Table of Contents

• An “ideal” world with lots of agents?
• The “myths”
• The Reality
• Our Pursuit of Happiness
• Coming back to life and work
• A Society with Agents

An “ideal” world with lots of agents?

Lately we’ve been witnessing many LLMs and Agents digging up cool research, finishing super complex coding tasks. Musk predicts they will make scientific discoveries by the end of the year. The world is heading toward agentic at full speed.

I cannot help but to picture a world like below.

AIs do all the work. They take care of us. We can day-dream all day long, and all of our problems will be solved automatically. How to educate kids? How to nail a project? How to plan a trip? They will all be taken care of.

The “myths”

Are we ready to sit back, relax, and enjoy our new chapter?

1. Much easier work.
2. More relaxed lifestyles.

How does that sound?

The Reality

I don’t think it’s going to work out like that. Not one bit.

I admit, we don’t have perfect LLMs/agents yet. But you would expect, that with all the automations already in place, our work and life should have gradually improved. But in my opinion? Not at all.

I work at a software company. You could argue that software development these days benefit the most from these ever advancing technologies. NASDAQ has surged over 10% from mid-2024 to mid-2025¹, largely driven by AI companies. Our daily routines are also automated to a great degree (coding, documentation, research).

But:

My job is actually getting harder. Nowadays, one could single-handedly execute a chunky projects in a week, which used to take quarters for a team. That sounds cool, but it also means lots of context switches between projects, lots of new changes in the system you need to adapt to. It also means our jobs actually fail more. Why? Imagine you working hard to increase the company revenue by 1%, and present it in front of everyone, just to find out another teammate increasing the revenue by 1.01%, with a proposal that is incompatible with yours. We all act faster than we talk. That is becoming the new norm.

Second, job is spilling into life. We used to have clear boundaries, and never needed to worry about work during “life” time. I mean, at the end of the day, “work” is performing a specific type of tasks during work hours, right? But what if the “task” keeps on changing? Old tasks keep on being automated: for example, a pretty web interfaces that used require one afternoon now gets done in one minute. The bars are higher too – and we need new skills, working on new tasks to reach these higher bars. When do we learn these new tasks? “Life” is an obvious solution: learning about new tools, keeping up with the tech trend, reading research papers during become our new “hobby” when we “take breaks”.

So, is this just a temporary state? I have two natural hypothesis.

Hypothesis one: maybe our AIs are not good enough yet. They cannot “fully automate” our jobs. Therefore we are still involved. Life is only miserable in the short term. Once we build an AI that is powerful enough, we’ll enter the eutopia, with zero worries on work.

Or …

Hypothesis two: maybe our expectation is wrong? Maybe jobs will keep getting harder regardlessly? And our lives will be eaten up by our jobs?

Before answering that question, I’d like to reflect a bit on how AI was designed to improve our lives.

Our Pursuit of Happiness

Building on one of my favorite books, Beginning of Infinity²: one way to look at our physical world, is to imagine that in the vast universe, there is a small circle, where humans are able to reach. Here by reach, I instead mean being able to achieve a desired outcome. For example, in the beginning, we were able to pick fruits, throw stones at animals. Lately, we are able to watch stars from faraway galaxies, send humans to the outerspace, edit our DNAs, and live chat with our friends from the other side of our planet. The unlimited possibilities of our reach forms this universe. Note that this universe is different from the actual physical universe we live in, though they correlate. It not only includes stars that we can reach in the physical world, but also includes the theories we can prove, and algorithms that we develop. It is a set of anything that humans are capable of doing.

In this universe, we may picture a “North Star”, which is the pure satisfaction. The pursuit of happiness, can be seen as finding a path toward that North Star. By reaching to that North star, one would feel the ultimate satisfaction. It is not the kind of short-lived satisfaction that Cocaine leads to, but a long-lasting one that leads to the ultimate satisfaction from birth, for a life time. Ideally, we want to get everyone onto that North Star. But we probably haven’t discovered that North star yet. Even if someone truly did discover, different people might have a different path toward that North Star.

The World of Reasoning

For thousands of years, humankind was far away from the North Star. We struggled with starvation, diseases, and miseries. In each generation, we tried hard to expand our reach, but overall humans were stuck far away from satisfaction. Until a couple hundred years ago, the discovery of science pushed humankind ever closer to it.

The discovery of science is essentially allowing our tentacles to expand from a different universe, reaching to our physical universe at an entirely different angle. From theories like Darwinism, physics, chemistry, biology, we are quickly finding our ways to solve problems that we never would have reached. We used to struggle feeding the plants enough energy, and we figured out ways to extract ammonia from crude oil. We used to die miserably from widespread plagues, which were cured by antibiotics. We used to cry for being far away from our relatives, which was solved by cameras connected with wifi signals. Science even allows us to predict things that we never would physically perceive, like Black holes, where we’ll never travel to, and parallel universes, which we would never see with our eyes.

A couple of simple steps in the world of reasoning can reach far in our physical world. Think about knowing when to plant seeds. It used to take generations of measurement, and praying to god, to be a good farmer. But now, by knowing that one year consists of ~365.25 days, I think anyone would be a better than average farmer if he/she time travels back to the ancient time. Discovering a new theory is hard, but applying it by a couple of steps, we now have fridges, GPS and many cool gadgets that solve huge problems. As David Deutsch writes: “The reach of explanations is potentially infinite.”³. This universe can ultimately solve a large amount of problems in our physical world.

AI: Expansion of the Worlds, On Steroid

When talking about AI, LLMs and Agents are two big buzz words. LLMs basically learn to mock humans in uttering words; whereas Agents mock humans in taking actions.

