Posts Tagged ‘AI’
“Technology doesn’t force us… it merely opens the door”*…
The estimable Tim O’Reilly reminds us to think deeply about how AI could and should turn out. He suggests that Jeff Ding‘s diffusion theory of the role of technology in great-power competition also applies to AI adoption– and that it suggests that companies obsessed with the frontier might be optimizing for the wrong thing…
In the 1980s, Japan led the world in semiconductors, consumer electronics, and computer hardware, the industries everyone assumed would decide the next phase of economic power. Japan won them and still did not overtake the United States in the information revolution that followed. Jeff Ding, a political scientist at George Washington University, opens his book Technology and the Rise of Great Powers with the history of the first and second industrial revolutions and the third, the information revolution. The explanation he gives for who wins and who loses applies to companies as well as it does to nations, and very much to the current trajectory of AI.
Ding contrasts two theories of how technological revolutions reshape economic power. The conventional one he calls the leading sector model, or LS theory. It goes like this: New technologies create fast-growing new industries like steel and railroads and automobiles and semiconductors, and the country that dominates invention in those sectors captures the monopoly profits and the upstream and downstream economic linkages that come with them. As the story goes, if you win the leading sector, you win the era. Britain won in the first industrial revolution through its mastery of steam power, and then was surpassed by the US in the second through its leadership in electrification, the internal combustion engine, and mass manufacturing. The US kept its lead over Japan in the information systems revolution not by competing in the “leading sector” of electronic hardware but by diffusing “up the stack” via software that took the power of computing into every sector of the economy. (OK, that last bit is my explanation of what happened rather than Ding’s, but it’s consistent with his theory.)
Leading Sector theory is pretty clearly the working hypothesis of today’s AI industry and the national strategy that is forming around that industry. The company and the country with the biggest and best models wins. Everyone else is an also-ran.
Ding offers another explanation, which he calls diffusion theory. He points out that general-purpose technologies, foundational ones like the steam engine, electricity, and the computer, don’t just create massive profits and productivity gains in a single industry but instead spread across the whole economy. National economic leadership comes not from inventing the new sector but from diffusing the general-purpose technology more quickly and more broadly than your rivals. This happens over decades. The win goes to whoever most successfully embeds the technology into a wide range of ordinary productive work. This is how the US kept its lead over Japan rather than being surpassed by it.
This is obviously aligned with the thinking of Arvind Narayanan and Sayash Kapoor in “AI as Normal Technology,” which Ding cites in his book.
A big part of what enables diffusion is what Ding calls skill infrastructure, the education and training systems that widen the pool of people who can actually work with the technology. When the priority is widespread adoption rather than invention, he argues, the institutions that matter are the ones that build engineering skill at scale, standardize good practice, and tie research to industry. He writes:
GPT diffusion theory highlights the importance of GPT [General Purpose Technology] skill infrastructure. Education and training systems that widen the pool of engineering skills and knowledge linked to a GPT. When widespread adoption of GPTs is the priority, it is ordinary engineers, not heroic inventors, who matter.
Music to my ears, as it should be to yours: “It is ordinary engineers, not heroic inventors, who matter.”
That is not how the current AI narrative goes. Everyone is fixated on the labs, the frontier models, and the most famous researchers. And that fixation shapes enterprise strategy. Inside many companies AI strategy is a procurement decision: Which model and which vendor and which flagship tool should we choose? Or it’s a moonshot to stand up a lab and build an impressive demo and hire your own famous developer. Both approaches treat AI as a sector to be won. Ding’s argument is that the breakthrough sector itself is not where the long-term value for national power lives. And I believe that the same applies to corporate success. The value is in how widely and how well the technology gets embedded into the work of the people you already employ. The company that puts AI to work in finance and support and legal and sales and operations, across every unglamorous process, as well as in product and engineering, outperforms its competitors and drives its industry forward.
