Published on September 30, 2024 9:13 PM GMT
TL;DR: AI will soon reverse a big economic trend.
Epistemic status: This post is likely more speculative than most of myposts. I'm writing this to clarify some vague guesses. Please assumethat most claims here are low-confidence forecasts.
There has been an important trend over the past century or so for humancapital to increase in value relative to other economically importantassets.
Context
Perplexity.ai says:
A 2016 economicanalysisby Korn Ferry found that:
Human capital represents a potential value of $1.2 quadrillion tothe global economy.This is 2.33 times more than the value of physical capital, whichwas estimated at $521 trillion.For every $1 invested in human capital, $11.39 is added to GDP.
I don't take those specific numbers very seriously, but the basicpattern is real.
Technological advances have reduced the costs of finding naturalresources and turning them into physical capital.
Much of the progress of the past couple of centuries has been due toautomation of many tasks, making things such as food, clothing,computers, etc. cheaper than pre-industrial people could imagine. Butthe production of new human minds has not at all been automated in asimilar fashion, so human minds remain scarce and valuable.
This has been reflected in the price to book value ratio of stocks. Ahalf century ago, it was common for the S&P 500 to trade at less than 2times book value. Today that ratio is close to 5. That's not an idealmeasure of the increasing importance of human capital - drug patentsalso play a role, as do network effects, proprietary data advantages,and various other sources of monopolistic power.
AI-related Reversal
AI is now reaching the point where I can see this trend reversing, mostlikely by the end of the current decade. AI cognition is substitutingfor human cognition at a rapidly increasing pace.
This post will focus on the coming time period when AI is better thanhumans at a majority of tasks, but is still subhuman at a moderatefraction of tasks. I'm guessing that's around 2030 or 2035.
Maybe this analysis will end up only applying to a brief period betweenwhen AI starts to have measurable macroeconomic impacts and when itbecomes superintelligent.
Macroeconomic Implications
Much has been written about the effects of AI on employment. I don'thave much that's new to say about that, so I'll just make a fewpredictions that summarize my expectations:
- For the next 5 years or so, AI will mostly be a complement to labor(i.e. a tool-like assistant) that makes humans more productive.Sometime in the 2030s, AI will become more of a substitute for humanlabor, causing an important decline in employment.Unemployment will be handled at least as well as the COVID-inducedunemployment was handled (sigh). I can hope that AI will enablebetter governance than that of 2020, but I don't want to bet onwhen AI will improve governance.
The limited supply of human capital has been a leading constraint oneconomic growth.
As that becomes unimportant, growth will accelerate to whatever limitsare imposed by other constraints. Physical capital is likely to be thelargest remaining constraint for a significant time.
That suggests a fairly rapid acceleration in economic growth. To10%/year or 100%/year? I only have a crude range of guesses.
Interest rates should rise by at least as much as economic growth ratesincrease, since the new economic growth rate will mostly reflect the newmarginal productivity of capital.
Real interest rates got unusually low in the past couple of decades,partly because the availability of useful ways to invest wealth waslimited by shortages of human capital. I'll guess that reversing thateffect will have some upward effect on rates, beyond the increase in themarginal productivity of capital.
AI Software Companies
Over the past year or so we've seen some moderately surprising evidencethat there's little in they way of "secret sauce" keeping the leadingAI labs ahead of their competition. Success at making better AIs seemsto be coming mainly from throwing more compute into training them, andfrom lots of minor improvements ("unhobblings") that competitors aremostly able to replicate.
I expect that to be even more true as AI increasingly takes over thesoftware part of AI advances. I expect that leading companies willmaintain a modest lead in software development, as they'll be a fewmonths ahead in applying the best AI software to the process ofdeveloping better AI software.
This suggests that they won't be able to charge a lot for typical usesof AI. The average chatbot user will not pay much more than they'recurrently paying ???
There will still be some uses for which having the latest AI software isworth a good deal. Hedge funds will sometimes be willing to pay a largepremium for having software that's frequently updated to maintain a2(?) point IQ lead over their competitors. A moderate fraction of othercompanies will have pressures of that general type.
These effects can add up to $100+ billion dollar profits forsoftware-only companies such as Anthropic and OpenAI, while stillremaining a small (and diminishing?) fraction of the total money to bemade off of AI.
Does that justify the trillions of dollars of investment that some arepredictinginto those companies? If they remain as software-only companies, Iexpect the median-case returns on those investments will be mediocre.
There are two ways that such investment could still be sensible. Thefirst is that they become partly hardware companies. E.g. they developexpertise at building and/or running datacenters.
The second is that my analysis is wrong, and they get enoughmonopolistic power over the software that they end up controlling alarge fraction of the world's wealth. A 10% chance of this result seemslike a plausible reason for investing in their stock today.
I occasionally see rumors of how I might be able to invest in Anthropic.I haven't been eager to evaluate those rumors, due to my doubts that AIlabs will capture much of the profits that will be made from AI. Iexpect to continue focusing my investments on hardware-orientedcompanies that are likely to benefit from AI.
Other Leading Software Companies
There are a bunch of software companies such as Oracle, Intuit, andAdobe that make lots of money due to some combination of their softwarebeing hard to replicate, and it being hard to verify that their softwarehas been replicated. I expect these industries to become morecompetitive, as AI makes replication and verification easier. Some oftheir functions will be directly taken over by AI, so some aspects ofthose companies will become obsolete in roughly the way that computersmade typewriters obsolete.
There's an important sense in which Nvidia is a softwarecompany.At least that's where its enormous profit margins come from. Thosemargins are likely to drop dramatically over the coming decade asAI-assisted competitors find ways to replicate Nvidia's results. A muchlarger fraction of chip costs will go to companies such as TSMC thatfabricate the chips. [I'm not advising you to sell Nvidia or buy TSMC;Nvidia will continue to be a valuable company, and TSMC is risky due tomilitary concerns. I recommend a diversified portfolio of semiconductorstocks.]
Waymo is an example of a company where software will retain value for asignificant time. The cost of demonstrating safety to consumers andregulators will constrain competition in that are for quite a while,although eventually I expect the cost of such demonstrations to becomesmall enough to enable significant competition.
Highly Profitable Companies
I expect an increasing share of profits and economic activity to comefrom industries that are capital-intensive. Leading examples arehardware companies that build things such as robots, semiconductors, anddatacenters, and energy companies (primarily those related toelectricity). Examples include ASML, Samsung, SCI Engineered Materials,Applied Digital, TSS Inc, Dell, Canadian Solar, and AES Corp (sorry, Idon't have a robotics company that qualifies as a good example; notethat these examples are biased by where I've invested).
Raw materiels companies, such as mines, are likely to at least maintaintheir (currently small) share of the economy.
Universities
The importance of universities will decline, by more than I'd predictif their main problems were merely being partly captured by a badideology.
Universities' prestige and income derive from some combination of thesethree main functions: credentialing students, creating knowledge, andvalidating knowledge.
AI's will compete with universities for at least the latter twofunctions.
The demand for credentialed students will decline as human labor becomesless important.
Conclusion
We are likely to soon see the end to a long-term trend of human capitalbecoming an increasing fraction of stock market capitalization. That hasimportant implications for investment and career plans.
Discuss
