Microsoft's "Sparks of Artificial General Intelligence" Study: Some Reflections

Since the release of GPT-4 a scarce month ago we have seen an abundance of comment on the capabilities of the system--overwhelmingly superficial comment, with little critical thought in evidence, such that a discerning observer may read a great deal of it without having much basis for judging its actual significance. (We are told over and over and over again that GPT-4 did well on a Bar Exam. That sounds impressive. But what does that really mean?)

However, a new study from a team of scientists at Microsoft, "Sparks of General Intelligence: Early Experiments With GPT-4," offers something more substantial. Working from a definition of intelligence laid out in a noted 1994 editorial produced by a large (52 member!) group of psychologists ("a very general mental capability that" implicitly "encompasses a broad range of cognitive skills and abilities" that includes "among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly and learn from experience"); a standard for what constitutes "Artificial General Intelligence" (as an artificial intelligence able to perform in the aforementioned ways at a level at least comparable to that of a human); and an understanding of how GPT-4 functions (as a neural network-based "Large Language Model" (LLM) trained on a vast body of web-based text "using at its core a self-supervised objective of predicting the next word in a partial sentence"); they devised a series of challenges which would be "novel and difficult" for such a program. In doing so they endeavored to challenge GPT-4 not only with a variety of demanding problems (coping with imagery, producing microcode, solving mathematical problems, etc.), but also highly idiosyncratic challenges that would require the chatbot to synthesize knowledge and skills from different areas to cope with problems for which its training was very unlikely to prepare it, demonstrating that more than memorization was involved, and that instead it possessed a "deep and flexible understanding of concepts, skills and domains." An excellent example is one test requiring it to present the proof of the infinitude of primes (aka "Euclid's theorem") written in the form of Shakespearean poetry, which forced the system to combine "mathematical reasoning, poetic expression and natural language generation." Still another was its being required to solve a riddle that required the system to come up with a conclusion based on the geographical location of the event described purely from clues that, it might be said, could only be handled on the basis of the "common sense" that has been such a challenge for AI developers. (Specifically it was expected to guess the color of a bear a hunter shot in a location where walking one mile south and one mile east walking one mile north brought him back to where he started--which can only be the North Pole, where from any point one mile south a mile back north brings one back to where they were originally.)

All of this established the Microsoft study then proceeds to detail the experiments and their results, and then present more general conclusions drawn from them. Ultimately the authors judged that "in all of these tasks, GPT-4's performance is strikingly close to human-level performance"--it even met the common sense challenge--such that they "believe that it could reasonably be viewed as an early (yet still incomplete) version of an artificial general intelligence (AGI) system" (emphasis added).

Because italicizing the relevant text doesn't seem to do it justice I am going to repeat it: "reasonably . . . viewed . . . as an early . . . version of an artificial general intelligence."

In other words, it is reasonable to say that not only is artificial intelligence here, but AGI is here. Now. As an actuality that millions of people are tinkering with.

Still, in looking at that statement one should not forget the qualifications. It may be reasonable to say GPT-4 is an AGI . . . but an early" and "still incomplete" version of it, and the team devotes as much time to explaining the limits of the AGI at hand as wowing us with its genuinely impressive capabilities. Key to their analysis is the distinction that psychologist Daniel Kahneman drew between "fast thinking" and "slow thinking." The former is automatic and intuitive, the latter controlled and rational--the kind where we have to consciously reason our way to a solution of a problem, a process which is, as implied by the terminology, slower, and more "effortful," but also likely to be more accurate and reliable.