The reason diffusion takes a long time is that it is an organizational problem and not a technical one…
[Tim elaborates, and specifies the requirements for successful management of what is an “enterprise transformation problem”; he then unpacks the geopolitics of AI. He concludes…]
… Sovereign AI is not just a matter of national power. It is a predictable consequence of diffusion. A technology that diffuses widely will be adapted by different societies, firms, and institutions to suit their own needs, values, and constraints. Sovereign AI is AI designed for diffusion, not just raw increases in capability.
This is one reason the arms-race framing is unhelpful. It encourages us to treat AI as if it were a weapons system or a scarce strategic asset. But if AI is closer to electrification, computing, or the written word, the important thing is how the technology is embedded into the ordinary life of economies and institutions, and whether that embedding happens in ways that increase agency broadly rather than concentrating it in a few hyperpowerful companies.
There are a few additional lessons we can take from the history of electrification. While motors became decentralized, factories stopped generating their own power and bought it from a centralized grid. The unit-drive revolution decentralized application, not generation. This limitation, which we are now working to overcome to some extent with decentralized solar generation, is perhaps ironically showing up most strongly in the strain that AI data centers are placing on the grid. Let’s learn from that misstep. You can diffuse AI into every workflow via API calls to a big centralized model, or it can be diffused by a network of smaller models that turbocharge every part of the economy.
We should design for a future of multiple AIs, not a single universal system. Different countries will want systems shaped by different legal regimes, languages, histories, and cultural assumptions. So will companies. So will professions and communities of practice. The instinct of some frontier labs is to imagine that the right answer is to homogenize the technology, purge it of bias, and offer a single sanitized intelligence layer for the world. But AI is a social and cultural technology. The differences are not a defect to be smoothed away.
We do need to think about standards and interoperability. The historical analogy that comes to mind is railroad gauge. When real world systems are built to incompatible standards, the result is not healthy diversity but decades of friction, kludges, and retrofitting. The same may prove true for AI. If we force the future into a choice between one universal model and a patchwork of disconnected sovereign systems, we will get the worst of both worlds. We need a layer between uniformity and fragmentation, which can come from standardized protocols that allow different models, tools, and institutions to interoperate without requiring them to become identical.
This is also why open source matters, but only if it is properly understood. Open source is not just about licenses. My earliest introduction to the shared development of software that now goes by that name came from the research community that grew up around Bell Labs’ Unix operating system despite AT&T’s proprietary (albeit permissive) licensing. Because of that experience, I became convinced that it was the modular, protocol-centric architecture of Unix that was a key driver of collaborative, internet-enabled software development.
Open source AI depends on far more than open models. It depends on the architecture of participation built into the systems above and around them: the protocols, servers, interfaces, and shared technical conventions that let many different actors build on common foundations. The Open Source AI Gap Map shows just how rich that open source AI ecosystem is becoming. But open source can also coexist with proprietary, de facto standards like the OpenAI and Anthropic APIs. Like the electric grid we are now beginning to rebuild, the AI future will be a mix of centralized and decentralized systems. Cooperation and competition can coexist. Different actors can build different systems, for different purposes, under different forms of governance, while still participating in a shared technical and economic order.
This is how the future can belong not just to the inventors of AI but to the people who make it usable, adaptable, interoperable, and worth adopting.
Eminently worth reading in full. AI for all of us: “Ordinary Engineers, Not Heroic Inventors,” from @timoreilly.bsky.social
Apposite: “How to talk about “AI” without adding to the anthropomorphization“
###
As we amplify access, we might we might spare a thought for someone who launched more than one central technology into braod diffusion: the Serbian-American electrical engineer and inventor Nikola Tesla; he died on this date in 1943. Tesla is probably best remembered for his rivalry with Thomas Edison: Tesla invented and patented the first AC motor and generator (c.f.: Niagara Falls); Edison promoted DC power… and went to great lengths to discredit Tesla and his approach. In the end, of course, Tesla was right.
