Is this time really different?
The AI productivity boom and why we actually may not be entering a recession
I wrote this article over 6 months ago but never actually published it. There’s been a lot keeping me busy between my job, the market and my personal life, but that is no excuse as I look back on this draft and realize it was sitting there in my drive waiting to be shipped.
Sometimes writers exhibit funny patterns of avoidant behaviour towards publishing when going through major changes in life. I still feel that there is much more to be said about AI and economic cycles than what’s been covered in this article, but that’s a story for the next piece.
Several months ago (June ‘23) when the Bitcoin spot ETFs were announced I spun up a bear rant that we are not yet back. That thesis remained correct up until the narrative around approving ETFs started intensifying in late October.
While I’m still not fully convinced we are back in the bull market, mainly because the money has yet to actually start flowing into these ETFs and speculation leading up to the news event can only last so long, I’m inclined to say that we may not see a full on recession like most doomsayers are predicting.
The reasoning for an upcoming recession is quite straightforward for the average macro enjoyer. Whenever inflation gets out of control and the Federal Reserve is forced to raise interest rates, the cost of money increases and economic activity slows.
A recession typically hits around 3 years after the first rate hike—that would be Feb-Mar 2025. The following chart is an extensive data-driven historical analysis into the pattern of events that typically follow. TL;DR we are deep in the muck of it now.
Typical sequence of macroeconomic events after the first rate hike
So instead of a bear rant, this time I'm going to explore the counterargument on a longer time frame because it is much more complex:
I would like to argue that we are in an era of change so pivotal, that even though macroeconomic history may have followed the same pattern over the last century, the map is not the territory.
History rhymes, but disruptive technology disrupts, and AI is doing that just now. Plus three global giants are currently engaged in kinetic, economic, and virtual proxy wars. It is in each party’s best interest to keep their respective economies chugging along, even if that entails borrowing from their future productivity.
Artificial Intelligence and the productivity boom
New technologies take time to prove themselves truly useful. It was not for several decades after the invention of electricity that it became common in households around the world. The printing press took centuries to have a significant cultural impact.
The industrial revolution witnessed a productivity boom that sped up life exponentially. The internet revolutionized distribution of information. The blockchain brought a layer of financialization and ownership to the internet.
Now the problem we are facing is harnessing these enormous powers we have created. The productivity boom of the industrial revolution is causing enormous strain on our planet’s resources. The abundance of information on the internet is polluting people’s mental health. The open permissionless nature of the blockchain is enabling scammers to take advantage of the unwitting.
If there’s only one way that AI will bring value to our society, it is as a curator and refiner of human forces. Proper allocation of resources and information will carry humanity across the brink of the next growth revolution.
It is a mythological force, creating mages out of common men as we hurtle towards singularity.
It took our jobs
AI has existed for decades (my mother even wrote a thesis on it in the 90’s), but it only became commonly known in Nov 2022 when OpenAI made ChatGPT a public-facing consumer application. Now this year a staggering 25% of US venture funding has gone into AI-related startups.
There's a lot of fear about AI replacing people's jobs, which is not unwarranted. A vast majority of "manual" labour computer tasks are already being replaced, from generic logo design to proofreading to basic accounting.
While many are rejoicing at the massive increase in productivity and savings in outsourcing, replacing jobs does not bode well for the general public. If society undergoes a massive "productivity boom" all of a sudden that leaves those at the bottom rung of the skill ladder out in the cold, our fearless leaders are going to have to find a way to placate the masses. Historically, a lack of economic growth has led to political instability and revolutionary rumblings.
Personally, I envision a future where automation of monotonous tasks causes a mass migration of labour towards humanitarian jobs in what could be called a hyper-personalization boom:
Healthcare becomes more accessible with greater availability of human care and support to provide personalized hospital experiences;
Therapy becomes increasingly abstracted with forms such as yoga, coaching, sound healing, etc. emerging to better equip humanity emotionally and spiritually to handle this transition;
Education becomes more effective with students studying in tight-knit focus groups and socratic discussions, tailor-made to individual learning needs;
…among countless other applications. All of which will be supported by mass data analytics and aggregation to understand individual needs, pinpoint issues, and further refine methods and approaches.
The question is how fast workers can adapt. McKinsey expects the trajectory of resilient and growing occupations to outpace “hit and declining jobs”, but exactly how smoothly that can happen is yet to be seen.
Lower skilled jobs may have not have anywhere to go immediately. They will need upskilling programs to ease with the transition where training is incentivized or subsidized.