As one might guess from the description of its functioning the word-predicting chatbot is a fast thinker which "essentially" may be said to "come up with an answer in . . . a single pass of the feedforward architecture." This works well with some problems, but not others, particularly those tasks which are "discontinuous" and so "require planning ahead . . . or a 'Eureka idea'"--and when faced with such tasks GPT-4's performance suffers, with the authors as one example of this another problem related to prime numbers. While GPT-4 did surprisingly well at explaining Euclid's theorem by way of Shakespearean poetry (inventively writing that explanation out as a dialogue between Romeo and Juliet) it did not do well at what would seem the much humbler task of simply giving us a list of the prime numbers between 150 and 250, with the authors arguing that this was because this is the kind of "slow thinking" task where most of us would "get out a scratchpad" and work it out--whereas GPT-4 has no such function, or even the basis for one. As this implies GPT-4 is simply not equipped to assess the quality of its own information or thought process, and not very good with context--deficiencies reflective of other lacks, like the absence of long-term memory, and faculties for making use of that memory through "continual learning." Together they leave it very sensitive to the form as well as the content of inputs--"the framing or wording of prompts and their sequencing in a session" easily throwing it even where tasks it can perform well are concerned. The authors of the study also stress that GPT-4 lacks the "confidence calibration" that would let it distinguish between when it is "guessing" and when it actually "knows"; and as many have remarked, prone to "hallucinating" false information. The result is that for its designers to improve this "early" and "still incomplete" AGI into something more complete requires that they add a capacity for "slow thinking" as well as "fast thinking," with the former overseeing the latter; long-term memory; a continual learning capacity to take full advantage of that faculty; and they add a shift beyond single word prediction toward "higher-level parts of [a] text such as sentences, paragraphs or ideas" (which they recognize may or may not emerge from a "next-word-prediction paradigm," even with these improvements).

Even as someone long inclining toward skepticism regarding the significance of the performance of systems like GPT-4 (I have tended to focus on how AI copes with the physical world, an area where the record has been rather disappointing) I have to admit that the case made in the study impressed me, rather more than anything I have heard or read about OpenAI's products since these started grabbing headlines last year. Indeed, I find myself thinking that here, after a great many false starts, AGI may be finally shifting from science fiction to science fact, with some of the capabilities that were supposedly "just around the corner" for decades finally arriving. (Indeed, in this month marking the forty-first anniversary of the Fifth Generation Computer Systems Initiative, it seems that we may be finally getting what was promised then, with important economic and cultural implications perhaps not too far off.)

However, I am also impressed by the constructive criticism offered herein, which seems to me well worth thinking about, particularly in regard for the necessity of a slow thinking function overseeing the more "autocomplete"-like function being demonstrated (the more in as, I suspect, that shift from single-word prediction to dealing with higher parts of text would likely depend heavily on it). If we accept GPT-4 as increasingly human-like when it comes to fast thinking, just where does AI research today stand with regard to the slow kind? And for that matter, where does it stand in regard to the integration of the two in a useful manner? Granting the case made here this would seem the next issue to think about--with the state and rate of progress determining whether the advance of AI research that has seemed so blistering in recent months accelerates, slows or even comes to another screeching halt of the kind that has renewed cynicism toward the field time and time again.

The Russian Armed Forces' Robotization Aspirations

Given both the escalation of interstate conflict in our time, in which tension between Russia and the West, and the Russo-Ukrainian conflict in particular, have been central; and the upsurge of interest in artificial intelligence research and its applications in the military as other spheres; the March 2023 report by the RAND Corporation's Krystyna Marcinek and Eugeniu Han regarding the Russian government's aspirations to the robotization of its military forces (Russia's Asymmetric Response to 21st Century Strategic Competition: Robotization of the Armed Forces) would seem especially timely.

The document, which runs to 132 pages all told, details the concept of robotization not only as understood generally but as it appears to be understood specifically among Russian military thinkers and policymakers; the Russian government's stated plans in this direction; the actual work done to achieve robotization of the Russian armed forces; and Russia's technological-industrial potential to achieve its goals.

As might be expected the mass of organizational, technical and even linguistic detail is considerable (the last seeing the authors decipher particular usages in the Russian literature for the benefit of Western readers who, even if among the very few possessing Russian language fluency, might not have a good grasp of the relevant terminology). Still, they manage to handle the material intelligibly, while in this moment when emotions are running high, and the old analytical baggage surrounding assessments of Russian industrial, technological and military performance can seem to weigh even more heavily than usual, the authors seem to me to manage to treat the subject intelligibly, and with some fair-mindedness.

The result is that they come to some interesting conclusions about differences in the thinking on military robotization as between Russia and the West. In particular, they argue that the Russian government is, if anything, more ambitious than the West with regard to robotization--that one can see this in their greater hope of replacing rather than augmenting human personnel with robots (hence, "robotization"); the signs they give of greater expectations of soon being able to deploy significant numbers of combat robots on the ground, like unmanned tanks; and their apparently greater willingness to countenance those systems using deadly force on an autonomous, no-human-in-the-loop basis. The authors also acknowledge the distance between the actual state-of-the-art in Russian, and everyone else's, robotics. And, as suggested by the title of their study itself (foregrounding "asymmetric response" while relegating mention of robotics to the subtitle), they stress the extent to which the government's ambitions reflect the difficulties of its situation--not least its demographic limitations relative to its rivals, and what its government may see as a geopolitical balance shifting disadvantageously in the near term. (Russia, after all, is a country of 150 million--as against a NATO alliance of some 900 million and growing, or a China of 1.4 billion--while even a Russian military devoted solely to homeland defense of the Russian Federation's territory would have that less than 2 percent of the world's population trying to hold a sixth of the world's landmass.)

