The problem with that is that the attention of readers remains limited. There's a problem somewhat akin to Baumol's Cost Disease here... One aspect of the market grows with technology while another remains stubbornly limited.
Unless you intend to have your own AI assistant read the posts for you and summarize them. (I am reminded of the Electric Monk in Dirk Gently's Holistic Detective Agency. Just as a dishwasher does all that rote, boring dishwashing for you, the Electric Monk does all the rote, boring _believing_ for you, to free up time for more productive activities.)
We'll use our AI assistants to summarize and prioritize the flood of Noahpinion articles. Our assistants could even learn to elaborate on things it believes we’ll be interested in. If we only want 2 minutes of content one day, then it will be particularly aggressive in filtering and summarizing. One day someone might ask for 4 hours, in which case their AI will aggressively expand and elaborate.
And of our course our AI assistants will handle the drudgery of commenting on our behalf. We can simply prompt, “write 10 paragraphs with authoritative references detailing why Noah is not only incorrect, but also deeply immoral.” And his AI will auto-reply. It will also provide Noah a daily summary of user comments; commonly just reporting, “you still have many foolish readers, but they’re slowly learning through our reinforcement.”
How do you patch a copilot generated bug that gets reused millions of times in low level code throughout the world? That's going to be an industry in itself.
Yep. I am a senior software engineer with tons of experience and AI code autocomplete is extremely useful, but not necessarily due to the comparative advantage factor. It's a search engine.
Using generative code AI, I can answer the question, "How would someone else write this thing that I know how to write?" and get a coherent answer that I can evaluate for effectiveness and possibly incorporate into my own technique.
Similarly, instead of trying to figure out the API of a particular complex package or system, I can just ask the AI to get me started. It might get things wrong, but it'll dump the most commonly used calls, routes, etc. in my lap.
Also, it teaches conventions. If I'm not exactly sure what the canonical way to, say, write a route in a new web api framework I'm presented with, I can just ask co-pilot and get a ton of useful context. This can sometimes be better than the official documentation for systems that are poorly maintained.
I think this holds true basically everywhere. You can ask generative AI to explain something to you and get basically a distillation of the median internet understanding of that topic as a crash course.
Yep! I’m much more interested in AI searching to find an existing library or API rather than helping me write new code. Similarly, I’d appreciate it helping me understand what it found. It could leverage its familiarity with my knowledge and experience to quickly get me up to speed with the new library or API.
I love posts like this! But who gets the wealth of the added productivity from those tasks that are automated? The Governance and wealth distribution of A.I. at scale, either leads to Dystopia or Utopia, depending on how this is handled.
That's what depends! The balance of the displacement effect, the productivity effect, and the reinstatement effect will be what determines whether capital or labor benefits more from the new technology...
Great question! I've wondered how much this might fundamentally shift how much income goes to capital vs. labor, and even begs the question of how much wealth is it possible to accumulate?
Hoping to explore these questions in a future newsletter myself!
Like many other things today, this sounds like another way for the educated / laptop class (of which Noah is part -- nothing personal, man) to distance itself from everyone else. Even Noah's title betrays this bias. If you're a roofer or a plumber, there's no "autocomplete" for you.
Just like the industrial revolution, it alters the labor/capital balance in ways unfavorable to labor. After all, if capital can become labor (via AI and robotics), that's pretty detrimental for labor. It's noteworthy that the last time a shift like this occurred, we got Karl Marx.
The 1980's called; they want their economic policy back.
If AI increases productivity of the laptop class significantly, this will cause either that class or the owner class to reap the vast majority of those benefits. Will they spend more? Sure, but the effects of the resulting inequality dwarf the effects of any increase in downstream income.
> If AI increases productivity of the laptop class significantly, this will cause either that class or the owner class to reap the vast majority of those benefits.
Not necessarily. There could easily be widespread unemployment of white collar workers at least in the short term. Industries adjust but it takes time. If we decided to import 10x the number of white collar workers for existing jobs it would likely drive their wages down. A massive and sudden productivity boost could easily look like the same thing. The white collar workers who kept their jobs would likely benefit enormously but a much larger group would lose their employment entirely.
As households get wealthier they spend more on status-seeking, positional goods which are intrinsically zero-sum: houses in the right areas, fine art and jewellery, and so on. Not so much relatively speaking on services that promote the flow of money around the economy.
I agree with Brian: this seems like it will inevitably increase the wage gap between the managerial/professional class and ordinary workers: salespeople, restaurant staff, laborers, truck drivers and other goods distribution worker, nurses and other low-paid health workers, even teachers and assistants.
(That's something like 30% of the workforce just in that list according to the BLS. There are more: janitors, groundskeepers, repair workers, construction workers, security guards, ...)
AI-based research and image generation could potentially give roofers and plumbers easy access to information about a building's interior layout, and records of previous maintenance or repair work, which might be as helpful to them as medical history and x-rays are to a doctor.
The OP is about the near term. In the long term, the impact of technology will be to make financial wealth much less important. The real question is what "near" and "long" actually mean in terms of years.
Will it though? After the industrial revolution people of different wealth strata can all eat their fill, wear warm clothing, heat houses automatically, etc. but it doesn't seem quite accurate to say financial wealth is much less important. Possibly technological developments will smooth out some of the differences between wealth strata we see today, but keep introducing new ones.
That is a joint responsibility of employer and employee. Ultimately, the goal is to end up with "free time" that can be used productively. Both the employer and employer can be wise with it and greatly increase opportunity for both, or it can be squandered. Viewing it as a zero-sum game with the focus on diverting the benefits to only one party is disastrous in the long run.
