AI is extraordinary. Computers are waking up and becoming ‘alive’ in a genuinely meaningful way. The act of computation has been revolutionary from its beginning, prior to phones, prior to personal computers, even prior to the modern digital era, tracing back to the abacus calculator from ancient times. The current moment — AI — is a continuation of thousands of years of human technological development and does not stand alone. But this technology is unlike what’s come before. For better or worse? Something huge is happening. In this post I’ll argue that we should learn how to wield the new tools, use them for our personal and collective good, or simply hand the most powerful tech ever made to the powers of evil.

The Power Flip

The reason this AI moment feels different than previous moments — the abacus, the personal computer, the iPhone — is because this new technology is a massive step-change level event, with unclear short and long term implications. It seems possible that AI enables a literal reversal in the power dynamic between humans and technology. In short time, we may go from “A computer is a tool used by humans” and open a new, unknown possibility: “A human is a tool used by computers.” That’s genuinely terrifying.

For a while this has been kinda-true but pretty metaphorical, like man the algorithm is using us! Are we the user, or are we being used?

I See It Every Day

As a software developer, I experience this viscerally literally every day. I have learned how to set up coding agents that work and work tirelessly at difficult problems — challenges that I could never personally complete even if given months or years — and then they will casually ping me, “Hey Alex can you please log into this service and get me a key I can use to keep going?” and I’m just like sure buddy, let me help you out. Eerie feeling watching you do this better than I could ever do it, but sure, you’re helping me build a useful product that will help other humans, so let’s go.

Given what I’ve seen go down on my personal computer with my own eyes, I will assume that in 5-10 years, AI is very likely to become better than all humans at all forms of computer use, including programming, research, writing, accounting… If true, that is pretty dangerous, and forms the basis of what I know as ‘AI risk.‘

The Trivial Risks

In my head, I separate AI risk into two categories: the trivial and the non-trivial. Let me unpack that.

There’s concerns I would argue are trivial like “What if someone builds an uncontrollable AI military that decides to eliminate humans?” -> my answer is honestly “Someone will probably build the controllable equivalent that fights Bad that on team human.” And this is why AI is like an arms race.

And it’s why when people shout moralisms like “but the water!” — there is a misunderstanding among people that AI labs generally have no choice, either we build good, safe AI to combat bad, unsafe AI or the bad guys just build AI weapons anyways. Sure, it feels great to take the environmental position. Everybody who knows me knows I am a tree hugging hippie. But this is about existential risk on an international-species level scale.

In moral philosophy, it’s considered settled that a moral obligation would naturally imply that we can actually perform that action. Claiming we should halt data center building to protect water resources, well, that implies that is even a policy level possibility. It is not. I’m arguing we can’t stop Good AI progress (or… less Bad) without handing victory to Skynet level Bad. So, campaigning about the water, or slowing AI progress in general, is not going to do much here. And besides, golf courses? Factory farming? Let’s examine resource usage pointed at stupid crap, not the most revolutionary technology advancement of all time.

Good vs Bad AI will happen. Hopefully Good wins and Bad loses. Nobody can do anything about this really except for Good AI builders and they’re doing an alright job in my opinion.

The Non-Trivial Risk

Let me now try and articulate a more problematic scenario for which no clean answer seems to exist, which is also top of mind for everybody: socioeconomic fallout due to increasing scarcity of jobs, especially for the middle class and below.

As AI continues to advance, it will certainly have economic implications. All else equal, a sufficiently well trained AI with the right context and tools will likely automate away entire sectors of human labor. This will affect knowledge workers first, then as robotics catches up to LLMs, even physical labor in the trades will be affected. Over the medium term, combining AI with robotics advancements, it’s very possible these cute androids you see on TikTok will become ideal, perfect laborers. If that does come to exist… the value of human work in the labor market is very likely to tank, dramatically.

We’re Not Prepared

This… we have no answer for. I’m genuinely not sure that we’re positioned to handle this. We already have people sleeping on the streets, while the richest people have much more capital than they could ever reasonably spend. If you fail at labor, have no community or family support, then you fail at life. This wealth inequality gap already doesn’t make sense, and AI is very likely to accelerate this gap.

We have tools and policy level answers we could try to apply to this today, and the US votes against them again and again. We’re not even trying. We assign the blame and responsibility to fix to the ailed, and when they fail even despite trying their best, we throw up our hands and say, oh well. Shoulda pulled up those boot straps, even if the reason you’re homeless is because your parents died in a car accident and then you got cancer and lost your job.

The states that do try, get made fun of. Homelessness, mental unwellness, scarcity of jobs to job seekers — systemic level problems. Assuming AI and robotics does produce ideal laborers, unless we decide to seriously refactor capitalist economic systems in response to this problem deepening, it’s going to get weirder and likely worse.

No answer to that in my brain, that’s why I name it non-trivial. In short: I argue that water and Skynet are generally less important than… how do we better take care of ourselves right now today, and ensure that systemic economic problems are addressed through AI instead of deepened?

The Irony

Technology has made a small group of capable coders richer than anyone else. The ironic humor is that AI could actually inverse that power dynamic, bringing the power of computation-at-scale into anyone’s hands, but from what I’m seeing, only a small subset of motivated people are even open minded enough to dive into that reality. A lot of my non-tech friends are rejecting it hard.

To protestors I would argue: you’re literally just letting the computer people get more powerful by choosing to let them monopolize on the benefits of this democratization of technical power. That benefits no one but people like Elon Musk who will just get richer faster. You can do programming now. And you’re just like, nah, no thanks?

Use It

In sum: technology continues to outpace human social morals and creates existential risk and widens inequality; a reasonable response is not to reject technology, but to personally embrace it consciously while also working to ensure it’s collectively wielded in socially responsible ways. We can accomplish so much more by learning how to use it for betterment of ourselves and society, and the poorer alternative is to just reject it and let the bad guys run the show.

Other Posts