, self-effacing discretion and devoted service to Her Majesty, and to the nation.
Hacker: “A modest wage”, did you say?
Sir Humphrey: Alas, yes.
Hacker: Humphrey, you get over £30,000 a year! That’s £7,000 more than I get.
Sir Humphrey: Yes, but still relatively the modest wage.
Hacker: Relative to whom?
Sir Humphrey: Well, Elizabeth Taylor, for example.
Hacker: Humphrey, you are not relative to Elizabeth Taylor. There are important differences.
Sir Humphrey: Indeed, yes. She didn’t get a first at Oxford.
Hacker: And you do not retire into obscurity?! You take a massive index-linked pension and go off to become directors of oil companies and banks.
Sir Humphrey: Oh, yes, but very obscure directors, Minister.
Hacker: You’re in no danger of the sack. In industry if you screw things up, you get the boot. In the civil service, if you screw things up, I get the boot.
Sir Humphrey: Very droll, Minister, now if you’ve approved the list…”
[Series Two (1981) Episode Two: Doing the Honours]
Sources
Much of the reading and sourcing of material for this was done across books from the Contraptions Book Club and some deep research help.
On the recent season of the show Clarkson’s farm, J.C. goes through great lengths to buy the right pub. As with any sensible buyer, the team does a thorough tear down followed by a big build up before the place is open for business. They survey how the place is built, located, and accessed. In their refresh they ensure that each part of the pub is built with purpose. Even the tractor on the ceiling. The art is in answering the question: How was this place put together?
A data-scientist should be equally fussy. Until we trace how every number was collected, corrected and cleaned, —who measured it, what tool warped it, what assumptions skewed it—we can’t trust the next step in our business to flourish.
Old sound (1925) painting in high resolution by Paul Klee. Original from the Kunstmuseum Basel Museum. Digitally enhanced by rawpixel.
Two load-bearing pillars
While there are many flavors of data science I’m concerned about the analysis that is done in scientific spheres and startups. In this world, the structure held up by two pillars:
How we measure — the trip from reality to raw numbers. Feature extraction.
How we compare — the rules that let those numbers answer a question. Statistics and causality.
Both of these related to having a deep understanding of the data generation process. Each from a different angle. A crack in either pillar and whatever sits on top crumbles. Plots, significance, AI predictions, mean nothing.
How we measure
A misaligned microscope is the digital equivalent of crooked lumber. No amount of massage can birth a photon that never hit the sensor. In fluorescence imaging, the point-spread function tells you how a pin-point of light smears across neighboring pixels; noise reminds you that light itself arrives from and is recorded by at least some randomness. Misjudge either and the cell you call “twice as bright” may be a mirage.
In this data generation process the instrument nuances control what you see. Understanding this enables us to make judgements about what kind of post processing is right and which one may destroy or invent data. For simpler analysis the post processing can stop at cleaner raw data. For developing AI models, this process extends to labeling and analyzing data distributions. Andrew Ng’s approach, in data-centric AI, insists that tightening labels, fixing sensor drift, and writing clear provenance notes often beat fancier models.
How we compare
Now suppose Clarkson were to test a new fertilizer, fresh goat pellets, only on sunny plots. Any bumper harvest that follows says more about sunshine than about the pellets. Sound comparisons begin long before data arrive. A deep understanding of the science behind the experiment is critical before conducting any statistics. The wrong randomization, controls, and lurking confounder eat away at the foundation of statistics.
This information is not in the data. Only understanding how the experiment was designed and which events preclude others enable us to build a model of the world of the experiment. Taking this lightly has large risks for startups with limited budgets and smaller experiments. A false positive result leads to wasted resources while a false negative presents opportunity costs.
The stakes climb quickly. Early in the COVID-19 pandemic, some regions bragged of lower death rates. Age, testing access, and hospital load varied wildly, yet headlines crowned local policies as miracle cures. When later studies re-leveled the footing, the miracles vanished.
Why the pillars get skipped
Speed, habit, and misplaced trust. Leo Breiman warned in 2001 that many analysts chase algorithmic accuracy and skip the question of how the data were generated. What he called the “two cultures.” Today’s tooling tempts us even more: auto-charts, one-click models, pretrained everything. They save time—until they cost us the answer.
