In Martin Wolf’s The Crisis of Democratic Capitalism, he observes how stagnant productivity creates a zero-sum dynamic in the economy. The book was published in early 2023 and the evidence would suggest the problem has only got worse since then:
In a country with fast increases in productivity everybody will get better off, unless inequality rises very quickly. But in A country with stagnant productivity, such as Italy over the last two decades or the UK over the last one and a half decades, the standard of living can rise for some only if the standard of living for others falls. This then becomes a zero sum economy. If A wins, B through C must lose.
This has been intensified by the post-Covid inflation shock. The headline figures are bad enough but the effect is segmented, hitting poorest householders and renters the hardest, with a tendency to underrepresent the cost of housing (albeit in a way that is methodologically complex and doesn’t permit easy generalisation). If we remain within this dynamic indefinitely then, at least in principle, we would see a continual immiseration of the vast majority of the population. This is partly inflationary (i.e. declining real wages for the majority of the population) and partly due to Piketty’s r > g dynamic (i.e. returns on capital grow faster than the economy as a whole). This means that those who already have wealth will tend to increase it faster while the majority of the population get poorer, not least of all through out competing the majority for assets like housing.
Unfortunately higher inflation is here to stay. The zero interest rate environment stabilised the economy after 2007/8 but at the cost of intensifying asset price inflation through the r > g dynamic. But that’s now broken down. The deflationary forces of the neoliberal era (access to global labour, cheap goods through international markets, the internet, debt financing of consumer spending etc) are exhausted or in reverse. So the forces and mechanisms keeping inflation down are gone, at exactly the point when there are renewed inflationary pressures: deglobalisation, climate change induced supply shocks and labour scarcity driven by ageing populations. My understanding is there’s significant uncertainty about how this will manifest sequentially but the climate crisis dimension seems inarguable to me.
I’ve recently taken to thinking of these two mechanisms as the enshittification of (political) economy: stagnant growth means that inflation and growing wealth inequality produce an economy which is ever more shit. The capacity to stabilise the system or invest decreases as the enshittification dynamic takes hold. It’s a vicious cycle and the more obvious it becomes, the less tenable neoliberal centrism becomes. This conjuncture can’t continue indefinitely. Something will break, sooner rather than later. The far-right represent one path forward, with an economic programme that ultimately amounts to mitigating the costs for a racialised in group while managing the social collapse through vicious attacks on the out group. The danger is that increasingly large sections of capital will come to endorse that, which I think is ultimately how we should read the difference between Trump 1 and Trump 2.
Ultimately something has to change about the political economy of this conjuncture. Increasingly I’m thinking ‘AI’ is positioned as a form of magical thinking for technocratic centrists who want to avoid this conclusion, functioning as the deus ex machina which could square the circle. If you get significant productivity growth going on again then you can keep the underlying political economy unchanged because the same mechanisms no longer operate in a zero-sum way: the rich can continue to get richer but the poorer can have their living standards sustained.
If you combine this investment in the status quo along with the ‘new cold war’ over AI in which it becomes a vector to manage relative American decline, you can begin to see the struggle over AI as the central question of political economy in the 2020s. This is the macro context in which we have to make sense of the micro and meso dimensions of LLMs being diffused. Obviously the shit is going to hit the fan at some point soon because the bubble cannot continue to growth indefinitely, at which point the underlying economic logic of the conjuncture will become a much more stark fault line of political economy i.e. do we try and reinflate the bubble or have the social fight which has been postponed for the last 20 years through a series of spatiotemporal fixes?
I feel slightly out of my depth with this terrain. The hill I will die on though is that there is some magic in LLMs even if it falls short of the magical thinking of the technocratic neoliberals. The problem is that the political economy organisations face under this conjuncture presses them to undercut exactly the conditions which are needed to realise the potential productivity gains: professional autonomy, freedom to experiment, respect for tacit knowledge and time to embed the experiments. The political economy which makes the productivity gains imperative simultaneously hinders the realisation of those gains.
Mark Carrigan 