The result is that the Russian government's high aspirations in the field of robotics can seem to come down to the hope of "wonder weapons" helping to salvage a worsening security situation--to an implausible degree. After all, overoptimism about the possible rate of progress aside, the fact remains that if Russia is not without some strengths here (among them a good STEM educational system, a relatively well-digitized economy and a genuinely serious government commitment here), it is far from clear how it would vault ahead of the rest of the world in this area given its not only being well behind other nations in even relatively easy areas like the less-demanding aerial drones to the point of relying heavily on foreign inputs for their production, numerous factors handicap it in any "race to AI superweaponry"--be it its having a much smaller industrial base than many other major states (its industrial disadvantage actually far exceeding its demographic disadvantage); its pointed weakness in many of the specific technological areas relevant to producing such weaponry (like the production of semiconductors, and their own required inputs); the difficulties for its activity posed by a sanctions regime increasingly incorporating the most important producers in those fields where it relies on imports (directed against not just Russia, but its important partner China); and the many enduring obstacles to its development broadly (like slight industry-academic collaboration, and the prospect of a "brain drain" as underemployed talent leaves the country for work abroad). Still, no one should be under any illusions about such problems being uniquely Russian, with, to all evidences, every major power currently and significantly a captive of technological hype as it copes with a world situation that is not all that any of them would like it to be.

Reporting the G-7 Economies

Americans are commonly stereotyped less interested in the outside world than other peoples--a function, some suppose, of the country's position in world affairs (its dominant status, which has others paying more attention to it than vice-versa) and more distinct features of its geography and culture (its continental scale, its being part of an already long predominant Anglosphere, its tendency to receive rather than send out immigrants, etc.).

Whether one sees the stereotype as containing any truth or not, the country's media seem to behave as if it were the case, adding a lessened attention to events in much of the world to its other biases--its preference for politics over policy, personality over material facts, narrative over nuance, and a centrist perspective squeamish about or hostile to contextualization and big-picture thinking, inclined to neoliberalism and neoconservatism, and deeply deferential to Establishment "expertise."

All this means that what coverage other countries get is of a very particular kind, as where the economic life of the Group of Seven advanced industrialized countries is concerned. It by and large approves Britain's being neoliberal trailblazer and "financial superpower" (and is much less interested in Britain's deindustrialization). It is less warm in its attitude toward Germany and Japan, but those two countries' economies are so big, powerful and dominant in their regions that the media are compelled to give them some heed--and if hewing to narratives of Eurosclerosis and Japanese "lost decades" for a long time, in the case of Germany especially, but also sometimes Japan, acknowledging manufacturing successes. (It also helps that playing up those countries' economic weight has been prominent in the American media's tendency to encourage those countries' "rearmament"--while given the campaign against China's IT sector there is no getting away from Japan's colossal profile in areas like chip-making inputs and robotics.)

By contrast there seems to me much less coverage of the other three G-7 members--France, Italy, Canada. Where the first two countries are concerned, and France particularly, it revels in neoliberal clichés about an "old Europe" of bloated governments, overgenerous welfare states, uppity, strike-happy workers, ever-low and eroding productivity and "competitiveness," and an oppressed entrepreneur class looking longingly across the English Channel, and still better across the Atlantic, toward FREEDOM! (Indeed, such cliché, which is to be found in France's own press as well as that of foreign countries, has reared its ugly head in American coverage of French protest against the raising of the country's retirement age--cliché consistently acknowledged even by those who would offer a more sympathetic take.) Italy is hazily perceived in a similar manner (the tendency to treat a whole continent as a nearly homogenous blob may not be as bad in Europe's case as in that of "Africa" or "Asia," but not as much better as one might think). Still, it is mostly ignored. And Canada is ignored even more completely--such that one has the irony of America neighboring one of the world's largest and most important economies, and its press paying almost no attention to the fact at all, such that I suspect that even Americans attentive to the relevant areas of the news know less about the economy of their own northern neighbor and North American Free Trade Agreement partner than they do any of the other countries.

