bunch of relevant links on social science/gradual disempowerment, research ideas, objections, etc.
https://x.com/LRudL_/status/1929067292245070210
After replacing jr employees, manager time will be the bottleneck, BC you need high context managers/approvers/directors to guide the work.
The upwards moving pyramid model doesn’t seem to factor in demand curves for the firm’s outputs. Depending on the shape of those curves, firms might instead massively increase the outputs of bottom-level employees, these might be directly productized or passed up to managers for further filtering and processing. Intra vs extra-firm demand curves is an interesting idea
How will/can unions of lower level white collar employees co-ordinate to stop the pyramid sinking? (Interestingness waterline link). Never writing stuff down, keeping context scarce will help for a while. More senior management might see the writing on the wall and form cartels, seems hard though since that’s more up to c-suite.
Korinek and Suh model wages as a function of fraction of automatable tasks. What else could you use? What other economic analyses are good? See also AGI could drive wages below subsistence level | Epoch AI https://epoch.ai/gradient-updates/agi-could-drive-wages-below-subsistence-level
Imagine that labor-replacing AI is here. There’s a massive scramble between countries and companies to make the best use of AI. This is all capital-intensive, so everyone needs to woo holders of capital. The top AI companies wield power on the level of states. The redistribution of wealth is unlikely to end up on top of the political agenda.2
2: 2 An exception might be if some new political movement or ideology gets a lot of support quickly, and is somehow boosted by some unprecedented effect of AI (such as: no one has jobs anymore so they can spend all their time on politics, or there’s some new AI-powered coordination mechanism).
- what are the coordination technologies that disproportionately favour labour organising? Which would be robust to better surveillance tech?
Even if individual consumers would pay for teachers. Govt or other more powerful actors lack incentives to invest in schools. So one decent thing to do would be build markets or other ways of provisioning goods that increase/invest in human capital or power than just need aggregate or individual consumer demand not an interceding state or corp. Funny how this stuff keeps rhyming with ancap crypto bo priorities lol.
- eh but remember the demand base would shift away from people so maybe this doesn’t matter
How does all this intersect with multi agent stuff Agent infrastructure
There are two main ways states completely break the resource curse: effective governance to redistribute resources, and economic diversification9 to create incentives for states to care about their people.
Resource curse in africa
More resource rich countries are worse economically because theyre incentivised to get extractive dictators and coups.
Its like extractive capacity (resources x extractive efficiency) is disruptive, and institutions need to be strong enough to capture it. In effect they must evolve in sync, as they do at the technological frontier.
The analogy with AI should make you more worried, as extractive capacity accelerates faster than institutions. Its like everyone with ai tech has the option of being an african dictator. The Alignment Problem and AI safety notes