Posts Tagged ‘humans’
“Don’t eat your seed corn”*…
AI doesn’t really “think.” Rather, it remembers how we thought together. Are we’re about to stop giving it anything worth remembering? Bright Simons with a provocative analysis…
We are on the verge of the age of human redundancy. In 2023, IBM’s chief executive told Bloomberg that soon some 7,800 roles might be replaced by AI. The following year, Duolingo cut a tenth of its contractor workforce; it needed to free up desks for AI. Atlassian followed. Klarna announced that its AI assistant was performing work equivalent to 700 customer-service employees and that reducing the size of its workforce to under 2000 is now its North Star. And Jack Dorsey has been forthright about wanting to hold Block’s headcount flat while AI shoulders the growth.
The trajectory has a compelling internal logic. Routine cognitive work gets automated; junior roles thin out; productivity gains compound year on year. For boards reviewing cost structures, it is the cleanest investment proposition since the internal combustion engine retired the horse, topped up with a kind of moral momentum. Hesitate, the thinking goes, and fall behind.
But the research results of a team in the UK should give us pause. In the spring of 2024, they asked around 300 writers to produce short fiction. Some were aided by GPT-4 and others worked alone. Which stories, the researchers wanted to know, would be more creative? On average, the writers with AI help produced stories that independent judges rated as more creative than those written without it.
So far, so on message: a familiar story about the inevitable takeover by intelligent machines. But when the researchers examined the full body of stories rather than individual ones, the picture became murky. The AI-assisted stories were more similar to each other. Each writer had been individually elevated; collectively, they had converged. Anil R Doshi and Oliver Hauser, who published the study in Science Advances, reached for a phrase from ecology to explain this: a tragedy of the commons.
Hold that result in mind: individual gain, collective loss. It describes something far more consequential than a writing experiment—it describes the hidden logic of our entire relationship with artificial intelligence. And it suggests that the most successful organizations of the coming decade will be the ones that do something profoundly counterintuitive: instead of using AI to eliminate human interaction by firing droves of workers, they will use it to create more human interaction. IBM has reversed course on its earlier human redundancy fantasies. I bet more will in due course…
[Simons sketches the history of humans’ intertwined development of both social/organizational and utile technologies, concluding…]
… What the chain reveals is a dependency the AI industry has largely declined to examine. The underlying intelligence of a large language model isn’t a function of its architecture, its parameter count, or the volume of compute thrown at its training. It is not even about the training data. It is a function of the social complexity of the civilization whose language it digested.
Each epoch advanced the cognitive frontier through something far richer and more complex than the isolated genius of an individual guru or machine. It did so through new forms of collective problem-solving. Think new institutions (the Greek agora, the Roman lex, the medieval university, the scientific society, the modern corporation, and the social internet) that demanded and rewarded ever more sophisticated uses of language.
The cognitive anthropologist Edwin Hutchins studied how Navy navigation teams actually think. In his 1995 book Cognition in the Wild, he wrote something that reads today like an accidental prophecy. The physical symbol system, he observed, is “a model of the operation of the sociocultural system from which the human actor has been removed.”
That is, with eerie precision, a description of what a large language model (LLM) really is, stripped of all the unapproachable jargon and mathematical wizardry. An LLM like ChatGPT is a model of human social reasoning with the human wrangled out. And the question nobody in Silicon Valley is asking with sufficient urgency is: What happens to the model when the social reasoning that produced its training data begins to thin?…
[Simons explores evidence that this may already be materially underway, then explores what that “atrophy” might mean …]
… If AI capability depends on the social complexity of human language production—and if AI deployment systematically reduces that complexity through cognitive offloading, homogenization of creative output, and the elimination of interaction-dense work—then the technology is gradually undermining the conditions for its own advancement. Its successes, rather than failures, create a spiral: a slow attenuation of the very substrate it feeds on, spelling doom.
This is the Social Edge Paradox, and the intellectual tradition it draws from is older and more interdisciplinary than most AI commentary acknowledges…
[Simons unpacks that heritage, and puts it into dialogues with recent thoughts from Dario Amodei, Leopold Aschenbrenner, and Sam Altman, concluding…]
… The Social Edge Framework outlined here is a direct counterpoint to Amodei, Aschenbrenner, and Altman. It is a program of action to counter the human redundancy fantasy. It challenges the self-fulfilling doom-spirals created by the premature reallocation of material resources to a vision of AI. I speak of the philosophy that underestimates the sheer amount of human priming needed to support the Great Recode of legacy infrastructure before our current civilization can even benefit substantially from AI advances.