The vast majority of engineers, finance sector, medical workers etc. today would have been shoveling manure, pulling a plow, and cutting firewood 200 years ago. The industrial revolution, especially on farms, has freed up a huge sector of humanity from those tasks to do other things. The people who were able to take advantage of the opportunity and were inclined to do so are immensely more wealthy today than they would have been in the cold and dark.
Anti-discrimination laws have been opening up opportunities for women and minorities since the 1950s. It is likely that Title IX and the reduction in Jim Crow segregation were cornerstones of the immense economic expansion in the US in the 1980s and 1990s as the beneficiaries entered the workforce. We continue to have opportunities to maximize our human capital. However, history is replete with examples of society-wide suppression of human capital due to bigotry and hate, so we must avoid that.
There's actually pretty good science that says almost all the income gains since the 1970's have come from women entering the workforce. Which is an illusion. A wife who stays home with the kids is economically invisible. A wife who works bumps GDP both with her earnings and her daycare consumption. It is this flaw in our GDP accounting that has allowed us to pretend most households' economic condition has been improving for the last 50 years.
Ernest has some good points-- a lot more people have a lot more choices than they did decades ago, and this on the whole has improved quality of life quite a bit.
But we do need to be mindful of the consequences of tech and policy change on the bargaining power of workers. The ideal situation of maximizing efficiently used free time (whether for work or leisure) as a cooperative game between employer and employee really only happens when workers on a whole are in a strong bargaining position, otherwise it's a race to the bottom like what we saw in the US in the 70s and 80s. A lot of factors combined to create an "oversupply" of labor, the consequences of which we are still dealing with today.
Fortunately, workers' bargaining power has been on a historic upswing since the mid-2010s. But we need to keep a close eye on making sure it doesn't get out of whack again. I write more about this here: https://2120insights.substack.com/p/wtf-happened-in-1971
Its much bigger than how it reflects in GDP. Women can now be economically independent of men. That is a massive change from just a few decades ago. Marriage is now a choice for professional women.
There are still the gender pay gaps, but that is more due to choices of careers and how society values those careers now than men being paid more than women for the same job.
I am in engineering. Women were non-existent in my field in the 1980s. We are now hiring them at an even pace with men and their pay scales are the same. They can now survey the potential mates and decide if they are suitable. They aren't forced to marry for survival anymore. A substantial percentage of young women professionals I know are single. Their existence and independence would have been highly unusual 40 years ago.
We now have the ability to find, educate, and hire top talent regardless of gender and race. That is a major competitive advantage over many other countries in the world. The talent pool is no longer just white men.
Equality has to improve in the lower income groups, but it has at least begun in the upper income levels which was not the case several decades ago. Distribution of child care etc. still needs to improve.
AI may also potentially be like the electric/gas stove, washer, dryer, and dishwasher. Those items revolutionized housework and turned many hours of work into minutes of labor time. The machines do the drudge work. They freed up massive amounts of time for housewives. For the hospitality industry, it required far fewer people which lowered costs and made hotels and restaurants much more accessible to the middle class.
What you are describing in this post is incredibly dystopian, dressed up with positive language to sound utopian.
Humans, in your telling, will still be around to *press the button* on the machine that does all the thinking, creating, adjusting and producing and then say "LGTM" like a pointy-haired middle-manager.
Then, as a reward for our efforts, we can go home and do an inferior job "for fun" for lack of any other purpose. Essentially this predicts that humans land in the role of permanent consumers, whose most important decision is which hobby to distract us from the gaping meaninglessness of existence.
The analogy of the boss still hiring a secretary even though he can type faster is very telling -- extrapolate that onto computers and people and you're describing a world where the AI is running everything and making all the important decisions and humans are still employed because we are occassionally useful at some minor task.
This post is able to spin this into a utopian story only by describing the process exclusively in terms of economics -- comparative advantage, utility, productivity, jobs, and so on. But I think it's relevant to most people's life satisfaction that you're predicting that every major endeavor that has given human beings meaning since at least the enlightenment, arguably for the last few millenia -- art, science, math, literature, invention, creativity, craftsmanship -- is about to be unceremoniously handed over to something else while we watch from a distance and get to fiddle with the output.
Yeah I absolutely agree. I was put off by the suggestion in the article that everyone will want to pursue creative work as a hobby instead of for money, but this is a super narrow view of the value hobbies provide. The assumption is that artists would happy just making the art and not being compensated for their work, but compensation is validation and can be part of the enjoyment of art even for hobbyists! I also question the authors' implication that all of this is unavoidable -- technology companies choose to develop and invest in LLMs. They're not inevitable but companies like OpenAI want us to think they are (of course, I know this is something of a moot point because they already exist, but it's worth asking why anyway).
Evgeny Morozov, I think, has a quote I like to steal about how bad society is at negotiating technological change. Roughly: If you try to raise an objection to a technology while it's still in development, you are rebuffed with "Well, it doesn't even exist yet, we'll deal with that problem when we get there." If you try to raise an objection to a technology after it's been released, the response is "Well, too late, it's already out there -- nothing we can do it about it now!"
To some extent, I do think there is a shape to technological progress that we can't completely control. But we can certainly control how we *respond* as a society to the technological changes -- nominally, we have an entire democracy whose purpose is explicitly to be the place where we negotiate how to respond to changes in the world. Whenever someone is pushing the "it's inevitable" view, they are actually making a political argument that something *shouldn't* be negotiated, discussed or regulated, presumably because *they would not like it to be.*
Yeah, I wasn't even addressing the class element here. To some extent, what Noah is describing here is just AI doing to the remaining middle class what capitalism has already done to the lower class. This definitely seems like it could lead to a winner-takes-all race to the bottom.
This is a horribly naive take on jobs. Replacing tens of thousands of good, well paying jobs as weavers with tens of thousands of wretched, poorly paying jobs as loom operators was not a wonderful thing for those tens of thousands of weavers. There's a reason workers turned to violence in a movement that only ended when the king intervened on the part of the works.
