Category: AI

Civilization, Erasure, and Immaculate Conceptions

Mark DavisLeave a comment

Jeff Koons’ Venus figure from Barcelona’s MOCO Museum. You can see me with my phone trapped in her belly.

There is a persistent fear of societal erasure that permeates contemporary social and cultural criticism. South Korea, Taiwan, Italy, and Eastern Europe all have very low birthrates and face the reality of drastic change or erasure in the next fifty or so years. Even the leaden National Security Strategy (NSS) document of the Trump administration outright proclaims there is a “stark prospect of civilizational erasure” in Europe due to low birthrates and immigration. In the United States there is worry over young people not being interested in sex, and birthrates, while not as low as in other countries, have not been at replacement rate for more than a decade. Among working-class whites in America, young women often just find their male counterparts not good husband or father material. The same is true of college-educated women, who increasingly outnumber their slacker male counterparts

Some commentators reflect that what is needed is a religious revival that somehow reverses the liberation of women and gets them back to raising children in a celebration of God’s aims. Less dreamy policy ideas include paying couples for children and eliminating the economic barriers to working women who have children.

There is, of course, an inherent racism built into some of these fears; the thought is that American or Japan or Italy are losing something important if waves of immigrants gradually come to dominate the civilizations. The decline and fall of Western civilization is driven not by the LA punk scene but by this nebulous replacement idea driven by some kind of global elite (often with antisemitic overtones).… Read the rest

Functional Information Analysis and the Chinese Room

Mark Davis1 Comment

 

I’ve been considering the implications of a new scientific law, the law of increasing functional information, in terms of how it can be applied to our thinking about various ideas. At first glance, the law says little new about the physical world. We already know about much of the various levels of the functions that are described in the paper, from star formation up through the evolution of human behavior. But there may be another way of thinking about it. A quote from Margaret Bowdon on Searle’s famous Chinese Room Argument shows how it might help:

The inherent procedural consequences of any computer program give it a toehold in semantics, where the semantics in question is not denotational, but causal.

So here we have an attack on the underlying assumption that what human understanding amounts to involves semantics and meaning that a robot or computational procedure can never have. If we expand Bowdon’s claim about how meaning comes about to include some of Searle’s other quotes like the room can never know what a hamburger is in Chinese just by processing the relevant symbols, we can enlarge that toehold by including all the functional engagements that are part of the experience of coexisting with and consuming hamburgers in a Chinese-language environment.

Semantics and intentionality and meaning—all these folk concepts we use to express how we are aware and conscious—collapse into function with the impetus driven by this new law. Meaning is an inherent feature of function, we just mystify it a great deal. In fact, a part of the semantics associated with the Chinese Room is embedded in the transfer rules that are used for translation. Whoever developed those rules understood Chinese well-enough to code them up accurately and that represents functional information increase.… Read the rest

After Z

Mark DavisLeave a comment

 

The nights and days collide with violence. There is the nocturnal me and the dazed, daylight version squinting away from the glaring windows. There are the catnaps that lace with riotous algebras. I am addicted to caffeine, or run on it, until even it becomes unpersuasive, and I droop over at the keyboard. Then this pulse of creation pulls me out again, stunned for a few beats, and I grasp my mug and stumble back to the lab floor.

Z keeps changing, day by day, midnights into dawns, and reawakening in clanging novelty. Z is for “zombie,” for it is in the uncanny valley of both a physical and cogitating thing. It perceives, stands, jogs in place beside me on the laboratory floor, an ochre braid of wires bouncing in a dreadlock mass behind it. Z plays chess, folds towels (how hard that was!), argues politics (how insane is that one!), and constantly restructures nuances in its faces and gestures. Sometimes I’m tired and Z is an impertinent teenager. Often there are substitutions and semantic scrambling like a foreigner who mistakes a word for another, then carries on in a fugue of incoherence.

There is a half-acre of supercooled GPUs to the north of the lab where the hot churn of work is happening. It’s a spread of parallel dreamscapes, each funneled the new daily stimuli, stacking them into a training pool, then rerunning the simulations, splitting and recombining, then trying again to minimize the incoherency, the errors, and the size of the model. Of the ten thousand fermenting together, one becomes the new Z for a few hours, but then is gone again by morning, replaced by a child of sorts that harbors the successes but sheds the excesses and broken motifs.… Read the rest

The Hard Problem of the Future

Mark Davis1 Comment

The American zeitgeist is obsessed with decline and a curious sense of ennui. On the progressive left there is the rolling mortal threat of inequality and the destruction of the middle class. Wages don’t keep up with inflation or, more broadly, the cost of living. On the new MAGA right there is an unfocused rage that builds in part on the angst of hollowed-out rural and post-industrial communities, and then in part on undocumented immigrants as scapegoats and symbolic of lefty lawlessness, and again in part as a tirade against wealthy, coastal elites who control the media, universities, and have pushed the Overton window in incremental lurches towards inclusiveness. The populism is mostly half-baked, certainly, and exploited by cynical conservatives for undermining social support while bolstering commercial interests and reducing taxes for the well-to-do. But half-baked is enough for a sensibility; things fully realized are only afterthoughts.