By “Great Recode,” I am paying homage to the simple but widely ignored fact that the overwhelming number of tools and services that advanced AI models still need to produce useful outputs for users are not themselves AI-like and most were built before the high-intensity computing era began with AI. In the unsexy but critical field of PDF parsing—one of the ways in which AI consumes large amounts of historical data to get smart—studies show that only a very small proportion of tools were created using techniques like deep learning that characterize the AI age. And in some important cases, the older tools remain indispensable. Vast investments are thus required to upgrade all or most of these tools—from PDF parsers to database schemas—to align with the pace of high-intensity computing driven by the power-thirst of AI. Yet, we are not at the point where AI can simply create its own dependencies.
Indeed, the so-called “legacy tech debt” supposedly hampering the faster adoption of AI has in many instances been revealed as a problem of mediation and translation. AI companies are learning that they need to hire people who deeply understand legacy systems to guide this Recoding effort. A whole new “digital archaeology” field is emerging where cutting-edge tools like ArgonSense are deployed to try to excavate the latent intelligence in legacy systems and code often after rushed modernization efforts have failed. In many cases, swashbuckling new-age AI adventurers have found that mainframe specialists of a bygone age remain critical, and multidisciplinary dialogues and contentions are essential to progress on the frontier. Hence the strange phenomenon of the COBOL hiring boom. New knowledge must keep feeding on old.
The Social Edge Framework says: yes, scaling matters, architecture matters, and compute matters. But none of these will continue to deliver if the social substrate—the complex, argumentative, institutionally diverse, perspectivally rich fabric of human interaction—is allowed to thin. And thinning is very possible…
… The Social Edge prescription is that organizations that hire more people to work in AI-enriched, high-interaction, and transmediary roles—where AI scaffolds learning rather than substituting it—will derive greater long-term advantage than those that treat the technology as a headcount-reduction device. In a world where raw cognitive throughput has been commodified, the value arc shifts to something considerably harder to replicate: the capacity to coordinate human intent with precision, speed, and genuine depth. That edge lies in trans-mediation and high human interactionism.
The AI industry is telling a story about the future of work that goes roughly like this: automate what can be automated, augment what remains, and trust that the productivity gains will compound into a wealthier, more efficient world.
The Social Edge Framework tells a different story. It says: the intelligence we are automating was never ours alone. It was forged in conversation, argument, institutional friction, and collaborative struggle. It lives in the spaces between people, and it shows up in AI capabilities only because those spaces were rich enough to leave linguistic traces worth learning from.
Every time a company automates an entry-level role, it saves a salary and loses a learning curve, unless it compensates. Every time a knowledge worker delegates a draft to an AI without engaging critically, the statistical thinning of the organizational record advances by an imperceptible increment. Every time an organization mistakes polished output for strategic progress, it consumes cognitive surplus without generating new knowledge.
None of these individual acts is catastrophic. However, their compound effect may be.
The organizations that will thrive in the next decade are not those with the highest AI utilization rates. They are those that understand something the epoch-chaining thought experiment makes vivid: that AI’s capabilities are an inheritance from the complexity of human social life. And inheritances, if consumed without reinvestment, eventually run out. This is particularly critical as AI becomes heavily customized for our organizational culture.
Making the right strategic choices about AI is going to become a defining trait in leadership. Bloom et al. cross-country research has long established that management quality explains a substantial share of productivity variance between teams and organizations, and even countries.
In the AI age, small differences in leadership quality can generate large differences in outcomes—a non-linear payoff I call convex leadership. The term is borrowed from options mathematics, where a convex payoff is one whose upside accelerates faster than the downside decelerates. Convex leaders convert cognitive abundance into structural ambition and thus avoid turning their creative and discovery pipelines into stagnant pools of polished busywork. Conversely, in organizations led by what we might call concave leaders—cautious, procedurally anchored, optimizing for error-avoidance—AI would tend to produce more noise than signal. Because leadership is such a major shaper of all our lives, it is in our interest to pay serious attention to its evolution in this new age.
The Social Edge is more than a metaphor. It is the literal boundary between what AI can do well and what it will keep struggling with due to fundamental internal contradictions. Furthermore, the framework asks us all to pay attention to how the very investment thesis behind AI also contains the seeds of its own failure. And it reminds leaders that AI’s frontier today is set by the richness of the social world that produced the data it learned from…
Eminently worth reading in full: “The Social Edge of Intelligence.”