Tesla patented over 300 inventions worldwide, though he kept many of his creations out of the patent system to protect their confidentiality. His work ranged widely, from technology critical to the development of radio to the first remote control. At the turn of the century, Tesla designed and began planning a “worldwide wireless communications system” that was backed by J.P. Morgan… until Morgan lost confidence and pulled out. “Cyberspace,” as described by the likes of William Gibson and Neal Stephenson, is largely prefigured in Tesla’s plan. On Tesla’s 75th birthday in 1931, Time put him on its cover, captioned “All the world’s his power house.” He received congratulatory letters from Albert Einstein and more than 70 other pioneers in science and engineering. But Tesla’s talent ran far, far ahead of his luck. He died penniless in Room 3327 of the New Yorker Hotel.
“The remedy is worse than the disease”*…
Your correspondent is a co-founder and director of Common Sense Media, an organization devoted to understanding the fraught media environment in which we live, to helping children, their families, and educators navigate it, and to advocating for “common sense” protections for kids. When we began, over 20 years ago, the terrain was largely TV, movies, books, and video games. They remain, of course, but the spotlight moved to social media, and now, to AI.
Advocacy, as it turns out, can be a tricky needle to thread. As Francis Bacon (and following him, a stream of others) observe, sometimes “the cure can be worse that the disease.” And that’s especially true when the “disease” one is fighting is the province of a powerful, richly funded cabal who market their moves as cures, even though, through their lens, disease is the desired state.
By way of one (chilling) example, Joe Wilkins outlines one dimension of the current state-of-play…
The White House and Congress are working on a deal to clamp down on what US citizens are allowed to say online. According to new reporting by Axios, the Trump administration is negotiating with key senators in an effort to shoehorn a massive legislation package which would limit states’ abilities to regulate AI in exchange for placing broad federal limits on digital speech.
Plenty of ink has been spilled about the Trump administration’s push to revoke AI regulation from individual states. Though the White House and its allies frame this as a matter of “safety” and “national security,” the timing is telling, coming as progressive state governments move to restrict the building of AI data centers and hold tech companies liable for harms their AI systems cause.
What makes this deal particularly insidious is the trade-off at its core. Per Axios, congressional lawmakers lead by Republican Marsha Blackburn are essentially offering to surrender their ability to regulate AI in exchange for three federal censorship bills: the Kids Online Safety Act, the NO FAKES Act, and a federal age verification mandate.
While the language around these three measures suggests common-sense cyber regulation, activists say they really amount to a massive censorship regime that is fundamentally anti-democratic.
In a blistering statement, the first amendment group Foundation for Individual Rights and Expression (FIRE) warns that “taken together, these bills would fundamentally change the internet as we know it.” That’s especially striking because FIRE is funded by conservative private interests like billionaire Charles Koch, meaning its opposition highlights the degree to which Trump is butting heads against fellow members of the US ruling class.
The Kids Online Safety Act (KORA) for example, would force social media giants to restrict lawful speech based on Federal Trade Commission regulations. FIRE complains that this would give the federal government too much power to hold platforms like Meta accountable for their harms — which, many would argue, is long overdue. But given that just a few corporations own the vast majority of web infrastructure and social media, KORA would give the Trump-controlled FTC major power to bend the internet to its will.
To put this power into perspective, just regulating Meta’s Instagram would impact about 71 percent of US citizens who say they regularly use the app.
It’s a powerful weapon, in other words — and like any weapon, it matters a great deal who wields it. Should the White House and Congress push KORA through, it’s likely it would functionally end the possibility of surfing the internet anonymously, while supercharging Trump’s efforts to criminalize left wing opposition groups in the US.
Whether it passes will depend on some shrewd maneuvering on the Trump administration’s part to secure Congressional support. But it all underscores a frustrating reality of AI regulation in the US: Americans overwhelmingly support stricter regulation on AI — but with the current cabal running the Oval Office, there’s no guarantee that the cure won’t be worse than the disease…
Common Sense Media continues to fight– both overt malignancies and the faux remedies that promise to be as bad or worse. So, I would argue, should we all.
Pending federal bills would fundamentally change the internet as we know it: “Trump Moves to Deeply Censor the Entire Internet,” from @joeonhere.bsky.social in @futurism.com.
A resonant reminder of the poisonous program to be resisted: “The Side That Won the Civil War is Now Banning Books About Why the Civil War Was Fought.”