If you look at the structure of the US job market, one notable change has been the decline in well paying mid-level skill jobs in urban areas. These have been replaced by low paying jobs, and our workforce has been increasingly bifurcated. Those jobs have been replaced by automation. I'll cite Autour's "Work of the Past, Work of the Future" on this.
There used to be a much larger market for trained secretaries, skilled shift managers, paralegals, resource allocation experts and so on. Those jobs have been computerized out of existence without need of artificial intelligence. AI is just going to make it worse, and if it does a much worse job than the humans it replaced, it won't matter to the handful of companies that dominate each industry.
I know you are trying to be a techno-optimist, but new technologies can have costly side effects that impact tens of millions and can lead to political instability. If we are serious about these technological transformations being good things, we need to make sure that they aren't bad things.
One thing that gives me pause about the AI revolution is that the first 100 years of the industrial revolution made a few people rich but were absolutely miserable for the working class. Like worse than subsistence farming.
I write code for a living. Or, at least, my job would be described as "developer" or "software engineer".
The amount of time I spend actually doing tasks at which GitHub Copilot would help is vanishingly small.
For instance, in a large system, changing one line of code often means reading and understanding hundreds more to understand whether your one-line change impacts anything else. GitHub Copilot can't help with that. Nor can it help translate the requirements from stakeholders into something implementable, or decide whether your unit and system tests are adequate.
As such, I don't think GitHub Copilot is going to result in major productivity improvements from developers. It's an incremental bump, no more.
But Copilot to comment your code is worthwhile. Or check where documentation needs to be updated due to code change and draft the PR. Or suggest a slightly more optimal implementation. "Autocomplete" is an apt term as it can help with completing related tasks to your primary task.
I’ve found copilot examples useful in the same way as a StackOverlfow answer. I wouldn’t want to blindly use either without review and revision, but both can provide a valuable example to help me understand the structure of a solution. That includes introducing library methods that I was unaware of as well as how multiple methods are commonly used together.
"Autocomplete for everything" is a good way of putting it. But I think we need to take a more nuanced look at the impact on the job market:
- In industries at the cutting edge, and where demand isn't yet saturated, AI will allow us to produce more value and we could see increased employment.
- But industries where demand is close to saturated (and costs need to come down), I imagine we'll see job cuts. Since AI will tend to make us more efficient rather than replace most jobs entirely, this could take the form of reduced hours, but the way employment is structured in the US (high health insurance costs being tied to jobs) are a barrier to this. So in some areas there will be technological unemployment.
It will be interesting to see where the S-curve of generative AI takes us, and how steep technological unemployment could be at its peak. Hopefully we can learn from history and have policies ready to keep the labor market strong if needed-- I took a closer look at what happened in the 70s and 80s here and hope to do a similar speculative analysis of the 2029s and 30s soon. At least this time, demographic trends will help keep the labor market tighter than it could be otherwise: https://2120insights.substack.com/p/wtf-happened-in-1971
It’s hard to know what it means for demand to be “saturated”. In 1910 you might have thought the demand for music was saturated if there were already bands performing in every bar. But once you have recorded music, suddenly there’s a whole new range of contexts where people would want to listen to music where you wouldn’t want a band to be present, like inside your own home at all hours of the day.
True, it's hard to know if a market is truly "saturated".
In the case of music though, the consequences to the labor market were similar. Musicians as a percentage of the workforce stopped growing after 1910, stagnating until about 1980 when demand for music teachers started rising: https://twitter.com/Ethan_Heppner/status/1598455264084168704
This is similar to my model of the world. A lot of AI is going to be more like the cotton gin, than replacing our jobs completely. I 'solved' a science problem three weeks ago, and as is standard in the ML world, have spent three weeks since then performance optimizing my code and laboriously testing every component and fitting it into a DAG. I'd have happily not spent the last week autistically testing and writing every small component to do this, which I've written a million times in the past at various past jobs.
In the future tech teams won't need 10 engineers on permanent infra work to keep things from crashing. They can each then go and support an entire product on their own, with the tooling to keep things running.
I mean, when you think about it, is the AI revolution a bigger revolution than LAPACK? I mean eventually sure, but I've worked with oldhats who've talked about implementing their own matrix libraries before starting work, now we don't do any of that.
The only people who are scared are those who stopped thinking a decade ago, and get by only on their esoteric knowledge of some old framework that they're paid to keep running. But those people are cringe and need to be automated away anyway.
LAPACK is a great example of universally valuable code that made almost all numerical programmers more efficient, especially people like me with a PhD in PChem who has written a lot of analysis/modeling code. Two other code libraries have made a huge impact on my research, lsode and FFTW. Both are such pinnacles of performance it has changed aspects of how science uses numerical codes. On one of the IBM mini-computers we ran code on the hand optimized by IBM FFT routines (roughly they were leased but instead you paid up front and then a required annual maintenance plan, $5k to $10k, per machine, per year). FFTW made the fastest FFT available for free. The ODE library lsode is so good at many kinds of stiff coupled ODE models there hasn't been any effort to develop something new. Both of these developments were pure human intelligence and they transformed one small part of science for the foreseeable future. Generative AI is not generative in the human creation realm. It is really just fancy pattern matching.
"The increased wealth that AI delivers to society should allow us to afford more leisure time for our creative hobbies."
And yet, that wealth never seems to go to the workers, and so we don't ever seem to have more leisure time.
I work at a content marketing company that's starting to introduce AI services to clients. On the front-end, they'll be getting more articles for less money. As far as I can tell, on the back end, that means the writers patching up Jasper's writing - it's prone to circuitous logic and inventing facts - will be working more hours, not fewer.