Consider also the complementary perspectives in “What will be scarce?,” from Alex Imas (via Tim O’Reilly/ @timoreilly.bsky.social)… and in the second piece featured last Monday: ““Curiosity Is No Solo Act.“
Apposite: “Some Unintended Consequences Of AI,” from Quentin Hardy.
And finally, from the estimable Nathan Gardels, a suggestion that Open AI’s recent paper on industrial policy for the Age of AI fills a vacuum left by an unimaginative political class and should be taken seriously, at least as a conversation starter: “OpenAI Proposes A ‘Social Contract’ For The Intelligence Age.”
* Old agricultural proverb
###
As we take the long view, we might recall that today is the anniverary of a techological advance that both fed the social edge and encouraged the build out of the technostructure from which today’s AI hatched: on this date in 1993 Version 1.0 of the web browser Mosaic was released by the National Center for Supercomputing Applications. It was the first software to provide a graphical user interface for the emerging World Wide Web, including the ability to display inline graphics.
The lead Mosaic developer was Marc Andreesen, one of the future founders of Netscape, and now a principal at the venture capital firm Andreessen Horowitz (AKA “a16z”)… where he has been become a major investor in, promoter of, and politicial champion of the current crop of AI firms.
“The secret of longevity is to keep breathing”*…
Shelly Fan, with news of a new study that proports to gauge the limits of longevity…
In 1997, Jeanne Calment passed away at the age of 122 and a half. The longest living human documented to date, she pushed the boundary of what was previously considered the maximum human lifespan.
Meanwhile, in 2023, Guinness World Records recognized Pat the mouse as the oldest mouse alive at a little over nine and a half years old—just a sliver in years compared to humans.
When it comes to lifespan, we mammals have an astonishing range. The common shrew lives less than two years; bowhead whales thrive for at least 211 years. Why the discrepancy?
Part of it, according to Dr. Steve Horvath and colleagues at the University of California, Los Angeles, comes down to epigenetics: the chemical tags attached to DNA that flip genes on or off. The type and position of these tags shift through major life events—puberty, aging—and even with dietary changes.
Unlike genetics, the study of genes coded in DNA, epigenetics better captures the “here and now” of gene expression as we go through life. Previously, Horvath and others have tapped epigenetics to develop “aging clocks” that predict a person’s biological age—that is, how old your body is biologically, rather than the number of candles on your birthday cake.
In a new study in Science Advances, Horvath’s team expanded their epigenetic clocks to predict three life-changing traits: gestation time—how long the next generation fully grows in the womb—puberty, and maximal lifespan.
“Many have suggested that epigenetic mechanisms play a role in determining lifespan,” wrote the team in the paper.
Taking advantage of data from the Mammalian Methylation Consortium, they analyzed one type of epigenetic modification in over 15,000 tissue samples across 348 mammals and developed multiple epigenetic predictors for the three life-history traits across species.
The predictors were reliable. When challenged with lifestyle and demographic factors often associated with changing epigenetic markers—for example, weight, race, and biological sex—they retained their accuracy. Surprisingly, even notable methods for extending lifespan in the lab, for example, caloric restriction, had little effect on the clock’s measures.
“This [epigenetic] signature may be an intrinsic property of each species that is difficult to change,” the team wrote…
More at: “New ‘Aging Clock’ Predicts the Maximum Lifespan of 348 Mammals Including Humans,” from @ShellyFan in @singularityu.
The underlying paper, “Epigenetic predictors of species maximum life span and other life-history traits in mammals,” is here.
* Sophie Tucker
###
As we age, we might send quixotic birthday greetings to Roy Walford; he was born on this date in 1924. A professor of pathology (also at UCLA), he was a pioneer in arguing for calorie restriction as a way of extending life (and a crew member of Biosphere 2.)
Walford died in 2004 at the age of 79 (though in fairness, his demise was a result of Lou Gehrig’s disease, which a could have been the result of low oxygen, high nitrous oxide levels in the Biosphere, causing the loss of brain cells).
“Don’t let us forget that the causes of human actions are usually immeasurably more complex and varied than our subsequent explanations of them”*…
Further, in a fashion, to yesterday’s post: Patricia Fara explains how the tension between religion and science as arbiters of knowledge came to head in the French Revolution, and how that inspired Lambert Adolphe Jacques Quetelet, a Belgian astronomer, mathematician, statistician, and sociologist, to introduce a radically new way of thinking about human beings:
… God had been forcefully excluded from astronomy during the French Revolution, when Pierre-Simon Laplace rewrote Newton’s ideas to create his deterministic cosmos, in which scientific laws govern every movement of every planet with no need for divine intervention. Inspired by this success, a Belgian astronomer called Alphonse Queteler decided that human societies are also controlled by laws. Each country has its own statistical patterns that remain constant from year to year–suicide and crime rates, for instance–and so Quetelet suggested that an ‘average man’ can consistently encapsulate a nation’s characteristics. Politicians should, Quetelet prescribed, operate like social physicists and try to improve average behaviour rather than worry about extreme anomalies. For him, variations from the statistical mean were–like planetary wobbles–imperfections to be smoothed out so that overall progress could be ensured.