And a look at the politicial economics that may drive it.
(Image above: source)
###
As we seek sanity, we might recall that it was on this date in 1788, with the ratification of the ninth state (New Hampshire), that the Constitution of the United States came into effect. It was, of course, briskly amended to include what we call the Bill of Rights— the first 10 of 27 amendments (so far).
“Common sense is not so common”*…
The Enlightenment is under attack by the Left and the Right. It can only be “saved,” Eliane Glaser argues, through use of its greatest legacy: permanent critique. And then there’s AI. After summarizing the critiques from both sides, she continues…
In consequence of this pincer-movement attack, the Enlightenment’s legacy is existentially vulnerable. It makes me deeply worried as someone whose entire career has been built on trying to understand and analyse the world around me – especially a world that still tries to confine thinking women to the realms of emotion and ‘personal experience’.
I believe that Enlightenment values are essential, but that we have largely forgotten how to make a good case for them: we need to rely on shared facts, tested by experiment; a public sphere where open discussion can take place; and the belief that discussion should be founded on reasoned argument. We need, moreover, to cherish the more political values of tolerance, freedom, human rights and the common good. Advocates for artificial intelligence have the temerity to claim that large language models are ushering in a ‘second Enlightenment’ (a claim that was uncritically echoed in a paper published by the World Economic Forum last year) when what we are in fact seeing is the destruction of the Enlightenment legacy under the false banner of its name. As the historian David Bell argued in The New York Times in 2025, AI is actually ‘shedding Enlightenment values’ by simply reinforcing ‘what we already think we know.’ In The Guardian,the journalist and geopolitical risk consultant Joseph de Weck warned that ‘AI is taking us back to the dark ages’, making us lazy, and stymying independent thinking.
The evidence suggests that we are going through a rapid de-enlightenment. Newspaper circulations, attention spans, and trust in forms of agreed knowledge are in freefall. Misinformation, disinformation and deepfakes are gaining ground. If we let go of the valuable aspects of the Enlightenment project, we open ourselves up to a world of AI blather, ‘my truth’ pronouncements, wobbly sentiment and unchecked power.
My unease with this parlous state of affairs has provoked me to go back and rethink the Enlightenment and what it has to offer. But, rather than unthinkingly recouping it as a mission, I want instead to tease out and weigh up its merits, to discern with nuance what is still fit for our times. I want to ask if it is possible to rescue the Enlightenment’s rallying power, and if it’s worth defending what the combined forces of Left and Right are coming together to attack. Are the Enlightenment’s deficiencies barnacles on an old ship, or integral to its design?…
And so she does. Do read on: “Flickering Enlightenment,” from @elianeglaser.bsky.social in @aeon.co.
* Voltaire
###
As we reclaim reason, we might spare a thought for a glorious product of the Enlightenment, Joseph Haydn; he died on this date in 1809. A composer of the Classical period, he was pivotal in the evolution of chamber music forms like the string quartet and piano trio, and is known as the “Father” of both the symphony and sonata forms. Haydn was a friend and mentor of Mozart, and a teacher of Beethoven; indeed, the Haydn, Mozart and Beethoven trio are sometimes referred to as the “First Viennese School.”
Schonberg wrote that Haydn “was the Classic performer par excellence, and in his long life, from 1732 to 1809, he grew up with the new musical ideas and, more than any one man, shaped them.”
“Gambling is a tax on ignorance”*…
And as Einstein observed, “two things are infinite: the universe and human stupidity; and I’m not sure about the universe.”
Gambling– and related specualtive investments– have always been, for the vast majority of punters, a sucker’s bet. But, as Paul Kedrosky explains, the growing prevalence of AI and the emergence of prediction markets have amplified that painful reality…
The return skew in prediction markets’ returns is startling. It is partly a function of their nature, but also of vibe-coding script kiddies attacking every market anomaly as quickly as it arises. Check a recent WSJ article for examples.