I thought I would share that my partner, an artist, has adopted the use of Midjourney exactly as described and is churning out artwork with the AI completing the task of composition of the piece for her. However she fears that we’re months away from AI advancing to the point of ironing out the weird kinks like the extra fingers and other uncanny valley aspects. At that point, surely any paying customer will just source their artwork straight from the AI?
For my own part I’m an actuary and would appreciate the Excel autocomplete. I wonder if the AI will be as prone to the catastrophic spreadsheet error as the human.
Yes, but in the case of Copilot, the catastrophic error will be in algorithm generated code used in millions of applications at a very low level. I can completely see corporate lawyers refusing to allow their engineers to alter algorithm generated code for liability reasons, even when engineers find obvious bugs: "it's code generated by an approved AI, who the hell do you think you are?" This happens in bureaucracies with processes all the time; now it will happen with low level code too.
I feel like artistic understanding of what is actually pleasing to human eye - as well as knowledge of (key)words that might describe it more precisely then "paint me a pretty picture" - will still be valuable enough for artist to remain as intermediate step.
And paying customers will likely still value their time more then to spend it on understanding prompting, generating, and curating hundreds to thousands of pictures. More so when they need some specific non-generic requirements.
Otherwise stock images would have already replaced artists.
Have you ever heard AI produced music? Today's programs will compose in the style of Bach or Chopin or Mozart, and you can tell the difference. The only training necessary was for the program to listen to the entire collected works of Bach and deduce the patterns that make something "Bach".
Knowing what is "pleasing to the human eye" is a basic pattern recognition problem. These were solved years ago.
If it is solved then it isn't yet well-integrated in current image generation in opposite direction. People take a lot of basics that come preinstalled with human brains for granted.
And generally people hold more context to what they might need beyond pleasing or their specific tastes - though like AIs they might need to be prompted to do so.
Humans can provide better "interface" to getting job done - assuming you can deal with all other complexities attached.
Interesting, but I think I agree more with the caveat than the rest of the piece. The technology seems to be improving so rapidly that what's predicted here might be a very brief phase before the AI takes over everything.
Also: tell us the truth. Is roon a person or something else?
I wonder if the pace of trust might actually lag significantly behind the technology. I wrote a couple of pieces exploring how much occupational change has happened in previous decades, based on Census data going all the way back to 1870:
Wow, I can't wait to pick out my favorite statistically likely choice for 8 hours a day! Oops, I meant 2 hours, because we'll all have more leisure time and get paid 4x as much :)
It's also very important to me that everyone has a job, because otherwise we couldn't get work done, which is necessary. So I'm very glad AI will allow me to keep working for the rest of my life, because I like working, but who doesn't? In fact, I hope AI creates more jobs so I can get two of them. My dream job is to generate uncontroversial and inoffensive listicles using AI, but I would settle for generating terms of service for software. Secretly, I'd love to make AI generated music that has lyrics that no reasonable person in a normal culture would find upsetting.
The future sounds so great! I can't wait to race down it with you all!
> “[T]here was a type of employee at the beginning of the Industrial Revolution whose job and livelihood largely vanished in the early twentieth century. This was the horse. The population of working horses actually peaked in England long after the Industrial Revolution, in 1901, when 3.25 million were at work. Though they had been replaced by rail for long-distance haulage and by steam engines for driving machinery, they still plowed fields, hauled wagons and carriages short distances, pulled boats on the canals, toiled in the pits, and carried armies into battle. But the arrival of the internal combustion engine in the late nineteenth century rapidly displaced these workers, so that by 1924 there were fewer than two million. There was always a wage at which all these horses could have remained employed. But that wage was so low that it did not pay for their feed.” - A Farewell to Alms
I like the sandwich model of work a lot. I think the next question is how much that improves productivity and what that means for employment in different fields. If an illustrator can now produce 10X as much output, does that mean we need 10X less illustrators? Maybe. I think having enough context to do a particular project well will still be a scarce, less scalable commodity.
I do think we'll be awash in generic text and image outputs (think SEO-style content and stock images). The value/market price of that type of media will go down massively, and that will lead to dislocations. It's also going to be extremely annoying to search for things on the web as the results get clogged up with 95 different AI-generated versions of an article, none of which have real differentiated insight. I could see generative AI hastening the decline of the current generation of search ranking algorithms. If I were Google, I would be hard at work on algorithms that can detect novel information in a piece of writing compared to superficially similar articles.
I'm less convinced by the point about GPU/AI scarcity. We don't see web searches being rationed now, and while AI auto-completion is more likely to be a paid service, I have a hard time seeing us "running out" of AI capacity.
Great points! It's interesting to consider ways in which AI might make us _less_ efficient, like making it harder to find good-quality information, or having to up IT security defenses in the face of more powerful AI-enabled bots.
This improved security probably extends to the physical world as well-- does advanced AI make it easier for a bad actor to "autocomplete" their design for a guided missile, or an engineered pandemic?
Looking forward to Noah's posts coming, like, four times a day, with the help of his AI assistant.
The problem with that is that the attention of readers remains limited. There's a problem somewhat akin to Baumol's Cost Disease here... One aspect of the market grows with technology while another remains stubbornly limited.
Unless you intend to have your own AI assistant read the posts for you and summarize them. (I am reminded of the Electric Monk in Dirk Gently's Holistic Detective Agency. Just as a dishwasher does all that rote, boring dishwashing for you, the Electric Monk does all the rote, boring _believing_ for you, to free up time for more productive activities.)
We'll use our AI assistants to summarize and prioritize the flood of Noahpinion articles. Our assistants could even learn to elaborate on things it believes we’ll be interested in. If we only want 2 minutes of content one day, then it will be particularly aggressive in filtering and summarizing. One day someone might ask for 4 hours, in which case their AI will aggressively expand and elaborate.