Such is the "quality" of the media that centrists endlessly sing as our salvation from fake news-purveying hordes.

On Italy's GPT-4 "Ban"

The Italian Data Protection Authority (Garante Per La Protezione Dei Dati Personali, or GPDP) has imposed an (I quote the official English translation of its statement on its web site) "immediate temporary limitation" on GPT-4.

According to the Authority's own statement (which you can see in the original, and in English translation, at its own web site, here) the action has nothing to do with the fears of AI-out-of-control of which so much is now being made. Rather the GPDP cites the more commonplace Internet regulation concerns of the protection of user privacy ("no information is provided to users and data subjects whose data are collected by Open AI; more importantly, there appears to be no legal basis underpinning the massive collection and processing of personal data in order to 'train' the algorithms on which the platform relies"), and the fear that children will be subjected to inappropriate content ("the lack of . . . age verification mechanism expos[ing] children to receiving responses that are absolutely inappropriate to their age and awareness").

One may speculate that these stated concerns do not exhaust the GPDP's concerns about the technology--and even that other concerns may actually be of higher priority than the ones stated. Still, that these are the ones presented is a reminder that, in what can seem a silly rush to see the release of GPT-4 as a bad sci-fi "rebellion of the robots" scenario, we may be overlooking humbler but quite important concerns--the more in as so much is being made of some of these exact concerns in relation to other technologies.

Notes on the 2023 British Defence Review, Thus Far

Government documents regarding official policy, contrary to what they should be, are not written for the intelligent citizen taking an interest in the public affairs with which every citizen ought to be concerned, but the "expert." One reflection of this is a lack of context for virtually everything important, which, especially in combination with the pompous verbosity that is a trademark of such documents' prose, leaves many a reader who takes the trouble to plow through it with a bland, platitudinous text offering little if any real insight into the important issue at hand.

It seems to me that British defence reviews are getting worse rather than better in this respect, with 2023's review document exemplary--the more in as it is a "refresh" of the 2021 document, which itself represented a break with the past. Titled Global Britain in a Competitive Age: the Integrated Review of Security, Defence, Development and Foreign Policy, it combined the review of defence with a much more comprehensive (and much longer) policy statement--and while generally being referred to as the defence review, relegated most of the more narrowly military stuff we actually expect from British defence reviews on the basis of what they previously looked like (like how the armed forces will be sized and structured, what will be done about equipment holdings and bases, etc., etc.) to a separate document, Defence in a Competitive Age.

So basically the thing we call a defence review, even as it swelled to greater proportions, did not actually contain much "defence review stuff" in it, while commentators generally refrain from calling the one with most of the defence review stuff in it something other than a defence review, the much more generic term "command paper," as Walter Sobchak would have it, the "preferred nomenclature." But with, as it happened, the critical stuff about Britain's government's plans to expand its nuclear arsenal mainly in the Integrated Review rather than the more defence review-type command paper.

Entirely logical, of course.

The pattern is being followed this year, with the Integrated Review out first (it appeared March 13), and the command paper to follow. However, when the 2021 Integrated Review came out the "command paper" came out the next day, so at least those interested in the issue had the details of the country's defence posture soon enough. This year, though, while the Integrated Review, which was itself subject to delay (coming out March 13, almost a week later than its initial March 7 due date ), will be followed by the "command paper" not by a single day but by a period of months--this expected out in June!

This very long delay seems the more consequential given the reports going back to February of squabbling among the government's senior policymakers over the armed forces' funding--a reflection, I suppose, of how between the country's post-Great Recession, post-Brexit, post-pandemic, post-Trussonomics economic situation, the reality of NATO's military confrontation with Russia as war rages in Ukraine, and the government's pretensions (prioritizing the Euro-Atlantic while somehow sticking with the "tilt to the Indo-Pacific"), the incoherence of the government's aspirations is impossible for even the senior functionaries to ignore. While it is clear that much remains to be decided, given the disparity between means and stated ends I see no squaring of this circle anytime soon, not with the crowd currently in office, nor for that matter any other allowed anywhere near 10 Downing Street--the more in as this month's Integrated Review appallingly opened with three paragraphs of the government congratulating itself at a moment when serious self-criticism was sorely needed.