The other issue is lack of a culture that communicates and shares a common language. Only in academic training is it possible to train a single person to understand the science, the instrumentation, and the statistics sufficiently that their research may be taken seriously. Even then we prefer peer review. There is no such scope in startups. Tasks and expertise must be split. It falls to the data scientist to ensure clarity and collecting information horizontally. It is the job of the leadership to enable this or accept dumb risks.
Opening day
Clarkson’s pub opening was a monumental task with a thousand details tracked and tackled by an army of experts. Follow the journey from phenomenon to file, guard the twin pillars of measure and compare, and reinforce them up with careful curation and open culture. Do that, and your analysis leaves room for the most important thing: inquiry.
The wave towered over me. Then the sound filled my ears. Not the calm breath of the waves; but it was surf music. I was maybe 3. Song names and artist names were beyond me. There was only the blue-green wave and the twang of the guitar.
I have chased music all my life. Just had to figure out the tools. The record player and the giant speakers taught my first lesson: pressing buttons was joy. In my pursuit I learned in about records, tapes, CDs, mp3, flac, streaming, Napster, torrents, Winamp, VLC, blanks, CD-R/RWs, compression, bit rates, conversion, transfer, backups, VPN, networking, impedance matching, DACs, amplifiers, calibration, ARC, fibre, buying, licensing, and streaming in approximate order.
I discovered that they were called The Ventures by accident. Late in the college years I watched Pulp Fiction and wanted all the music. This one wasn’t quite home but it was the right street. It was surf music.
The hunt was on. Only a notion of the song and the confidence that I would know it when I heard it. I didn’t know the name of the album only that it had a big wave on the cover. It took me the better part of 6 months, on slow DSL, trawling all the sources I knew. Listening for that drum fade-in. Then one day I found it.
It’s been decades since the record player stopped spinning. I’ve moved a dozen times, the records were lost. I am the default A/V guy and love the role. Now I live in one of the surfiest places on the planet, the current still pulls but I walk, don’t run.
I got told off by The Paris Review today. Maybe it wasn’t necessarily directed at me, but as they say in the, now old, new lingo, I felt attacked. You see, recently, drawing on the well of inspiration that is history I succeeded in writing a poem, but not just any poem. I wrote a ghazal.
Those who know me for any amount of time are madeawareofmytasteforwritingpoetry. It’s usually pretty bad but I persist, cause why not. The OG is long gone anyway. The ghazal is an especially ambitious type of poetry to be taken up my someone with my modest talents. To make matters worse, as I learned today, the ghazal is really well suited for the Urdu. For all practical matters, I know only English.
By me! San Diego Botanic Garden, California Poppy (I think).
For anyone with any little interest in love and romance, being born in South Asia is a special kind of blessing. We are lucky to have had Urdu poetry reach its peak here. Urdu is perhaps the perfect medium to transmit mischief, passion, pain, longing, and the myriad other emotions which are handmaidens to big Love. Not any kind of expert, but all my life I’ve consumed shayari, sher, ghazals, whether in mainstream Bollywood or in sparkling corners of the internet.
Armed with the internet, full of inspiration, my trusty editor, Mir ChatGPT, in the other tab. I decided it was time to go all in. The Ghazal was to be written. It was, it follows all the rules, I even make a self reference in the last couplet as is the tradition, but it lacks oomph. A good sher, a good ghazal, should pierce you and make you blush for it’s andaaz, mischief and audacity.
Mine… well, you can read it here yourself, don’t forget to play the tiny desk concert, it is lovely.
Definitely read The Paris Review article for it’s a great take of view from a writer who transfers the styles of poetry in one language to another.
When printing was invented, Europe suddenly had access to all the books that had existed until that point in history. This included everything from mystical texts to astronomical observations. Having no guides to judge quality, some people went off on the deep end. Giordano Bruno is sometimes referred to as the forefather of modern cosmology. He was not. An extreme case, he took mystical click-bait, mixed it with the then-contemporary Copernican theories, and, without any data, invented the infinite universe. Eventually, culture adapted and people started to compare and organize all the data. This act of orienting and place-making led to the scientific revolution.Printing created too much information and we had to learn how to handle it. Today we are in a similar position.
Berry Pickers (1873) by Winslow Homer. Original from The National Gallery of Art. Digitally enhanced by rawpixel.
Still in the early days of the internet we sometimes lost the ability to tell signal from noise. Recently Hank Green posted this video where he makes his thesis that we aren’t addicted to content, but are instead starving for information. This strikes me as true.