And of our course our AI assistants will handle the drudgery of commenting on our behalf. We can simply prompt, “write 10 paragraphs with authoritative references detailing why Noah is not only incorrect, but also deeply immoral.” And his AI will auto-reply. It will also provide Noah a daily summary of user comments; commonly just reporting, “you still have many foolish readers, but they’re slowly learning through our reinforcement.”
you have a very witty AI! thanks
My AI assistant will summarize the articles and feed them directly into my brain using a machine-brain-interface.
The problem with training an AI on existing code is that it won't necessarily generate _good_ code, and if you as an engineer are not sharp enough to catch that, you can wind up propagating bugs. https://www.spiceworks.com/it-security/security-general/news/40-of-code-produced-by-github-copilot-vulnerable-to-threats-research/
Literally the example that Noah gives has a bug in it!
Yeah.
How do you patch a copilot generated bug that gets reused millions of times in low level code throughout the world? That's going to be an industry in itself.
I agree, but unfortunately that doesn't make it worse than human software engineers.
Yep. I am a senior software engineer with tons of experience and AI code autocomplete is extremely useful, but not necessarily due to the comparative advantage factor. It's a search engine.
Using generative code AI, I can answer the question, "How would someone else write this thing that I know how to write?" and get a coherent answer that I can evaluate for effectiveness and possibly incorporate into my own technique.
Similarly, instead of trying to figure out the API of a particular complex package or system, I can just ask the AI to get me started. It might get things wrong, but it'll dump the most commonly used calls, routes, etc. in my lap.
Also, it teaches conventions. If I'm not exactly sure what the canonical way to, say, write a route in a new web api framework I'm presented with, I can just ask co-pilot and get a ton of useful context. This can sometimes be better than the official documentation for systems that are poorly maintained.
I think this holds true basically everywhere. You can ask generative AI to explain something to you and get basically a distillation of the median internet understanding of that topic as a crash course.
Yep! I’m much more interested in AI searching to find an existing library or API rather than helping me write new code. Similarly, I’d appreciate it helping me understand what it found. It could leverage its familiarity with my knowledge and experience to quickly get me up to speed with the new library or API.
I love posts like this! But who gets the wealth of the added productivity from those tasks that are automated? The Governance and wealth distribution of A.I. at scale, either leads to Dystopia or Utopia, depending on how this is handled.
That's what depends! The balance of the displacement effect, the productivity effect, and the reinstatement effect will be what determines whether capital or labor benefits more from the new technology...
Great question! I've wondered how much this might fundamentally shift how much income goes to capital vs. labor, and even begs the question of how much wealth is it possible to accumulate?
Hoping to explore these questions in a future newsletter myself!
Like many other things today, this sounds like another way for the educated / laptop class (of which Noah is part -- nothing personal, man) to distance itself from everyone else. Even Noah's title betrays this bias. If you're a roofer or a plumber, there's no "autocomplete" for you.
Just like the industrial revolution, it alters the labor/capital balance in ways unfavorable to labor. After all, if capital can become labor (via AI and robotics), that's pretty detrimental for labor. It's noteworthy that the last time a shift like this occurred, we got Karl Marx.
Since labor is a complement to AI (no AI is going to fixing roofs or sinks anytime soon) their skills should become more valuable.
Sorry, but you lost me there. How will rising, AI-driven, productivity of the laptop class benefit working class, real world, physical laborers?
As total wealth increases, consumers will spend more overall, including on services not replicable by AI.
The 1980's called; they want their economic policy back.
If AI increases productivity of the laptop class significantly, this will cause either that class or the owner class to reap the vast majority of those benefits. Will they spend more? Sure, but the effects of the resulting inequality dwarf the effects of any increase in downstream income.
There's more to life than Pareto-optimality.
> If AI increases productivity of the laptop class significantly, this will cause either that class or the owner class to reap the vast majority of those benefits.
Not necessarily. There could easily be widespread unemployment of white collar workers at least in the short term. Industries adjust but it takes time. If we decided to import 10x the number of white collar workers for existing jobs it would likely drive their wages down. A massive and sudden productivity boost could easily look like the same thing. The white collar workers who kept their jobs would likely benefit enormously but a much larger group would lose their employment entirely.
As households get wealthier they spend more on status-seeking, positional goods which are intrinsically zero-sum: houses in the right areas, fine art and jewellery, and so on. Not so much relatively speaking on services that promote the flow of money around the economy.
I agree with Brian: this seems like it will inevitably increase the wage gap between the managerial/professional class and ordinary workers: salespeople, restaurant staff, laborers, truck drivers and other goods distribution worker, nurses and other low-paid health workers, even teachers and assistants.
(That's something like 30% of the workforce just in that list according to the BLS. There are more: janitors, groundskeepers, repair workers, construction workers, security guards, ...)
AI-based research and image generation could potentially give roofers and plumbers easy access to information about a building's interior layout, and records of previous maintenance or repair work, which might be as helpful to them as medical history and x-rays are to a doctor.
The OP is about the near term. In the long term, the impact of technology will be to make financial wealth much less important. The real question is what "near" and "long" actually mean in terms of years.
I think the point is that there's already a firmly entrenched technology that could lead to the dystopic outcome: rent-seeking.
Will it though? After the industrial revolution people of different wealth strata can all eat their fill, wear warm clothing, heat houses automatically, etc. but it doesn't seem quite accurate to say financial wealth is much less important. Possibly technological developments will smooth out some of the differences between wealth strata we see today, but keep introducing new ones.
That is a joint responsibility of employer and employee. Ultimately, the goal is to end up with "free time" that can be used productively. Both the employer and employer can be wise with it and greatly increase opportunity for both, or it can be squandered. Viewing it as a zero-sum game with the focus on diverting the benefits to only one party is disastrous in the long run.