Quetelet had introduced a radically new way of thinking about human beings. As one of his admirers put it, ‘Man is seen to be an enigma only as an individual, in mass, he is a mathematical problem.’ Quetelet’s successors took his ideas in many different directions. For one thing, his work was valuable politically because it could be interpreted in different ways. While conservatives insisted that little could be done to alter the current system, radicals accused governments of impeding the natural course of progress, and Utopians–such as Karl Marx–envisaged harmonious societies governed by nature’s own laws guaranteeing improvement. Data collection projects proliferated, and statisticians searched for laws governing every aspect of life, ranging from the weather to the growth of civilization, from stock market fluctuations to the incidence of disease. Many scientists took their ideas from Quetelet rather than from abstract textbooks–but they added their own twist. Whereas Quetelet regarded individual deviations from the norm as errors to be eliminated, scientists set out to study how variations occur…
An excerpt from Fara’s Science: A Four Thousand Year History, via the invaluable Delanceyplace.com (@delanceyplace): “God, Science, and Data.”
* Fyodor Dostoevsky, The Idiot
###
As we focus on frames, we might spare a thought for a man who kept his eye on the individual, Wilhelm Reich. A medical doctor and psychoanalyst, he was a member of the second generation of analysts after Sigmund Freud. Reich developed a system of psychoanalysis concentrating on overall character structure, rather than on individual neurotic symptoms. His early work on psychoanalytic technique was overshadowed by his involvement in the sexual-politics movement and by “orgonomy,” a pseudoscientific system he developed. He also built a device he called a cloud buster, with which he claimed he could manipulate the weather by manipulating the “orgone” in the atmosphere. Reich’s claims aroused much controversy; and he was taken to court for fraud by the Food and Drug Administration (FDA). The court ordered his books and research burned and his equipment destroyed. Reich was sentenced to prison where he died of heart failure on this date in 1957.
“Many of the things you can count, don’t count. Many of the things you can’t count, really count”*…
Still, we count… and have, as Keith Houston explains, for much, if not most of human history…
Figuring out when humans began to count systematically, with purpose, is not easy. Our first real clues are a handful of curious, carved bones dating from the final few millennia of the three-million-year expanse of the Old Stone Age, or Paleolithic era. Those bones are humanity’s first pocket calculators: For the prehistoric humans who carved them, they were mathematical notebooks and counting aids rolled into one. For the anthropologists who unearthed them thousands of years later, they were proof that our ability to count had manifested itself no later than 40,000 years ago.
…
Counting, fundamentally, is the act of assigning distinct labels to each member of a group of similar things to convey either the size of that group or the position of individual items within it. The first type of counting yields cardinal numbers such as “one,” “two,” and “three”; the second gives ordinals such as “first,” “second,” and “third.”
At first, our hominid ancestors probably did not count very high. Many body parts present themselves in pairs—arms, hands, eyes, ears, and so on—thereby leading to an innate familiarity with the concept of a pair and, by extension, the numbers 1 and 2. But when those hominids regarded the wider world, they did not yet find a need to count much higher. One wolf is manageable; two wolves are a challenge; any more than that and time spent counting wolves is better spent making oneself scarce. The result is that the very smallest whole numbers have a special place in human culture, and especially in language. English, for instance, has a host of specialized terms centered around twoness: a brace of pheasants; a team of horses; a yoke of oxen; a pair of, well, anything. An ancient Greek could employ specific plurals to distinguish between groups of one, two, and many friends (ho philos, to philo, and hoi philoi). In Latin, the numbers 1 to 4 get special treatment, much as “one” and “two” correspond to “first” and “second,” while “three” and “four” correspond directly with “third” and “fourth.” The Romans extended that special treatment into their day-to-day lives: after their first four sons, a Roman family would typically name the rest by number (Quintus, Sextus, Septimus, and so forth), and only the first four months of the early Roman calendar had proper names. Even tally marks, the age-old “five-barred gate” used to score card games or track rounds of drinks, speaks of a deep-seated need to keep things simple.