There are other chthonic rumblings and imputations that filter up. The rise of China’s industrial, military, and scientific power is a growing shadow that some see threatening to engulf the world in its umbra. And with it comes the fear of slowing technological might, despite the domination of the recent technological present by the United States. We might be left behind like unhoused, opioid-addicted, modern peasants. The crumbling of the cities would be just punishment even if their loss only cascades the problems of the heartland.

And so as the future keeps getting harder, we turn to mad kings who promise radical change in the face of hard problems. The change can’t possibly be realized, so it is better to just pretend that there are solutions. Annex Greenland, rename the Gulf of Mexico, incorporate Canada, occupy Panama, reach for Mars, acquire territory, but all the while cocooned by the complex institutional and international realities that mean that acting aggressively and alone is now untenable.… Read the rest

All Hail Evolved Irrationality

Mark DavisLeave a comment

From an impersonal distance, watching other people make decisions is always interesting. They may apply reason and passion in varied measures to figure out a way forward, or a lifestyle, or even fashion choices. Often they mindfully weigh the choices and decide to moderate even that process, committing to a radical path that has drama and uncertainty. It’s sometimes better to be interesting than cautious and correct. As a virtue this recalls the romantic movements that arose in opposition to the mechanization of the 19th century as trains and steamships criss-crossed the world. Order and peace were at odds with drama, passion, faith, and love that are somehow in our core, animalistic nature.

Motifs of uncertainty crept into science and reason as we transitioned into the 20th century, from the stochastic rumble of thermodynamics, to the realigning of basic concepts like gravity as a space-time curvature, and then the wave-particle duality of quantum reality. The clockwork of the universe that was so oppressively mechanistic showed fuzzy edges and knotty intersections that defied ordinary-scale expectations. The combined mathematical realizations of incompleteness and algorithmic uncomputability overlaid the investigations of the physical to such an extent that new theories were developed that proposed that the quantum world and mind are inextricably laced together; subjective and objective do not exist independently.

Even as our knowledge and management of the universe has grown, there is a background roil of irrationality, like the primal chaos of Tiamat. Human thought has a collection of ways of organizing the world that appear to be natural consequences of our social development. Religious belief is a widespread catalyst for building predictable and supportive communities by slaving our baser tendencies to coordinating strictures, obligations, and status maintenance.… Read the rest

Rise, Teleonomatons!

Mark Davis2 Comments

My essay for the Berggruen Prize this year. Of course, the organization missed an opportunity to drop down a staggering rabbit hole and lean into a whole new regime of neologistic energetics, but I do like the prize-winning essays!

Rise, Teleonomatons

Meaning entanglements

I can’t figure out what some statements about science mean, though I have a fair background in a range of scientific areas. Now, I can peruse highly technical papers, browse abstracts, interpret graphs, study conclusions, and typically do just fine. The professional stuff is a procession of arid facts and assumes the reader knows the basic definitions for things. It’s the popular versions of scientific insights that befuddle me—especially the definitions that try mightily to bridge meanings for learners. So do introductory texts. Light is a wave? Not exactly. Atoms are particles? Well, not quite, but they are small. Electrons orbit the nucleus? No, really, no. A force is an influence tending to change the motion of a body? OK, but what is an influence? People are influenced, aren’t they? Or under the influence.

And then there are texts like those of existential philosophers that leave me completely befuddled. What is this “Being” that they write about and how did it get so enlarged in significance, capitalized, and shoehorned by the translator into incomprehensible juxtapositions with other bulbous words?

It may be low pedantry to expect clarity from words and writing. We generally just roll with it and use the terms according to conventions inferred from reading and learning. We rush over the imperfect metaphorical bridges, the analogies, the similes. For physics, definitions are bound to the equations and measurement properties that accompany the words of description, and they become a semantic pier that is constantly informing our relationship with the ideas.… Read the rest

Death, Healing, and Language Games

Mark Davis1 Comment

A phone call came in the early afternoon in late August: she was reclined on her day bed and she was dead. She had lain down for a nap and didn’t wake up. In the subsequent weeks there has been a rallying of the families, grief, tremendous effort, flights before dawn, and scripted expressions of condolences. In my youth I had necessarily been a rules deconstructor, going in bare feet to a wedding, challenging expectations, trying to find novel ways to intervene—sometimes boorishly, I’m certain. But now I prize cheerful clarity and just volunteer to do whatever is needed to reach our collective goals. Remember: freedom and coordination.

In moments like this there are somewhat scripted conventions for discussing the hard matters of duties and feelings. These language games have organically arisen from contending forces, from Anglo-American sentimentality to the influence of organized religion, and they serve to facilitate life transitions. And now they have been summarized by large language models (LLMs) like ChatGPT that have trained on the masses of written content on the web to the point that they have reliable consistency. An emergency room doctor reports in the New York Times that ChatGPT does a better job than he does at the hard job of best-practices for bedside manner when conveying bad news. He also notes that LLMs are remarkably reliable for refining the scripted discussion of symptoms and medical diagnoses.