The vast majority of engineers, finance sector, medical workers etc. today would have been shoveling manure, pulling a plow, and cutting firewood 200 years ago. The industrial revolution, especially on farms, has freed up a huge sector of humanity from those tasks to do other things. The people who were able to take advantage of the opportunity and were inclined to do so are immensely more wealthy today than they would have been in the cold and dark.
Anti-discrimination laws have been opening up opportunities for women and minorities since the 1950s. It is likely that Title IX and the reduction in Jim Crow segregation were cornerstones of the immense economic expansion in the US in the 1980s and 1990s as the beneficiaries entered the workforce. We continue to have opportunities to maximize our human capital. However, history is replete with examples of society-wide suppression of human capital due to bigotry and hate, so we must avoid that.
There's actually pretty good science that says almost all the income gains since the 1970's have come from women entering the workforce. Which is an illusion. A wife who stays home with the kids is economically invisible. A wife who works bumps GDP both with her earnings and her daycare consumption. It is this flaw in our GDP accounting that has allowed us to pretend most households' economic condition has been improving for the last 50 years.
Ernest has some good points-- a lot more people have a lot more choices than they did decades ago, and this on the whole has improved quality of life quite a bit.
But we do need to be mindful of the consequences of tech and policy change on the bargaining power of workers. The ideal situation of maximizing efficiently used free time (whether for work or leisure) as a cooperative game between employer and employee really only happens when workers on a whole are in a strong bargaining position, otherwise it's a race to the bottom like what we saw in the US in the 70s and 80s. A lot of factors combined to create an "oversupply" of labor, the consequences of which we are still dealing with today.
Fortunately, workers' bargaining power has been on a historic upswing since the mid-2010s. But we need to keep a close eye on making sure it doesn't get out of whack again. I write more about this here: https://2120insights.substack.com/p/wtf-happened-in-1971
Its much bigger than how it reflects in GDP. Women can now be economically independent of men. That is a massive change from just a few decades ago. Marriage is now a choice for professional women.
There are still the gender pay gaps, but that is more due to choices of careers and how society values those careers now than men being paid more than women for the same job.
I am in engineering. Women were non-existent in my field in the 1980s. We are now hiring them at an even pace with men and their pay scales are the same. They can now survey the potential mates and decide if they are suitable. They aren't forced to marry for survival anymore. A substantial percentage of young women professionals I know are single. Their existence and independence would have been highly unusual 40 years ago.
We now have the ability to find, educate, and hire top talent regardless of gender and race. That is a major competitive advantage over many other countries in the world. The talent pool is no longer just white men.
Equality has to improve in the lower income groups, but it has at least begun in the upper income levels which was not the case several decades ago. Distribution of child care etc. still needs to improve.
AI may also potentially be like the electric/gas stove, washer, dryer, and dishwasher. Those items revolutionized housework and turned many hours of work into minutes of labor time. The machines do the drudge work. They freed up massive amounts of time for housewives. For the hospitality industry, it required far fewer people which lowered costs and made hotels and restaurants much more accessible to the middle class.
What you are describing in this post is incredibly dystopian, dressed up with positive language to sound utopian.
Humans, in your telling, will still be around to *press the button* on the machine that does all the thinking, creating, adjusting and producing and then say "LGTM" like a pointy-haired middle-manager.
Then, as a reward for our efforts, we can go home and do an inferior job "for fun" for lack of any other purpose. Essentially this predicts that humans land in the role of permanent consumers, whose most important decision is which hobby to distract us from the gaping meaninglessness of existence.
The analogy of the boss still hiring a secretary even though he can type faster is very telling -- extrapolate that onto computers and people and you're describing a world where the AI is running everything and making all the important decisions and humans are still employed because we are occassionally useful at some minor task.
This post is able to spin this into a utopian story only by describing the process exclusively in terms of economics -- comparative advantage, utility, productivity, jobs, and so on. But I think it's relevant to most people's life satisfaction that you're predicting that every major endeavor that has given human beings meaning since at least the enlightenment, arguably for the last few millenia -- art, science, math, literature, invention, creativity, craftsmanship -- is about to be unceremoniously handed over to something else while we watch from a distance and get to fiddle with the output.
Brave New World. I'm sure Centrifugal Bumble Puppy will provide meaning for my life.
Yeah I absolutely agree. I was put off by the suggestion in the article that everyone will want to pursue creative work as a hobby instead of for money, but this is a super narrow view of the value hobbies provide. The assumption is that artists would happy just making the art and not being compensated for their work, but compensation is validation and can be part of the enjoyment of art even for hobbyists! I also question the authors' implication that all of this is unavoidable -- technology companies choose to develop and invest in LLMs. They're not inevitable but companies like OpenAI want us to think they are (of course, I know this is something of a moot point because they already exist, but it's worth asking why anyway).
Evgeny Morozov, I think, has a quote I like to steal about how bad society is at negotiating technological change. Roughly: If you try to raise an objection to a technology while it's still in development, you are rebuffed with "Well, it doesn't even exist yet, we'll deal with that problem when we get there." If you try to raise an objection to a technology after it's been released, the response is "Well, too late, it's already out there -- nothing we can do it about it now!"
To some extent, I do think there is a shape to technological progress that we can't completely control. But we can certainly control how we *respond* as a society to the technological changes -- nominally, we have an entire democracy whose purpose is explicitly to be the place where we negotiate how to respond to changes in the world. Whenever someone is pushing the "it's inevitable" view, they are actually making a political argument that something *shouldn't* be negotiated, discussed or regulated, presumably because *they would not like it to be.*
Ah important point, thank you for the thoughtful response!