The companies behind the social internet drown us in noise with just enough signal to keep you coming back. That signal, that hit, is a hint at information that provides orientation. Opportunities for conversation and belief challenging interactions are difficult to experience. As explored in a previous post, as humans are geographical creatures. Phones and the internet are a real part of our environment. Without sufficient places for orientation, we are left glassy eyed, lost. To see why that ‘information hunger’ feels so visceral, consider the simple ladder that links raw signal to a basic survival drive:
Signal → Information → Orientation → Biology
Signal is any pattern in the environment—visual, auditory, textual—that stands out from background noise. On social platforms this might be a headline, a notification badge, or an unexpected data point.
Information is signal that has been parsed and interpreted. Your brain (or a community) attaches meaning and relevance: “This headline matters to my work,” or “That data point contradicts my belief.”
Orientation is what information enables: a clearer, updated internal map of “where I stand and what to do next.” It answers “How does this fit with what I already know?” and “Which way should I move—intellectually, emotionally, physically?”
Biological need is the evolutionary pressure behind all of this: organisms that build accurate mental maps survive. Humans feel discomfort when our maps are fuzzy (disorientation) and relief or pleasure when new information sharpens them.
A few years ago, my corner of the internet got into waldenponding and promptly logged off. Just kidding. The failure of modern waldenponding makes it clear that this move of turning away from the social internet is not the answer. That would be like giving up on books because there were too many of them. The internet and the social internet in general do provide opportunities Instead, engaging with curiosity allows us to orient ourselves. Having an information shaped content diet opens up a path to a healthier mind. While society learns to put on the right kind of controls as we have on sugar and tobacco, how can we learn to have fun on the internet?
The hunt for knowledge and discovery, even of trivia is immensely enjoyable. Socratic problem solving is a team sport. Everyone has narrow views of the world and our thinking may be based on shaky knowledge. Social internet has so far made our eagerness to win the top emotion in online discourse, Socratic inquiry can transform that into collaborative inquiry. To arrive at better knowledge we must be willing to talk, listen, challenge, and accept. It is only by comparing notes that we open up a topic, a space, for exploration. Each of us and our thoughts are a place in the world. Places create orientation and orientation has the potential to create progress. While progress may not be guaranteed, not engaging in inquiry guarantees disorientation and formlessness.
While printing turned information into data, the social internet has turned information into noise. Social internet companies have tuned our culture to produce low signal-to-noise “content”. As Hank Green put it, we do hunger for information. We hunger because information is orientation. Orientation is a primal biological need to help us navigate our physics-virtual environment. The internet is a place where people share freely and welcome warm interactions. To turn away from the internet because of the culture tuning is the wrong move. The internet has too much to give, engaging from a posture of inquiry is the way. Inquiry satisfies that inner need for place creation and orientation.
This post continuesthe series on Socratic Thinking, turning the space-and-place lens inward to examine the mind itself. Human minds can be thought of as an imperfect place with the ability to create their own insta-places to navigate ambiguity.
On the Trail (1889) by Winslow Homer. Original from The National Gallery of Art. Digitally enhanced by rawpixel.
Exploration in any real or conceptual space needs navigational markers with sufficient meaning. Humans are biologically predisposed to seek out and use navigational markers. This tendency is rooted in our neural architecture, emerges early in life, and is shared with other animals, reflecting its deep evolutionary origins 1,2 . Even the simplest of life performing chemotaxis uses the signal-field of food to navigate.
When you’re microscopic, the territory is the map; at human scale, we externalise those cues as landmarks—then mirror the process inside our heads. Just as cells follow chemical gradients, our thoughts follow self-made landmarks, yet these landmarks are vaporous.
From the outside our mind is a single place, it is our identity. Probe closer and our identity is nebulous and dissolves the way a city dissolves into smaller and smaller places the closer you look. We use our identity to create the first stable place in the world and then use other places to navigate life. However, these places come from unreliable sources, our internal and external environments. How do we know the places are even real, and do we have the knowledge to trust their reality? Well, we don’t. We can’t judge our mental landmarks false. Callard calls this normative self-blindness: the built-in refusal to saw off the branch we stand on.
Normative self-blindness is a trick to gloss over details and keep moving. Insta-places are conjured from our experience and are treated as solid no matter how poorly they are tied down by actual knowledge. We can accept that a place was loosely formed in the past, an error, or is not yet well defined in the future, is unknown. However, in the moment, the places exist and we use them to see.