Governments that step in to alleviate the pain of transition, just like many did during covid, will be able to reposition their labour force to absorb the shock. If this occurs in the form of a Universal Basic Income or other form of wage cushion, it will be inevitably inflationary, i.e. it would require the creation of more money, which would not lead to a recession.
But is AI even useful enough?
One of my favourite counterarguments to the thesis of AI disruption is “Sure, but AI makes mistakes all the time. It’s overhyped.”
Overhyped I can certainly agree with (does Nvidia really solve $1 trillion worth of problems? [Update: Now $3 trillion]) but that is the nature of virtually any economic cycle in history, from tulips to NFTs.
Yes, AI makes mistakes, but that is to be expected with any nascent technology. “It’s 98% operator,” as we say in the blue collar industry whenever someone blames the equipment for a mistake. Proper verification of the accuracy is entirely in the hands of the person operating it.
Some examples I’ve come across of people saving themselves countless hours of work:
Idea generation: A founder told me the other day he likes to test ideas by speaking with Pi, running that output through Claude, and then feeding questions into ChatGPT. Think about the impact on the speed of innovation if we can refine startup ideas at scale.
Branding: A large part of branding is the discovery process. Instead of days or weeks of back and forth with an agency, trying to get through a desired vision, make Midjourney chew through as many prompts as you can imagine. Shortlist, agree on a general art direction, and send off the best outputs to the designer to refine for a final version.
Writing: Anyone who says ChatGPT can't write doesn't know how to prompt it. Sure, it isn't creative but it can come up with a great framework for content if you make it prompt you first with questions to understand the context, audience, and style needed. It is especially good at formulaic texts, like product specifications documents or reports.
Legal: ChatGPT is excellent at drafting legal contracts if you take it section by section and loop your feedback. This can save hours of billing on legal advice because you can create a template for the legal experts to further refine.
Coding: I have never coded in my life but I am now spinning up telegram bots and staking nodes. Granted, it takes a lot of time to do proper troubleshooting, but if you know what you want to build, it massively accelerates time to ship.
Interview prep: Let this video of a guy interviewing as an aerospace engineer speak for itself, even if it is a joke.
“It frees the mind” said a developer about using AI, you need to worry less about less meaningful things and you can be more creative and think about problem to solution.
It is obvious that AI is not a silver bullet but only a tool that helps reach the final goal quicker. Human discretion is critical in ensuring the outcome accurately addresses the problem.
Financial stability is a matter of national security
In an era of changing world order, capital is only loyal to the bottom line.
The openness of American open capital accounts is a double-edged sword—money can flow in from anywhere in the world but it can flee just as easily.
And as sentiment starts to turn, nations are disincentivized from holding US bonds, especially if they would rather sponsor innovation locally. Furthermore, US dollar hegemony is facing disruption by CBDCs introducing alternative global settlement systems.
Huawei's breakthrough with their new smartphone shows they are perfectly capable of manufacturing chips of similar standards without depending on western technology. While I believe that innovation will remain concentrated around the US for a while, it will continue to disperse as rival nations try to build their own local versions of Silicon Valley. This will take time, however, as fostering a culture of risk-taking and entrepreneurship is no easy feat.
We are now in a new kind of cold war, where instead of an arms race, it’s a technology race. It’s a race to get control over the fastest, smartest, most powerful technology before anyone else can.
“I'm having specifically a lot of schizophrenic conversations with people in DC right now where if I talk to them and China doesn't come up, they just like hate tech, they hate American tech companies, they hate AI, they hate social media, they hate this, they hate that, they hate crypto, they hate everything and they just want to like punish and like ban and like they're just like very very negative.
But then if we have a conversation half an hour later when we talk about China then the conversation is totally different: now we need a partnership between the US government and American tech companies to defeat China. It’s like the exact opposite discussion.”
— Marc Andreessen
Losing the lead in the AI race would prove fatal for the US. Entering a recession means funding would dry up. In today's hypermobile world, founders with innovative ideas can easily hop on a plane and fly out to meet their friendly overseas sovereign wealth funds, who are more than happy to throw mountains of money at innovators to build locally.
Unless of course they regulate it and close the system, as has happened time and time again throughout history in the rise and fall of information empires. Considering the existential threat that AI poses to humanity, it would not come as a surprise if it was declared a matter of national security.
And so we have several massive forces tugging on each other simultaneously from different directions. One is the disruptive capabilities of AI and its threat to displace a significant amount of the workforce, triggering the need for a massive capital injection from central banks to support the masses. The other is the superpowers of the world stage jostling to protect their own interests through increasing abstractions of proxy wars.
AI could tip the scales of the future of humanity. How and in which direction is yet to be seen.