These two most popular AI techniques are putting the expansion from both our universes on steroids. In the world of reasoning, LLMs offloads the brain-power needed from humans to machines. In the physical world, Agents, on the otherhand, automates our actions with motors. Automation is powerful. It means the steps no longer feeds on humans, which feeds on animals and plants (and most importantly, coffee), which feeds solely on solar energy. It means all the steps can directly feed on electricity, which is rich and available from so many different sources. It also means intelligent work, the once privileged human activity, has detached from human’s emotional suffering, like many would experience in today’s office politics, and from exertions in various dangerous jobs.

We, humans, are approaching our north star, the pure satisfactions, at an increasing pace.

Or are we, really?

Why am I not feeling it?

Well, that all comes down to: what is satisfaction in the first place? Where exactly is our North Star? Or, if we really know what is our North Star, would it mean it will absolutely be within our reach, ultimately?

Approaching Satisfaction

Now let’s take a look from the goal backward. One pattern we experience from human history is that whenever one problem is solved, another one emerges. We solve hunger, by growing lots of plants, and then more of us die from plagues. We then invented antibiotics, to eliminate plagues, then more of us suffer from poverty. We then automate things, to get everyone rich, then more of us suffer from mental stress. Satisfaction from the previous period seems like the starting point for the pursuit of the next level of satisfaction. So where exactly is the true satisfaction? Maybe the true satisfaction does not exist? Then, are we striving for nothing?

Or, maybe true satisfaction does exist. And we may already know what it is. But it could still mean we can never get there. For example, let’s say the true satisfaction is knowing every digit of Pi (well, the real satisfaction is definitely more complicated than this: for example, avoiding death at every moment). It takes some work to know the first three digits, and then the fourth digits, the fifth and so on. Every step gets us closer to this known goal. However, Pi has an infinite number of unrepeated digits. Now, even if computers/AI helps us to discover new digits at whatever speed, he still makes finite progress during his lifetime. So, upon his death, if we zoom out: what’s the difference between discovering 100 digits v.s. 100000? He still seem to be stuck in the beginning anyways.

Coming back to life and work

So, where are we headed? I have my own speculations, heavily influenced by Harari’s “Sapiens”⁴. As Harari observes: “The Agricultural Revolution was history’s biggest fraud. The species that benefited from it were not humans, but rather wheat, rice and potatoes.”⁵ Perhaps we’re witnessing a similar pattern with AI.

Era	Work Schedule	Job Requirement	Training Needed to Solve Problems	Satisfaction (1–10)
Hunter-Gatherer	4 hours a day	Hunt, Gather	1 year	5
Farming	All day every day	Plow, plant seeds, etc.	10 years	5
Industrial Revolution	All day every day	Many different jobs, more philosophers :D	20–30 years	5
AI Era (speculative)	All day every day	Even more fine-grained jobs (speculation)	20–30 years (speculation)	5
Superintelligence (speculative)	All day every day	?	??	5

• As hunter-gatherers, we work limited hours per day, which required minimum training.
• As Farmers, we found us some tough jobs, requiring us to work all day every day. But the nice thing is it increased our populations drastically.
• As workers, we found us even more tough jobs, working day and night, attending schools for half of our adult lives. This way, we start to fund more interests, more philosophers. More niche jobs start to emerge (e.g. Youtubers selling nail cutters).

The same thing that inherit throughout the generations is probably our low satisfaction to lives. We always foresee a better life. The insatiable desire drives our pursuit toward “better” lives. We work hard every day, maybe never seeing the “great life” by ourselves, as if we are working selfishlessly for our descendents.

In AI era, I personally expect us to still work all day every day. Tools will help us create a even richer set of fine-grained jobs that we can’t imagine from now. Each of us is objectively getting closer to our North Star: better health, better income, even better mental health. But I also expect all of us to be as dissatisfied as our ancesters. I think of this as human nature.

Maybe we will ultimately get to an era of “Superintelligence”, where machines surpass us in all skills. There, maybe each of us is still trying to figure out our unique positions in the society, in our own pursuit of happiness. As a result, we still work all day, every day. On what? I don’t know. All that I know is, probably we are almost as unhappy as we are now.

A Society full of Agents

I don’t know for sure about the real “Super-intelligence future”. But I think it is definitely a fascinating topic that I cannot stop thinking.

The three possibilities of an “agent” society as I see:

Scenario 1: Lazy Human	Scenario 2: Working as Colleagues	Scenario 3: Iron Man Fleet
AI does all the work while humans relax and enjoy leisure time	Humans and AI work together as partners, each contributing their strengths	Humans control multiple AI agents like Tony Stark commanding his Iron Man suits

Scenario 1: I don’t think the “lazy human” future that I pictured would emerge, for everyone. That is just the human nature as I elaborated above.

Scenario 2: Lately, Harari had comments that agents will work as our colleagues. He described them as similar to “immigrants”, who will take away some of our jobs, maybe having different political stances than ours, but will greatly improve our lives at the same time. As Harari puts it: “AI will be like having millions of immigrants entering the job market… they will compete with us for jobs, but they will also create new opportunities and make our societies richer.”⁶

Scenario 3: I personally, though, am leaning more toward a future of a society of Iron-men. Each of us have abilities greatly enhanced by AI (like Jarvis in the movie), as well as the robotic arms, and all the fancy deadly gadgets that we carry in our pants. But when you think of Iron man, you don’t think of a rich guy lying on the beach, with robots changing his diapers for him. Iron man has his own troubles, his own struggles. He works even harder than before he suited up, to bring peace to himself and to others. He faces more severe challenges, changing context every minute, adapting to Jarvis, while upgrading Jarvis as he goes.

I concede: there is not an “average human” living on the planet. These scenarios aren’t mutually exclusive either. I think changing diaper could be a useful feature for Jarvis too. So, what is our choice? Or, does our choice really matter that much in a world of “super-intelligence”?

---

References