Understanding and accepting that our minds work this way is a key tenet of Socratic Thinking. It makes adopting the posture of inquiry much easier. Socratic inquiry begins by admitting that everyone’s guiding landmarks may be made of semi-solid smoke.
1Chan, Edgar, Oliver Baumann, Mark A. Bellgrove, and Jason B. Mattingley. “From Objects to Landmarks: The Function of Visual Location Information in Spatial Navigation.” Frontiers in Psychology 3 (2012). https://doi.org/10.3389/fpsyg.2012.00304
“A farmer has to cut down trees to create space for his farmstead and fields. Yet once the farm is established it becomes an ordered world of meaning—a place—and beyond it is the forest and space.” — Yi-Fu Tuan
Thinking itself is place-making: the act of converting undifferentiated possibility into navigable meaning.
A place comes into being the moment we interrupt undifferentiated space. Place-making is fundamentally an act of interruption. Space is thought of as possibility but is unavailable without the signposts of place. When a place is created we impose a way of looking, being, and acting on the space of choice. The place you pick to navigate your space defines the identity you will inhabit during your quest. Every tool is a micro-place: it frames what can be thought and forecloses alternative moves. They enforce the kind of thoughts that can be had, the type of exploration that can be done, and configures space in an opinionated way.
Two-masted Schooner with Dory (1894) by Winslow Homer. Original from The Smithsonian. Digitally enhanced by rawpixel.
Picking a tool commits us to a world view. Consider the space of ‘good TV shows’. Family, friends and culture have made the choice of what good means. When Netflix suggests shows it uses your watching history as a probe to create place so that every individual is always watching ‘good’ shows. The pure possibility space of the search bar is disrupted by the suggestions provided.
Like algorithmic curation, Socratic dialogue also interrupts space, it is interrogation as cartography. Socratic thinking is also an act of interruption and making concrete what was nebulous. It’s asking us to specify which show, if we claim to love TV. Socratic thinking (henceforth referred to as just thinking) starts by probing that which does not need questioning, the answers that are obvious the ones that everyone knows. This may seem foreign at first glance but we do this all the time, say we make a list of our favorite TV shows, someone always says you are missing this or that show and that this list is completely wrong. This kind of disagreement leads to the shared quest of answering the question, ‘What is it to be entertained?’.
Thinking pursues knowledge through the act of stabilizing answers to such questions by creating places in those unexamined areas. Discussion allows us to map. There is usually no well defined answer for such questions, if there were, they would simply be problems that we could solve with a google search. The quest stops when the parties involved are satisfied that they have arrived at an answer. Thinking is the act of place-making by taking something that was ungraspable and tying it down with knowledge. Place is, after all, an “ordered world of meaning” and we can use these places to create home bases from which to explore.
Even without other people simply engaging with the reality of the universe is sufficient for thought. Places are stable systems which provide a surface on which your thoughts and hypothesis can be tested. Even if there is no other person around and you’re simply engaged with looking at the world can uncover a new truth tied down by knowledge.
Thinking is the process of updating beliefs based on the mini places that make up the space that you’re interrogating. Each place is a noisy pointer to the underlying truth, and each updating of belief allows you to get closer to the knowledge you seek.
I had the strangest conversation with my son today. There used to be a time when computers never made a mistake. It was always the user that was in error. The computer did exactly what you asked it to do. If something went wrong it was you, the user, that didn’t know what you wanted. After decades of that being etched in today I found myself telling him that computers make mistakes, you have to check if the computer has done the right thing and that is actually ok. A computer that hallucinates also provides a surface for exploration and seeking answers to questions.
Boys Wading (1873) by Winslow Homer. Original public domain image from National Gallery of Art
In her book, Open Socrates, Agnes Callard draws our attention to the differences between problems and questions. I’ll get to those in a bit, but the fundamental realization I had was that until recently all we could use computers (CPUs, spreadsheets, internet) for was solving problems. This started all the way back with Alan Turing when he designed the Turing test. He turned the question of what is it to think into the problem of how do you detect thought. As Callard mentions, LLMs smash the Turing test but we still can’t quite accept the result as proof of thinking. What is thinking then? What are problems? What are questions? How do we answer questions?
Problems are barriers that stand in your way when you are trying to do something. You want to train a deep learning algorithm to write poetry, how to get training data is a problem. You want something soothing for lunch, getting the recipe for congee is the problem. The critical point here is that as soon as you have the solution, the data, the recipe, the problem disappears. This is the role of technology.