Notes on the Comeback of Conscription

As the reaction to a deepfake video of Joe Biden reviving the draft has reminded news-watchers, conscription is appearing on the agenda across much of the Western world. Sweden, scarcely after ending it, has already reinstituted it in a limited way (having a draft, but so far limited call-ups). Meanwhile France, where President Emmanuel Macron who had wanted to mandate "direct experience of military life" for the young for at least a month, now enrolls its young people in a "national service" program where they are "encouraged" to partake of a military component that can seem at least a quarter-step in that direction. Meanwhile Germany's Minister of Defense, while demanding ever more resources for a program of German rearmament already without any point of comparison since at least the founding of the Bundeswehr in the 1950s, openly speaks of the end of conscription as a "mistake."

In the U.S. there has been less discussion as yet, for ample reason, not least an insular geography that makes a manpower-intensive army relatively less important than a technology-intensive navy than is the case for a country with long land borders with big military powers; and a vast and relatively youthful population base that makes it easy to raise large land armies on even a volunteer base. (Compared with the biggest West European power, Germany, the U.S. has four times as big a population, and five times as many people in the critical "military age" category of 18-24--some 30 million to Germany's 6 million.) However, the most immediately important fact is that the U.S. armed forces were never reduced from their Cold War level of strength to anything like the same degree as the European nations. The U.S. cut its armed forces by a third after that conflict--while the cut was closer to two-thirds in the case of the Federal Republic of Germany (FRG), and much deeper where fighting forces were concerned. Where in the Cold War the FRG's army had a dozen active-duty divisions, including six armored divisions, today it has just three divisions--not a single one of which is a proper armored division on the old standard, because of how the country's "tank park" shrank, from some five thousand tanks to perhaps not even two hundred and fifty--a 95 percent contraction. It is likewise with the reserves whose mobilization would come ahead of any draft. Where the Bundeswehr's massive reserve once permitted its ground-forces component to by itself put over 1 million soldiers into the field when fully mobilized, now it would manage 70,000--a 93 percent drop. And so does it go, more or less, with France, and Britain, and Italy.

The result is that if really intent on in short order restoring the capability to fight a conventional conflict on any significant scale (even one well short of what they were prepared for in the Cold War) the European states have few alternatives to conscription, as compared with a U.S. which would not have to go anywhere near so far, so soon, still possessing as it does massive conventional war-fighting capabilities (even where ground forces are concerned, plausibly disposing of a greater mechanized fighting capability than Europe's powers bring to the table combined). However, whether other Western states actually will follow Sweden's example will have much to do with how the international scene evolves--and what the publics of those countries will accept. The media is characteristically eager for the most militarized response to the international situation of today--but it is far from clear that the broader public is as enthusiastic. Moreover, the attractiveness of the idea falls in proportion to one's expectation that they will be the one actually drafted--"Back in my day"-growling old people who don't much like young people, and know they themselves are safe from call-up, very readily support the idea--but young people not so much. (After all, if they wanted to be in the armed forces they would likely already be there.) Indeed, forcing large numbers of young people into military service could be exactly what catalyzes the relatively marginal opposition to the conflict seen to date into something much more difficult for supporters of the very conflict for which they drafted them.

Australia and the AUKUS Submarine Agreement

Ever since the announcement of the AUKUS partnership and the Australian nuclear submarine program for which it would allow there has been speculation about just how Australia would actually be supplied with such submarines--or at least, what at least the initial plan for such supply would be. In the wake of a joint statement by U.S. President Joseph Biden, British Prime Minister Rishi Sunak, and Australia's Prime Minister Anthony Albanese at Naval Base Point Loma in San Diego the period of speculation is over.

Simply going by the official transcript of the statement by the three figures--if also distilling these speakers' all too characteristically rambling, repetitive and public relations-laden rhetoric into hard fact--one gets the following:

Australia will buy three Virginia-class nuclear-powered (but conventionally-armed) boats starting in the early 2030s, with an option on two more. These acquisitions will then be followed by a British-designed next generation of likewise nuclear-powered but conventionally-armed boats incorporating British, Australian and U.S. technology that both Britain and Australia will use, and which will "share components and parts with the U.S. Navy," to the point of communicating "with the same equipment"--the "AUKUS SSN." Australia will build its copies of those AUKUS boats domestically, and sustain them domestically. They will also be commanded by the Australian Navy as a "sovereign capability"--apart from Australia's reliance on imported nuclear fuel, as it will not produce its own (that capacity, of course, necessarily entailing the capacity to produce fissile material, which would be a breach of its non-proliferation commitments). The first of those boats will be delivered in 2042, with new deliveries every three years on, and an ultimate goal of eight boats.