Yeah, I wasn't even addressing the class element here. To some extent, what Noah is describing here is just AI doing to the remaining middle class what capitalism has already done to the lower class. This definitely seems like it could lead to a winner-takes-all race to the bottom.
This is a horribly naive take on jobs. Replacing tens of thousands of good, well paying jobs as weavers with tens of thousands of wretched, poorly paying jobs as loom operators was not a wonderful thing for those tens of thousands of weavers. There's a reason workers turned to violence in a movement that only ended when the king intervened on the part of the works.
If you look at the structure of the US job market, one notable change has been the decline in well paying mid-level skill jobs in urban areas. These have been replaced by low paying jobs, and our workforce has been increasingly bifurcated. Those jobs have been replaced by automation. I'll cite Autour's "Work of the Past, Work of the Future" on this.
There used to be a much larger market for trained secretaries, skilled shift managers, paralegals, resource allocation experts and so on. Those jobs have been computerized out of existence without need of artificial intelligence. AI is just going to make it worse, and if it does a much worse job than the humans it replaced, it won't matter to the handful of companies that dominate each industry.
I know you are trying to be a techno-optimist, but new technologies can have costly side effects that impact tens of millions and can lead to political instability. If we are serious about these technological transformations being good things, we need to make sure that they aren't bad things.
Well maybe we'll get a new AI driven labor movement. That's the only optimistic thing I can think of!
One thing that gives me pause about the AI revolution is that the first 100 years of the industrial revolution made a few people rich but were absolutely miserable for the working class. Like worse than subsistence farming.
I write code for a living. Or, at least, my job would be described as "developer" or "software engineer".
The amount of time I spend actually doing tasks at which GitHub Copilot would help is vanishingly small.
For instance, in a large system, changing one line of code often means reading and understanding hundreds more to understand whether your one-line change impacts anything else. GitHub Copilot can't help with that. Nor can it help translate the requirements from stakeholders into something implementable, or decide whether your unit and system tests are adequate.
As such, I don't think GitHub Copilot is going to result in major productivity improvements from developers. It's an incremental bump, no more.
But Copilot to comment your code is worthwhile. Or check where documentation needs to be updated due to code change and draft the PR. Or suggest a slightly more optimal implementation. "Autocomplete" is an apt term as it can help with completing related tasks to your primary task.
I’ve found copilot examples useful in the same way as a StackOverlfow answer. I wouldn’t want to blindly use either without review and revision, but both can provide a valuable example to help me understand the structure of a solution. That includes introducing library methods that I was unaware of as well as how multiple methods are commonly used together.
"Autocomplete for everything" is a good way of putting it. But I think we need to take a more nuanced look at the impact on the job market:
- In industries at the cutting edge, and where demand isn't yet saturated, AI will allow us to produce more value and we could see increased employment.
- But industries where demand is close to saturated (and costs need to come down), I imagine we'll see job cuts. Since AI will tend to make us more efficient rather than replace most jobs entirely, this could take the form of reduced hours, but the way employment is structured in the US (high health insurance costs being tied to jobs) are a barrier to this. So in some areas there will be technological unemployment.
It will be interesting to see where the S-curve of generative AI takes us, and how steep technological unemployment could be at its peak. Hopefully we can learn from history and have policies ready to keep the labor market strong if needed-- I took a closer look at what happened in the 70s and 80s here and hope to do a similar speculative analysis of the 2029s and 30s soon. At least this time, demographic trends will help keep the labor market tighter than it could be otherwise: https://2120insights.substack.com/p/wtf-happened-in-1971
It’s hard to know what it means for demand to be “saturated”. In 1910 you might have thought the demand for music was saturated if there were already bands performing in every bar. But once you have recorded music, suddenly there’s a whole new range of contexts where people would want to listen to music where you wouldn’t want a band to be present, like inside your own home at all hours of the day.
True, it's hard to know if a market is truly "saturated".
In the case of music though, the consequences to the labor market were similar. Musicians as a percentage of the workforce stopped growing after 1910, stagnating until about 1980 when demand for music teachers started rising: https://twitter.com/Ethan_Heppner/status/1598455264084168704
Great post!
This is similar to my model of the world. A lot of AI is going to be more like the cotton gin, than replacing our jobs completely. I 'solved' a science problem three weeks ago, and as is standard in the ML world, have spent three weeks since then performance optimizing my code and laboriously testing every component and fitting it into a DAG. I'd have happily not spent the last week autistically testing and writing every small component to do this, which I've written a million times in the past at various past jobs.
In the future tech teams won't need 10 engineers on permanent infra work to keep things from crashing. They can each then go and support an entire product on their own, with the tooling to keep things running.
I mean, when you think about it, is the AI revolution a bigger revolution than LAPACK? I mean eventually sure, but I've worked with oldhats who've talked about implementing their own matrix libraries before starting work, now we don't do any of that.
The only people who are scared are those who stopped thinking a decade ago, and get by only on their esoteric knowledge of some old framework that they're paid to keep running. But those people are cringe and need to be automated away anyway.
LAPACK is a great example of universally valuable code that made almost all numerical programmers more efficient, especially people like me with a PhD in PChem who has written a lot of analysis/modeling code. Two other code libraries have made a huge impact on my research, lsode and FFTW. Both are such pinnacles of performance it has changed aspects of how science uses numerical codes. On one of the IBM mini-computers we ran code on the hand optimized by IBM FFT routines (roughly they were leased but instead you paid up front and then a required annual maintenance plan, $5k to $10k, per machine, per year). FFTW made the fastest FFT available for free. The ODE library lsode is so good at many kinds of stiff coupled ODE models there hasn't been any effort to develop something new. Both of these developments were pure human intelligence and they transformed one small part of science for the foreseeable future. Generative AI is not generative in the human creation realm. It is really just fancy pattern matching.