When we work with computers to solve problems we are essentially handing off the task to the computer without caring that the computer wants to or even can want to write poetry or have a nice lunch. So we ask the LLM to write code, we command google to give us a congee recipe. Problems don’t need a shared purpose, only methods to solve them to our satisfaction. Being perpetually dissatisfied with existing answers is the stance of science.
Science and technology are thus tools to move towards dealing with questions. Unlike problems which dissolve when you solve them, questions give you a new understanding of the world. The thing with asking questions is that there is no established way, at least in your current state, to solve them. Thus asking a question is the first step of starting a quest. In terms of science the quest is better understanding of something and you use technology along the way to dissolve problems that stand in your way.
AI lets us explore questions with, rather than merely through, computers. Granted that most common use of AI is still to solve problems, LLMs and their ability to do back and forth chat in natural language does provide the affordance to ask questions. Especially, the kind that seem to come pre-answered because we are operating from a posture where not having an answer would dissolve the posture altogether.
The Socratic Co-pilot
As a scientist, the question “what is it to be a good scientist?” comes pre answered for me. Until I am asked this question I have not really thought about it but rush to provide answers. Scientists conduct experiments carefully, they know how to do use statistics, they publish papers and so on. However, this still does not answer what it is to be a good scientist. Playing this out with an AI, I assert “rigorous statistics,” the AI counters with an anecdote on John Snow’s cholera map and I’m forced to pivot. None of these by themselves answers the root question, but it allows generation of some problems which can be answered or agreed on. This is knowledge.
Knowledge draws boundaries, or as I have explored earlier, creates places around the space that you wish to explore. In the space of “being a good scientist”, we can agree that the use the scientific method is an important factor. Depending on who you are, this could be the end of quest.
Even if no methodology exists for a given problem, simply approaching any problem with an inquisitive posture creates a method, however crude. In his book What Is It Like to Be a Bat? Thomas Nagel tackles an impossible to solve problem but a great question, through the process of a thought experiment. If I were to undertake this, I may try to click in a dark room, hang upside down. Okay, maybe not the last bit, but only maybe. Even this crude approach has now put me in the zone to answer the problem. Importantly my flapping about has created surface area where others can criticize, as Nagel was. Perhaps future brain-computer-interface chips will actually enable us to be a bat. However, lacking such technology, this is better than nothing as long as you are interested in inquiring about the bat-ness.
This kind of inquiry, this pursuit of answering questions is thinking. Specifically, as Callard puts it, thinking is “a social quest for better answers to the sorts of questions that show up for us already answered”. Breaking that down further it’s social because it’s done with a partner who disagrees with you because they have their own views about the question. It’s a quest because the both parties are seeking knowledge. The last bit about questions being already answered is worth exploring a bit.
Why bother answering questions you already have answers to? This is trivial to refute when you know nothing about a subject. For example let’s say you knew nothing about gravity and your answer to why you are stuck to the earth cause we are beings of the soil and to the soil we must go, the soil always calls us. If that is the worldview then you already have the answer. The only way to arrive at a better answer, gravity, is to have someone question you on the matter. Refuting specific points based on their own points of view. This may come in the form of a conversation, a textbook, a speech etc. I suspect this social role may soon be played by AI.
Obviously hallucinations themselves aren’t great but the ability to hallucinate is. In the coming years I expect AI will gain significant amounts of knowledge access not just in the form of training but in the form of reference databases containing data broadly accepted as knowledge. In the process we will probably have to undergo significant social pains to agree on what Established Knowledge constitutes. Such a system will enable LLMs to play the role of Socrates and help the user avoid falsehoods by questioning the beliefs held by the user.
Until now computers couldn’t play this role because there wasn’t enough “humanness” involved. In the bat example, a bat cannot serve as Socrates or as the interlocutor to a human partner because there isn’t a shared world view. LLMs, trained on human generated knowledge would have enough in common to provide a normative mirror. The AI comes with the added benefit of having both infinite patience and no internal urge to be right. This would allow the quest to provide an answer that is satisfactory to the user searching at every level of understanding. LLMs can be useful even before they gain the ability to access established knowledge. Simply by providing a surface on which to hang questions the user can become adept at the art of inquiry.
So the next time you have a chat with your pet AI understand that it starts as a session of pure space. Each word we put in ties down the AI to specific vantage points to help us explore. Go ahead—pick a question you think you’ve already answered and let the machine argue with you.
Aneesh Sathe 