In the meantime Australian personnel have been undergoing "nuclear power training" in the U.S., while "[b]eginning this year, Australian personnel will embed with U.S. and UK crews on boats and at bases in our schools and our shipyards," and U.S., and British, nuclear submarines will make more port visits to Australia, on the way to the establishment of "a rotational presence of U.S. and UK nuclear-powered subs in Australia to help develop the work force Australia is going to need to build and maintain its fleet" later in the decade.

Rounding out the vision of tri-national cooperation there is also the expectation that, Britain and Australia's sovereign but identically equipped and closely collaborating fleets ("[o]ur submarine crews will train together, patrol together . . . maintain their boats together"), and the similarly linked U.S. fleet, will go a long way to realizing that ideal not only of partnership, but efficiency-achieving homogeneity even as each retains control of its assets.

One may add to this that these figures, particularly Australian PM Albanese, also made much of the submarine program from an "industrial policy" standpoint, claiming that "[t]he scale, complexity, and economic significance of this investment is akin to the creation of the Australian automotive industry in the post-World War Two period."

Of course, what we are hearing reported in the press now is not limited to what was presented in the official statement. There are, of course, questions about just how "sovereign" the Australian boats really will be. (As things stand, Australia's nuclear industry is virtually nonexistent--the capacity to build and service such boats as yet nonexistent, and its actual establishment yet to be seen.) There is also the matter of the immensity of the resources invested. Were the cost of the program over the longer haul is concerned we are hearing figures in the $250 billion (U.S. dollars) range for the next three decades. This sum is about equal to 15 percent of Australian GDP today (circa $1.6 trillion), and equal to some $4 trillion in the case of an economy the size of that of the U.S.--exceeding any comparable expenditure the U.S. has made--and even if spread out over three decades, with the tendency of price tags like these to rise set aside, and generous allowance for the possibility of economic growth aided by the "stimulus" of the program, there is no getting away from the sheer size of the bill relative to the country's resources, with that this implies. (Even those who are supportive of such high military expenditure may, in fact, wonder if this is really the best way of getting "value for the money.") And of course there is the geopolitical rationale for the program. The sole reference in the entire transcript to China was mention of the U.S. having "safeguarded stability in the Indo-Pacific for decades to the enormous benefits of nations throughout the region, from ASEAN to Pacific Islanders to the People’s Republic of China." Yet there would be no discussion of such a project at all were it not for the concern among the country's policymakers, and the governments of its allies, about the possibility of conflict with China --and not all are sanguine about a militarized course. (Indeed, many in Australia, and elsewhere, are convinced that confrontation is the wrong path--however much news outlets like the Sydney Morning Herald, with its "Red Alert" series, insist otherwise.)

The result is that if the joint statement answered some questions a great many others remain to be decided--the more in as a very great deal can happen over the very long time frame assumed in this program, which, even if going entirely as planned and scheduled, will not deliver the first Australian-built AUKUS sub to the country's navy for two decades.

After all, just how much resemblance does the world of 2023 bear to the expectations of two decades prior?

The Upcoming British Defence Review

The British government was supposed to have a defense review out on March 7--one the more eagerly anticipated because of the fact that, as its appearing a mere two years after the last rather dramatic review indicates (in the post-Cold War era, recalling the 1994 Defence Costs Study after the monetary-fiscal crisis of the early '90s, the "New Chapter" to the 1998 Strategic Defence Review after the September 11 attacks), it was expected to address a painful shock, or rather several of them. The protraction of the pandemic, Britain's especially weak economic condition afterward, the worsening of matters by "Trussonomics"--and of course, the war in Ukraine, amid a generally deteriorating international scene--make it easy to picture changes coming.

The review, of course, does not seem to have appeared as of the time of this writing (March 9). On March 7, the day it was supposed to actually be out, we were told that it "will be published in the coming days" (around the time of the Spring Budget 2023, which is scheduled for March 15)--against a backdrop of speculation that dispute within the government over policy and its resourcing was likely to compel such a delay--the Ministry of Defence apparently objecting to being asked to do more without a greater infusion of cash.