The same dynamic is now spreading across retail-dominated markets. A driver is how AI lowers the cost of systematic exploitation and exploration to near zero. What used to require infrastructure, data pipelines, and bearded quants is now accessible via off-the-shelf models, APIs, and loosely stitched “agent” workflows doing … stuff that even their users don’t fully understand.
The result isn’t democratization of returns. It is wider participation, of a sort, alongside the rapid re-concentration of profits. A small subset of users—those willing to iterate fastest, monitor continuously, and deploy capital programmatically—capture gains, with everyone else just liquidity.
They scrape sentiment, parse new information, and reprice positions in seconds, compressing the half-life of mispricings. That doesn’t eliminate inefficiency, but changes who harvests it. The edge shifts from insight to speed, coverage, and execution discipline—areas where even modest automation compounds quickly, and edges disappear overnight.
Prediction markets are simply the cleanest expression of this trend because they combine thin liquidity, discrete outcomes, and high retail participation. But the same pattern is visible in options flow, single-stock volatility events, and even online poker, which AI increasingly dominates.
As AI tools continue to scale, expect this to get worse: a small cohort running semi-automated strategies extracting semi-consistent edge, and a much larger base supplying them returns. Under the pressure of AI prevalance, markets don’t flatten, the return gradient steepens to a cliff…
Fewer and fewer winners take more and more of the pot. The mechanics of concentration: “AI is Eating Markets” from @paulkedrosky.com.
* Warren Buffett
###
As we contemplate concentration, we might note that today is Mother’s Day. As noted yesterday, the observance became official on that date in 1914. But the quest to honor moms began a good bit earlier. On this date in 1908, Anna Jarvis held a memorial for her mother at St. Andrew’s Methodist Church in Grafton, West Virginia, the location of the International Mother’s Day Shrine. But her quest to create Mother’s Day had begun three years earlier when her mother Ann, a lifelong activist, died.
Ann had tried to start a “Mother’s Remembrance Day” in the mid-19th century. On her passing, Anna enlisted the support of retailer extraordinaire John Wanamaker, who knew a merchandising opportunity when he saw one, and who hosted the first Mother’s Day ceremonies in his Philadelphia emporium’s auditorium. In 1912, Anna trademarked the phrases “second Sunday in May” and “Mother’s Day”, and created the Mother’s Day International Association. By 1914, she and Wanamaker had built sufficient support in Congress to score the Congressional Resolution noted yesterday. (President Wilson, who was by current accounts uninterested in the move– distracted as he was by the beginnings of his ultimately unsuccessful effort to keep the U.S. out of the troubles in Europe that became World War I– nonetheless knew better than to take a stand against moms.)
“Everything is ephemeral, both that which remembers and that which is remembered”*…
Sally O’Reilly on “gray literature,” why it fascinates her… and how an early “AI” attempt to harvest it misses the mark…
In my modest collection of gray literature, the specialist title that comes closest to a blockbuster is Jean Aspin’s Vaginal Examination: A Unique Pocket Guide (ca. 1980s). Or perhaps it’s Dovea Genetics’s Beef Directory (2014).
Aspin was a community midwife in Luton and Dunstable University Hospital’s maternity wing. Her pocket guide is a well-produced, ring-bound, wipe-clean, tongue-shaped booklet, published by the baby milk company Cow & Gate. Its Latinate lists, labeled diagrams, and die-cut holes of increasing diameters, representing vaginal dilation, step a midwife through the assessment of fetal skull position during labor. The Beef Directory promises “Rock Solid Beef Genetics.” It peddles not anonymous meat but the sperm of individual bulls with names that sound like variety acts: Tonroe Lord Ian! Utile Ben! Virginia Andy! Vagabond! Mornity Handyman! Pinocchio! Seaview Tommy! Atok Socrates! Kilowatt D’Ochain! Immense D’Yvoir! It is richly illustrated, suitably glossy, and a chilling ode to muscle. (Behold the bulging rumps of Belgian Blues!)