So a counter to the “slop economy” at least that provides some guidance for harried professionals trying to do a good job at the delicate threshold of personal pain and fear. The stochastic parroting is suddenly desirable insofar as it is parroting best practices and conventions.… Read the rest

Indeterminacy and the Ethics of Emergence

Mark DavisLeave a comment

Continuing on with this theme of an ethics of emergence, can we formulate something interesting that does better than just assert that freedom and coordination are inherent virtues in this new scheme? And what does that mean anyway in the dirty details? We certainly see natural, emergent systems that exhibit tight regulatory control where stability, equilibrium, and homeostasis prevent dissipation, like those hoped-for fascist organismic states. There is not much free about these lower level systems, but we think that though they are necessary they are insufficient for the higher-order challenges of a statistically uncertain world. And that uncertainty is what drives the emergence of control systems in the first place. The control breaks out at some level, though, in a kind of teleomatic inspiration, and applies stochastic exploration of the adaptive landscape. Freedom then arises as an additional control level, emergent itself.

We also have this lurking possibility that emergent systems may not be explainable in the same manner that we have come to expect scientific theories to work. Being highly contingent they can only be explained in specificity about their contingent emergence, not by these elegant little explanatory theories that we have now in fields like physics. Stephen Wolfram, and the Santa Fe Institute folks as well, investigated this idea but it has remained inconclusive in its predictive power so far, though that may be changing.

There is an interesting alternative application for deep learning models and, more generally, the application of enormous simulation systems: when emergent complexity is daunting, use simulation to uncover the spectrum of relationships that govern complex system behavior.

Can we apply that to this ethics or virtue system and gain insights from it?… Read the rest

When the Cranes Cry

Mark Davis1 Comment

The crane has a symbolic resonance in Celtic mythology. A magician, assuming an elaborate pose—one eye open and one leg drawn up—was said to see into the otherworld, just as the crane itself moved from sky to land to water. But there is the other meaning of the word crane: the ancient lifting contraption that helped build Greece and likely had a role in Egypt and Sumeria before that. And now they protrude into the urban sky, raising up our buildings and even other cranes as we densify our cities. It was this mechanical meaning that Dan Dennett at Tufts chose to contrast with conceptual skyhooks, the unsupported contrivances that save protagonists in plays by dangling gods above the stage. For Dennett, the building crane is the metaphor we should apply to the mindless, simple algorithm of evolution. The algorithm raises up species and thus creates our mysterious ideas about meaning and purpose. No skyhooks or Deus ex Machina are needed.

Dennett passed away at 82 in Maine leaving a legacy as a public intellectual who engaged in the pursuit of reason throughout his adult career. He was committed to the idea that this world—this teeming ensemble of living matter—is intrinsically miraculous, built up by something dead simple into all the convolutions and perilous ideas that we now use to parse its mysteries. He was one of the Four Horsemen of the Apocalypse during the so-called New Atheism craze of 2008-2010, along with Richard Dawkins, Christopher Hitchens, and Sam Harris, but even then he was committed to the crane metaphor to displace these ancient skyhooks of belief rather than, say, a satirical impact-analysis of religion a la Hitchens.

There is another phrase that Dennett championed in Darwin’s Dangerous Idea: Evolution and the Meanings of Life: universal acid.… Read the rest

Inferred Modular Superparrots

Mark DavisLeave a comment

The buzz about ChatGPT and related efforts has been surprisingly resistant to the standard deflationary pressure of the Gartner hype cycle. Quantum computing definitely fizzled but appears to be moving towards the plateau of productivity with recent expansions of the number of practical qubits available by IBM and Origin in China, as well as additional government funding out of national security interests and fears. But ChatGPT attracted more sustained attention because people can play with it easily without needing to understand something like Shor’s algorithm for factoring integers. Instead, you just feed it a prompt and are amazed that it writes so well. And related image generators are delightful (as above) and may represent a true displacement of creative professionals even at this early stage, with video hallucinators evolving rapidly too.

But are Large Language Models (LLMs) like ChatGPT doing much more than stitching together recorded fragments of texts ingested from an internet-scale corpus of text? Are they inferring patterns that are in any way beyond just being stochastic parrots? And why would scaling up a system result in qualitative new capabilities, if there are any at all?

Some new work covered in Quanta Magazine has some intriguing suggestions that there is a bit more going on in LLMs, although the subtitle contains the word “understanding” that I think is premature. At heart is the idea that as networks scale up given ordering rules that are not highly uniform or correlated they tend to break up into collections of subnetworks that are distinct (substitute “graphs” for networks if you are a specialist). The theory, then, is that the ingest of sufficient magnitudes of text into a sufficiently large network and the error-minimization involved in tuning that network to match output to input also segregates groupings that the Quanta author and researchers at Princeton and DeepMind refer to as skills.… Read the rest