We will have to wait until then to see how the government will try to square the circle of its ends and means--or rather, pretend to do so. Since the nineteenth century Britain has been consistently, severely, overstretched militarily--the long-range big-picture history one retrenchment after another. Especially since 1945 the retrenchments have come thick and fast--in part because the cuts to commitments always significantly lagged the cuts to forces, which in turn lagged the cuts to the armed forces' budget, the reduction of which was never in line with the reduction in what Britain could afford. The result was that the situation quickly compelled a new round of adjustments--generally, cuts--of all of the above, each inadequate in the same way as before, Britain requiring its armed forces to do more than was reasonable, while being penurious with the forces it did have, and even that penurious spending more than the British fiscal state or the British economy on which it stood could bear (even, one might add, as British budget-cutters inflicted increasing pain on their populations with ever-greater stringency in services and social programs, down through the death rate-elevating 2010-2019 round of austerity), in a vicious circle that just went on and on.

There is not much left to cut these days, with British forces, which have shrunk almost every year since the Korean War, now under 150,000 strong, with an army that cannot boast a single proper heavy division (as against the three armored divisions it had in Germany during the Cold War), and a fleet stretched to the limit merely providing escort to its handful of carriers and nuclear ballistic missile submarines.

How could this be when Britain has been, and remains, one of the world's biggest spenders? The story is a complex one, but a key part has been British governments' hanging on to certain superpower-type trappings--a nuclear deterrent, a global base network (extending from Alberta to Kenya, from Cyprus to Oman, from the Falklands to Singapore), a relatively robust defense-industrial base that has the country building its own tanks and warships and fancying itself building sixth-generation super-fighters. Surprisingly recent decades have seen the government bring back things it had admitted that it could not afford back in the 1960s--not least the aforementioned big carriers (the seeds of which lie in the 1998 review), and an "east of Suez" orientation (emergent with the "Global Britain" rhetoric and alarums about China, the essential thrust of the "tilt toward the Indo-Pacific" of the 2021 review). Meanwhile, ever-extending the gap, Larry Elliott's quips about a Britain with a "fantasy island" economy sinking toward Third World status looks less and less like hyperbole all the time. (Crunching the numbers recently it seemed to me that the first industrialized nation is approaching the condition of a developing nation, certainly if the manufacturing-intensiveness of the economy is the standard.)

The more critical, quite fairly, speak of post-imperial "grandeur" prevailing over the government's realistically living "within its means," all as the same government wrecks the economic foundation providing those means for decade after decade. Still, severe as the situation has become, I do not think retrenchment or rationalization are to be expected from the coming review--just as I have no confidence that the current British government, nor one headed by its Parliamentary opposition, is likely to shift the country's economic course to any appreciable degree. Rather I expect that the new policy will be grandiosely described, materially overambitious, inadequately resourced, and (the more so for combination with that dismal economic situation) necessitate overhaul in another very short period of time--likely, before 2028. Because in our time the pattern of Establishment decision-making, in all things and everywhere, is "If it's broke, don't fix it."

Kristin Tate, Artificial Intelligence and the Decline of the White Collar: Thoughts

Recently suggesting that artificial intelligence might displace "white collar" workers before it displaced "blue collar" workers performing more mobility and dexterity-demanding manual labor I had the impression that I was unlikely to see anyone arguing along similar lines in a mainstream forum anytime soon. Naturally it was a surprise to see Kristin Tate making a version of the case in The Hill.

Considering Ms. Tate's argument I find her very bullish on the prospect--more than I think is warranted. I have simply seen artificial intelligence oversold far too highly for far too long to be anything but guarded toward the more recent claims for "the coming artificial intelligence economic revolution." (Remember Ray Kurzweil back at the turn of the century? Or how profoundly the hacks in the press misunderstood the Frey-Osborne study from 2013, creating the material for panic out of their highly qualified, rather uncertain conclusions?) I also think Ms. Tate overlooks the important fact that (as Frey and Osborne noted, not that anyone was paying attention to what they actually said) its being technically feasible to do something does not make it automatically worthwhile to do that thing--and so underestimates the economic and cultural and political obstacles to actually using these technologies to substitute for human labor, even were the performance of the technology perfectly adequate from a purely technical standpoint. Ms. Tate, for example, acknowledges that humans might be preferred in face-to-face jobs, and counts a certain amount of legal work among them--but overlooks the extent to which professionals, especially those belonging to the more powerful professions, can be expected to resist their replacement by AI (as was demonstrated in their reaction to the plans to have the first "robot lawyer" plead a "client's" case in a California court back in February).