Among the most niche in my collection of niche titles is the UK Ministry of Defence’s Corrosion: R.A.F. and A.A.C. Aircraft (1966), a bone-dry primer on the control, rectification, and treatment of nine types of corrosion. The Kent County Constabulary’s booklet Special Constabulary Inter-Divisional Competition (1971) is possibly the least read of all. Copied from typewritten documents, with hand-drawn diagrams, and stapled between two pieces of medium-weight red card, the booklet was produced “to enable officials and spectators to follow the progress of the Competition and the fortunes of the teams” during a public event at a Kent police station.3 Fun-seekers watched on as teams, comprising police officers from different divisions within the county, underwent an inspection of uniforms and accoutrements, competed in a quiz, and responded to a hypothetical incident at a demonstration involving a vicar, an unconscious policeman, a drug-addled youth, and an old man with a loaded shotgun.
Gray literature is a diffuse genre. Informational at base, its tone might tend toward bouncy sales patter or flinty authoritativeness. Visually, it ranges between perfunctory pragmatics, rickety flamboyant amateurism, and the polish of corporate comms. The most reliable way to identify an item’s grayness is by its function and milieu. According to the 2010 Prague definition, established at the 12th Annual Conference on Grey Literature and Repositories,
Grey literature stands for manifold document types produced on all levels of government, academics, business and industry in print and electronic formats that are protected by intellectual property rights, of sufficient quality to be collected and preserved by library holdings or institutional repositories, but not controlled by commercial publishers i.e., where publishing is not the primary activity of the producing body.
Gray literature does not have the market or cultural value of a novel or textbook. It is not an end in itself, but facilitative paraphernalia of some other endeavor—midwifery, policing, animal husbandry, war. This vicariousness, and its heterogenous forms, makes it notoriously difficult to place in library catalogues. Should a practical primer on the mitigation of corrosion in airplanes be placed under Dewey Decimal class “671: Metalworking & primary metal products,” “387: Water, air & space transportation,” or “358: Air and other specialized forces”? When a bull sperm directory is a matter of genetics, food production, and commerce, which can it be said to be about? Gray literature isn’t made with libraries or bookshops in mind. It strides out into the world to do an honest day’s work. None of this hanging around on hushed shelves waiting to impart knowledge in the abstract. It’s got sperm to tout, babies to birth, aircraft to maintain, a policed public to mollify.
I find these publications compelling by their very existence and, for the most part, unreadable. Their content slides off my mind. Gray literature’s high and narrow window onto specialist processes is anathema to traditional general-interest non-fiction publishing, which delivers information like a tap dispenses safely managed water—filtered, chlorinated, and piped into your very own quarters. Gray literature is a sploshing bucket of someone else’s water, murky with unfamiliar vocabulary, its means of application not always entirely obvious. Each publication is an invitation to speculate on a sector’s operations, to marvel at the specificity of other people’s knowledge and the focus of their working lives. My paltry library gestures toward the infinite complicatedness of human activity and the vast, disorganized array of murky buckets out of which the materiality of our lives somehow continues to emerge.
I have recently acquired some items that confuse the already untidy category of grayness. While seeking out books on theatrical quick-change (more on that another time), I came across the Webster’s Timeline History series and, out of curiosity, bought the three cheapest of the second-hand editions available: Wallpaper, 1768–2007; Secrecy, 393 BC–2007; and Bristol, 1000–1893. They are collations of excerpts, references, and citations that feature their titular word or phrase, and there are thousands of them. The series’s aggregate subject matter reads like the archest of list poems, the word associations of a disheveled mind, or dying humanity’s life flashing before its eyes…
Do read on for a fascinating/horrifying/illuminating tale all-too-relevant to our times– the story of the Webster’s Timeline History series…
On gray literature and Webster’s Timeline History books: “The First Tomato to Know Everything,” from @sosallyo.bsky.social in @cabinetmagazine.bsky.social.
###
As we hold onto the human, we might (in preparation for tomorrow) remind ourselves that it was on this date in 1914 that President Woodrow Wilson issued a proclamation declaring the first national Mother’s Day. The previous day, May 8, Congress had designated the 2nd Sunday in May as Mother’s Day and had requested the proclamation.
You must be logged in to post a comment.