Counting in the prehistoric world would have been intimately bound to the actual, not the abstract. Some languages still bear traces of this: a speaker of Fijian may say doko to mean “one hundred mulberry bushes,” but also koro to mean “one hundred coconuts.” Germans will talk about a Faden, meaning a length of thread about the same width as an adult’s outstretched arms. The Japanese count different kinds of things in different ways: there are separate sequences of cardinal numbers for books; for other bundles of paper such as magazines and newspapers; for cars, appliances, bicycles, and similar machines; for animals and demons; for long, thin objects such as pencils or rivers; for small, round objects; for people; and more.
Gradually, as our day-to-day lives took on more structure and sophistication, so, too, did our ability to count. When farming a herd of livestock, for example, keeping track of the number of one’s sheep or goats was of paramount importance, and as humans divided themselves more rigidly into groups of friends and foes, those who could count allies and enemies had an advantage over those who could not. Number words graduated from being labels for physical objects into abstract concepts that floated around in the mental ether until they were assigned to actual things.
Even so, we still have no real idea how early humans started to count in the first place. Did they gesture? Speak? Gather pebbles in the correct amount? To form an educated guess, anthropologists have turned to those tribes and peoples isolated from the greater body of humanity, whether by accident of geography or deliberate seclusion. The conclusion they reached is simple. We learned to count with our fingers…
From an excerpt from Houston’s new book, Empire of the Sum: The Rise and Reign of the Pocket Calculator: “The Early History of Counting,” @OrkneyDullard in @laphamsquart.
* Albert Einstein
###
As we tally, we might send carefully calculated birthday greetings to Stephen Wolfram; he was born on this date in 1959. A computer scientist, mathematician, physicist, and businessman, he has made contributions to all of these fields. But he is probably best known for his creation of the software system Mathematica (a kind of “idea processor” that allows scientists and technologists to work fluidly in equations, code, and text), which is linked to WolframAlpha (an online answer engine that provides additional data, some of which is kept updated in real time).
“Homo sapiens, the only creature endowed with reason, is also the only creature to pin its existence on things unreasonable”*…
How, Sarah Constantin asks, did we humans get so smart?
If you zoom way out and look at the history of life on Earth, humans evolved incredibly recently. The Hominidae — the family that includes orangutans, chimpanzees, bonobos, gorillas, and humans — only arose 20 million years ago, in the most recent 0.5% of evolutionary history.
Within the Hominidae, in turn, Homo sapiens is a very recent development [see image at top]. We appeared 800,000-300,000 years ago, or in the last 1.5%-5.3% of hominid history.
If you look at early hominid “technological” milestones like tool use or cooking, though, they’re a lot more spread out over time. That’s interesting.
There’s nothing to suggest that a single physical change in brains should have given us both tool use and fire, for instance; if that were the case, you’d expect to see them show up at the same time.
Purposeful problem-solving behaviors like tool use and cooking are not unique to hominids; some other mammals and birds use tools, and lots of vertebrates (including birds and fish) can learn to solve puzzles to get a food reward. The general class of “problem-solving behavior” that we see, to one degree or another, in many vertebrates, doesn’t seem to have arisen surprisingly fast compared to the existence of animals in general.
However, to the extent that Homo sapiens has unique cognitive abilities, those did show up surprisingly recently, and it makes sense to privilege the hypothesis that they have a common physical cause.
So what are these special human-unique cognitive abilities?…
“Is Human Intelligence Simple? Part 1: Evolution and Archaeology,” from @s_r_constantin. Part 2 is here.
* Henri Bergson
###
As we study our species, we might send self-examining birthday greetings to Giambattista Vico; he was born on this date in 1668. A political philosopher, rhetorician, historian, and jurist, Vico was one of the greatest Enlightenment thinkers. Best known for the Scienza Nuova (1725, often published in English as New Science), he famously criticized the expansion and development of modern rationalism and was an apologist for classical antiquity.
He was an important precursor of systemic and complexity thinking (as opposed to Cartesian analysis and other kinds of reductionism); and he can be credited with the first exposition of the fundamental aspects of social science (and so, is considered by many to be the first forerunner of cultural anthropology and ethnography), though his views did not necessarily influence the first social scientists. Vico is often claimed to have fathered modern philosophy of history (although the term is not found in his text; Vico speaks of a “history of philosophy narrated philosophically’). While he was not strictly speaking a historicist, interest in him has been driven by historicists (like Isaiah Berlin).
You must be logged in to post a comment.