I might also add that the hint of gloating Schadenfreude toward the college-educated "elite" and celebration of "blue collar values," especially to one alert to her politics, makes it easy to picture her reveling in the anticipated ruin of a group of people she despises--or perhaps simply "trolling" them with the thought of their ruin--coming in ahead of critically-minded caution.

Still, the essential argument is not just well worth thinking about but, if we start actually seeing chatbots and other technologies really make their mark, likely to become more commonplace--with the same going for the consideration of their implications.

Revisiting Chris Hedges' Death of the Liberal Class

Back when it came out I read and reviewed Chris Hedges' Death of the Liberal Class.

More recently I have had occasion to revisit its argument.

As I observed in the initial review, despite the book's title Hedges' focus is so much on liberals' actions, as against the larger, often highly inimical political scene in which they had to operate, that he tells only part of the story, with that part perhaps more fittingly titled "The Suicide of the Liberal Class." Still, in spite of that close attention to their conduct--and his getting a good deal right, like the way in which liberals consistently ganged up with the right to beat up on the left, and the way in which a condition of permanent war has a corrosive effect on liberalism (with Cold War anti-Communism combining the two)--there were significant limitations to his analysis. In particular his address of the relevant mechanisms seemed inadequate--I would now say, because of the weaknesses of his understanding of liberalism itself.

Consider that "liberalism" as presented in the work of mid-century theoreticians such as Arthur Schlesinger Jr. and Daniel Bell, with a key part of their story their response to the horrors of the world war era--fascism, the Holocaust, etc.. There was a leftist reading of all this--that the whole old order of capitalism, nation-states, and the rest, was bankrupt and irredeemable and something better had to be built. However, there was also a conservative reading of all this--namely that it was trying to be build something better that was the cause of the problem. Schlesinger and Bell (among many, many others) inclined to the latter view, with all its implications not only for the socialism that had once attracted them, but plain and simple liberalism. Liberalism, after all, is founded on the efficacy of reason as a way of understanding and directing social life, and accordingly the importance of social criticism and the possibility of progress. However, these "centrists" instead despaired of reason, and uncomfortable with the criticism founded on it, because they were pessimistic about progress--with this going so far as a quasi-theological belief in the existence of "evil," and disdain for the mass as a "swinish multitude." Indeed, that particular liberal form of government, democracy, became something which they were uncomfortable, such that they increasingly endeavored to limit it--insisting on a political discourse stripped of ideology and values and anything else that might call into question the system, or arouse discontented masses, from which has emerged that form of discourse we now speak of as "pluralist" and civil. They also shifted away from that other liberal standard, universalism, toward particularism, with (certainly as seen in the historiography of Daniel Boorstin) the U.S. a country uniquely gifted with a tradition of such discourse in contrast with the ideology-addled Old World.

All of these views--the attitude toward reason, social criticism and progress; the belief in evil and suspicion of masses; the suspicion of democracy, to the point of their being eager to bound it; the stress on difference over universality--together comprise the classical conservative package to such a degree that centrism must be recognized as an adaptation of classical conservatism to the circumstances of mid-twentieth century America. Given its influence it is far to say that American "liberalism" traded their old liberal philosophical foundations for conservatives ones that increasingly came to define American liberalism (as, to his credit, that other mid-century "liberal" intellectual Richard Hofstadter acknowledged in the 1950s).

Of course, that said these liberals still had their differences with the Taft-MacArthur-Goldwater right, but that did not diminish the fact of their conservatism, rather leaving them emphasizing a different side of the conservative tradition. Theirs was the conservatism which "pruned the tree"--which was prepared to compromise and make adaptations to defend what it deemed essential, as against the more uncompromising, or even reactionary, conservatism that Hofstadter, was to call "pseudo-conservatism." And of course, as the Cold War ran its course, the stalling of the post-war boom compelled a reconsideration of American social arrangements, and the Communist menace that had been the main reason to compromise disappeared, conservative centrists saw less reason to do so--and moved rightward accordingly.

Accordingly centrism/liberalism's aligning itself with the right against the left was not a matter of the corruptibility of liberals, but their most natural and predictable course of action, especially in the circumstances liberals went on facing over the twentieth century, and into the twenty-first.