"The increased wealth that AI delivers to society should allow us to afford more leisure time for our creative hobbies."
And yet, that wealth never seems to go to the workers, and so we don't ever seem to have more leisure time.
I work at a content marketing company that's starting to introduce AI services to clients. On the front-end, they'll be getting more articles for less money. As far as I can tell, on the back end, that means the writers patching up Jasper's writing - it's prone to circuitous logic and inventing facts - will be working more hours, not fewer.
As they say, we sure live in interesting times.
I thought I would share that my partner, an artist, has adopted the use of Midjourney exactly as described and is churning out artwork with the AI completing the task of composition of the piece for her. However she fears that we’re months away from AI advancing to the point of ironing out the weird kinks like the extra fingers and other uncanny valley aspects. At that point, surely any paying customer will just source their artwork straight from the AI?
For my own part I’m an actuary and would appreciate the Excel autocomplete. I wonder if the AI will be as prone to the catastrophic spreadsheet error as the human.
Yes, but in the case of Copilot, the catastrophic error will be in algorithm generated code used in millions of applications at a very low level. I can completely see corporate lawyers refusing to allow their engineers to alter algorithm generated code for liability reasons, even when engineers find obvious bugs: "it's code generated by an approved AI, who the hell do you think you are?" This happens in bureaucracies with processes all the time; now it will happen with low level code too.
I feel like artistic understanding of what is actually pleasing to human eye - as well as knowledge of (key)words that might describe it more precisely then "paint me a pretty picture" - will still be valuable enough for artist to remain as intermediate step.
And paying customers will likely still value their time more then to spend it on understanding prompting, generating, and curating hundreds to thousands of pictures. More so when they need some specific non-generic requirements.
Otherwise stock images would have already replaced artists.
Have you ever heard AI produced music? Today's programs will compose in the style of Bach or Chopin or Mozart, and you can tell the difference. The only training necessary was for the program to listen to the entire collected works of Bach and deduce the patterns that make something "Bach".
Knowing what is "pleasing to the human eye" is a basic pattern recognition problem. These were solved years ago.
If it is solved then it isn't yet well-integrated in current image generation in opposite direction. People take a lot of basics that come preinstalled with human brains for granted.
And generally people hold more context to what they might need beyond pleasing or their specific tastes - though like AIs they might need to be prompted to do so.
Humans can provide better "interface" to getting job done - assuming you can deal with all other complexities attached.
Interesting, but I think I agree more with the caveat than the rest of the piece. The technology seems to be improving so rapidly that what's predicted here might be a very brief phase before the AI takes over everything.
Also: tell us the truth. Is roon a person or something else?
I wonder if the pace of trust might actually lag significantly behind the technology. I wrote a couple of pieces exploring how much occupational change has happened in previous decades, based on Census data going all the way back to 1870:
https://2120insights.substack.com/p/job-growth-across-the-decades
It will be interesting to see how the 2020s-2030s compare. But if we measure occupations based on willrobotstakemyjob.com's index, progress has actually just been pretty steady since 1970: https://2120insights.substack.com/p/a-preview-of-coming-attractions
Wow, I can't wait to pick out my favorite statistically likely choice for 8 hours a day! Oops, I meant 2 hours, because we'll all have more leisure time and get paid 4x as much :)
It's also very important to me that everyone has a job, because otherwise we couldn't get work done, which is necessary. So I'm very glad AI will allow me to keep working for the rest of my life, because I like working, but who doesn't? In fact, I hope AI creates more jobs so I can get two of them. My dream job is to generate uncontroversial and inoffensive listicles using AI, but I would settle for generating terms of service for software. Secretly, I'd love to make AI generated music that has lyrics that no reasonable person in a normal culture would find upsetting.
The future sounds so great! I can't wait to race down it with you all!
Comparative advantage doesn't always save you.
> “[T]here was a type of employee at the beginning of the Industrial Revolution whose job and livelihood largely vanished in the early twentieth century. This was the horse. The population of working horses actually peaked in England long after the Industrial Revolution, in 1901, when 3.25 million were at work. Though they had been replaced by rail for long-distance haulage and by steam engines for driving machinery, they still plowed fields, hauled wagons and carriages short distances, pulled boats on the canals, toiled in the pits, and carried armies into battle. But the arrival of the internal combustion engine in the late nineteenth century rapidly displaced these workers, so that by 1924 there were fewer than two million. There was always a wage at which all these horses could have remained employed. But that wage was so low that it did not pay for their feed.” - A Farewell to Alms
I like the sandwich model of work a lot. I think the next question is how much that improves productivity and what that means for employment in different fields. If an illustrator can now produce 10X as much output, does that mean we need 10X less illustrators? Maybe. I think having enough context to do a particular project well will still be a scarce, less scalable commodity.
I do think we'll be awash in generic text and image outputs (think SEO-style content and stock images). The value/market price of that type of media will go down massively, and that will lead to dislocations. It's also going to be extremely annoying to search for things on the web as the results get clogged up with 95 different AI-generated versions of an article, none of which have real differentiated insight. I could see generative AI hastening the decline of the current generation of search ranking algorithms. If I were Google, I would be hard at work on algorithms that can detect novel information in a piece of writing compared to superficially similar articles.
I'm less convinced by the point about GPU/AI scarcity. We don't see web searches being rationed now, and while AI auto-completion is more likely to be a paid service, I have a hard time seeing us "running out" of AI capacity.
Great points! It's interesting to consider ways in which AI might make us _less_ efficient, like making it harder to find good-quality information, or having to up IT security defenses in the face of more powerful AI-enabled bots.
This improved security probably extends to the physical world as well-- does advanced AI make it easier for a bad actor to "autocomplete" their design for a guided missile, or an engineered pandemic?