The AI Revolution

Today, I will share the final part of what has become a five-part summer letter to IMA clients. The letter is quite lengthy, so for easier reading, I have divided it into smaller parts. In the previous four parts, I discussed the economy, the Magnificent 7, past booms and busts and EVs. In this part, I will discuss AI from the following question:

Question: AI is changing a number of industries. Citi released a report predicting that 54% of jobs in the banking industry could be automated with AI.  How is IMA able to invest in a way to take advantage of AI disruption?

In these client letters, I am not selling anything; they are written to IMA clients, who have already bought into what we are doing. I don’t like sanitizing my letters (rewriting them into articles), as I don’t learn anything from doing it, so I am leaving them as I wrote them.

The AI Revolution

Now, let’s look at AI.

Let’s state the obvious: The discussion of AI quickly falls into a domain bordering on Sci-Fi. My thoughts here are only marginally shaped by scientific facts.

If you look at technological progress over the centuries through the lens of productivity, the improvement was tiny, barely noticeable. Our ability to do the same tasks maybe improved a few basis points a year for millennia. People figured out fire, learned how to use levers, created a wheel.

Then the pace of technological progress started to accelerate with the first and second industrial revolutions – steam engine, internal combustion engine, electricity, microprocessors, internet – productivity went from improving a few percent a century to a few percent a year. The beauty of productivity growth is that it is the magic that improves our lives. Technology allows us to spend fewer resources as we grow tomatoes and build new houses.

Serendipitously, as I was typing this I received an email from Edward Chancellor, who, in his article titled “The Trouble with Prosperity,” argues that productivity growth has slowed down to 1% from 3% since the 1970s. Edward thinks that the main causes were: (1) prosperity – the wealthier we get, the less hard we want to work, (2) the larger role governments play in the economy (Western governments have run deficits for nearly fifty years), and (3) regulation – it is up tenfold.

I hope AI can reverse this. Back to the future.

I have been battling between two conflicting frameworks here.

On one side, the technological progress we observed over the last hundred years has changed the nature of employment. We used to have one-third of the population working on farms, but today only a tiny percentage work there – a single combine replaces dozens of workers. 

I am quite sure that this transition was painful for many farmers, but it was positive for society as a whole. While a transition happens, we can only see the jobs we will lose but don’t yet know the jobs that will be created, because those occupations are yet to be invented. Just look how many jobs that we take for granted today did not exist 50 or 100 years ago – laparoscopic surgeon, data scientist, Chief Listening Officer – I kid you not, this one was suggested by ChatGPT (maybe it can see the future).

On another side, as Larry Summers pointed out, the biggest difference between AI and all other advancements is that “the wheel cannot create another wheel” but AI can build AI. There’s a good chance that AI will add exponentially to growth – the rate of productivity growth can accelerate. Again, I am in the Sci-Fi, not Sci-Fact, domain.

AI’s Impact on the Job Market

Most technological improvements of the past have attacked blue-collar jobs. The only exception I can think of is Xerox replacing typing pools – rooms of women were replaced by a single copy machine. AI is “collar-blind”, as it is coming for both blue- and white-collar jobs. I wanted to say that repetitive, pattern-recognition tasks are going to be swallowed by AI. But ChatGPT showed that AI can generate content, and thus creative jobs are not safe, either. At least it is going to increase productive output. In other words, powered by AI tools, a designer will be able to do the job of two or … pick a number. 

Add AI to robots and automation, and warehouses are only going to have humans turning the “on” switch. AI’s attack on blue-collar jobs doesn’t require much imagination; Amazon and others are fast moving that way. Robots will start stocking shelves at supermarkets at night. Amazon warehouse workers who are complaining that they have to walk ten miles a day will start complaining that Netflix ran out of new movies for them to watch, as bots will be doing all the walking and shipping while they are unemployed, flipping channels.

The composition of skills in the US will change over time – low-skilled labor will have a hard time finding a job. (Plumbers, electricians and other tactile skilled jobs have infinite security.) Today we are importing a lot of low-skilled labor through legal and illegal immigration to the US. A decade from now these people will have a hard time finding work. (That’s not a political observation but an economic one.)

Uber and the Threat of Robotaxis

As of now, only one company in our portfolio faces potential disruption by AI – Uber. The threat lies in the hypothetical scenario where millions of robotaxis suddenly appear on the roads. Uber’s value stems from its fragmented, two-sided network. If the car supply on the street becomes dominated by a few large conglomerates (like Tesla or rental car companies, for instance), Uber’s value proposition diminishes significantly.

However, robotaxis remain more science fiction than fact at this point. The technology isn’t quite there yet, functioning only in limited, discrete areas. While Tesla’s self-driving software has made significant strides recently, we’re far from the point where consumers would feel safe inputting their destination and then losing themselves in TikTok for the journey.

Self-driving technology relies heavily on sensors, which have limitations. For instance, my Tesla’s drive assist becomes inoperable in rain or snow, when cameras can’t clearly see the road. Moreover, there’s a trust factor: Many of us would gladly pay a premium for the assurance of a human driver getting us safely from point A to B, rather than risking an empty, self-driving car. The legal framework for self-driving vehicles is also nonexistent. In the event of an accident, who bears responsibility – this robotaxi or that one?

A standalone Tesla robotaxi service is unlikely to succeed. Tesla will struggle to maintain a deep pool of supply on their own, further reinforcing the value of Uber’s established network. Consumers aren’t specifically seeking robotaxis; they want reliable transportation that arrives quickly (key word) when summoned. 

As robotaxis gradually begin to appear on roads in the coming years, they’ll probably become another option on the Uber app (rather than operate independently), alongside the UberX, Comfort, or Luxury selections.

Societal Implications of AI

What will the economy and society look like when AI has taken over the office and bots are running everything else? Let’s say we institute some kind of basic income system and now people don’t have to work. What is that going to do to our society? Will people lose the meaning in their lives that today is fulfilled by working?

But maybe I am looking too far into the future. In the meantime, it is a given that AI will bring big change.

I tell my kids that it’s not what AI will bring to us that will matter, but what we’ll bring to AI. Our job is to be open-minded, embrace AI, and not see it as a threat but as a friend. It’s going to happen whether we like it and embrace it or not, so we may as well benefit from it.

From an investment perspective, we have a long list of companies from robotics to automation to energy that will benefit from AI. The issue is that today they are fully valued, like the rest of the market.

Key takeaways

• The AI revolution represents a potential leap in productivity growth, possibly reversing the slowdown observed since the 1970s and accelerating technological progress exponentially.
• Unlike previous technological advancements, the AI revolution is “collar-blind,” affecting both blue- and white-collar jobs, including creative professions, which could lead to significant changes in the job market and skill requirements.
• The impact of the AI revolution on specific industries, like transportation (e.g., Uber and potential robotaxis), remains uncertain due to technological, trust, and regulatory challenges.
• The AI revolution raises profound questions about the future of work, societal structure, and personal fulfillment, potentially necessitating new economic models like universal basic income.
• From an investment perspective, while many companies stand to benefit from the AI revolution, current market valuations make it challenging to find undervalued opportunities in this space.