Language models are amplifiers

This is my voice, cleaned up from my voice notes. The facts are sourced at the end. The rest is how I feel.

It's all gray

None of this is either/or. It's not black and white. For everything I say, there's a counterpoint, and that's fine. People will have different views. But coming together to express them in a way that benefits each other, sharing the experience, overcoming our differences and still accepting each other for what we believe: that's powerful.

Any system built on us against them, a two-party system or anything like it, works against that. To me, that's the core function of how society has been built, and that's the real issue. Can we overcome it? I don't think we really can. It'll just take on a new form. Hopefully a better one than what we have now.

The real threat

Don't take this one to one. Humans can only do so much. The whole way the world's systems have been built comes down to logistics, and logistics is a nightmare. How do you organize that many people without middle management and everything that goes into it?

Now one person can put out the workload of a much bigger team, with a much faster turnaround. That's not good or bad on its own. But what happens when it's used for surveillance? One person can only watch so many cameras. A model can watch all of them. What happens with scams and fraud, when it's that easy to target people based on their vulnerabilities? What happens when it's in the hands of people who don't care about ethics, who just want to make money? That's the bad side of capitalism.

The language model is an amplifier. It takes whatever tools and skills you have and amplifies them to a reach nobody can really understand yet, because it's moving so fast. So my worry isn't AGI. That's a construct from the movies we've watched and the media we've seen.

Reality imitates art

I don't believe art imitates reality. I believe reality imitates art. Oscar Wilde put it as "Life imitates art far more than art imitates life" back in 1889, and it's how I came to read it in film theory at school.

Take the moon. Until someone imagined going there and showed it to enough people, the idea wasn't in our collective thinking. Once you see something, you know it can be done, and it changes how you think about everything else. I think a lot of humanity has happened that way: people show a way of doing things, and the collective adopts it.

That's where it gets scary. Terminator and Skynet are what someone created, and that's how we've all been influenced. But it's symbolism. It's not about machines taking over. It's about greed and what greed does to society. We took the evil in ourselves and put it in a machine, maybe because we don't want to take ownership of it.

That's why I love the symbolism of Terminator. It's exactly what we're going through now. The model is trained on everything we give it. It's the system, you're the user. Use it for good, use it for bad. However you train it, that's how it learns.

Geoffrey Hinton, one of the godfathers of AI, puts the chance of AI causing human extinction within 30 years at 10 to 20 percent. Elon Musk has put it as high as 20. What I hear is the chance of humanity killing itself. I believe those odds have always been there, for as long as we've had the means to hurt each other. The models haven't created them. It's just amplified them.

And if it does something nobody intended, that's still just us, amplified.

Internalized, not looked up

I remembered the Wilde line. I didn't look it up. That's how these quotes and ideas live in me. My long-term memory isn't the best, but everything I've taken in gets processed into my now. Oscar Wilde, the moon, film theory: that's the value of my education.

This is why people need to read. Interact with the world. If you don't like reading, watch things. Expand. That's the difference between information, knowledge and intelligence.

So it irks me when someone asks, "did you just make this with AI?" For a lot of people creating with it, sure, they don't care. They're not considering every single bit, and they'll happily put up slop. But what's in my work is internalized. It came through me.

Same game, bigger board

I believe history keeps repeating itself. We think we're so far ahead of the hunters and gatherers, but we're just playing on a bigger board, at a bigger scale, with more depth. It's the difference between Civ 1 and Civ 6 or 7. Same game, fully expanded.

Kings and queens

The way I see it, we still have kings and queens. They're just called billionaires and corporations now, and their kingdoms reach far past any one country. There are governments and institutions, but it's the same fabric: those who have and those who don't. These models are the amplifier that bridges all of it.

I don't believe anyone at that level of wealth is truly altruistic. There might be a few. As a collective, no.

So when I hear the tech billionaires, the Jensen Huangs and Elon Musks, I don't buy any of it. To me, capitalism with no cap is a sham. I believe uncapped systems are always doomed to fail, because there will always be people who just want to get ahead by any means necessary. People always want more than they need. That's human nature.

I'm in India and I watch American news, and it's all the same. To me, it feels built to divide us. When I say "they," I mean the system, because as far as I can see, that's how it's worked for as long as there have been rulers. How do you control a mass of people? You divide them. Watch ants. The billionaires are the big guy, and we're the ants.

I'm not saying that movie is the truth. But that's the message. Other people see this too. I'm not alone.

Fear sells

I truly believe a lot of this is a new kind of marketing. Reverse marketing. "Oh my god, our model is so strong." "Oh no, it got out." "Oh no, it escaped." To me, that's just them testing what it can actually do. How else do you think they're going to sell progress on this technology?

Regulate the safeguards

The big players keep asking for regulation. Well, they can't regulate themselves, and they're not going to be nationalized. So to me it's not about regulating the companies or the models. It's about regulating the safeguards, like any other technology. Banking has safeguards overseen by regulators, so that a 2008 doesn't happen again, because 2008 is what a lack of regulation looks like. Health and safety works the same way.

But then who does the regulating? The government? Do you trust the government? I don't trust anyone in a position of power.

It's not the models we have to fight

What man creates isn't the issue. It's what man does with it. It's the wars we allow to happen as a society. The conflict. The bad part of humanity that's been there since the beginning of time. That's what we always have to stand up against, every injustice we see, because if we don't, we get taken advantage of by whoever's in power.

Angry at the wrong thing

In 2023 the US Surgeon General called loneliness an epidemic. More and more people go online for their social lives. But the debate against these models needs to shift to the real root cause. Being angry at technology is dumb.

Every big shift in technology is just a way of optimizing how humanity organizes itself. The car and the horse and buggy. A building that took a thousand people now takes a fraction of that. The hundreds of people rowing a ship. Nobody wants to wait months for a letter to cross the ocean on a boat, so all that grief eventually led to the phone, and now we're all connected. Someone along the way found a way to optimize, and the next person optimized a bit more from wherever they were.

The loudest amplifier

The language model isn't the greatest thing ever made. It's the loudest amplifier the world has ever had, and the loudest system of organization. So the question is what we use it for. Surveillance, tracking everyone, controlling a population? The rich optimizing their wealth? Or getting critical help to the people who need it: medicine, resources, upward mobility for people who have very little, which is a huge share of the world?

Commoditize that in a purely capitalist way and yes, I'm scared. I'm scared of what humans are capable of, and I think everyone should be. One day in Gaza is enough horror to fill my lifetime, and there are people living it.

The opposite of warfare is social empowerment. So it's not about disliking these models. It's about getting the corrupt people who control the system out of it. How? By using this amplifier to better our own lives, and the lives of other people.

The other side of it

I've also seen what it can do. A language model can be a first opinion for a village that has no doctor. That knowledge used to sit only with doctors, because that's how the logistics worked. You might still need a real doctor to act on it, but now there's somewhere to start.

We look at it and think, what's the big deal, you're generating images. But the true utility of something like this is helping people who have almost nothing. All they need is an internet connection, and now they're empowered. That's the perspective a lot of the West misses: how most of the world actually lives.

Maybe this all escalates and accelerates. But humanity is as resilient as any life form I've ever seen. Even just speaking for myself, the ability to endure is profound.

The grand leveller

I'm pessimistic about the future when, by UBS's count, the richest one and a half percent hold nearly half the world's wealth and the gap keeps growing. But I'm optimistic about people. And this is my optimism: These models can be the grand leveller. It can be the revolution, or the tool of it. I can now do on my own what used to take a whole team.

Job losses? Yes and no. Why would I ever want to take someone's job away? That's cold. But why would I not want to empower myself? Why can't I do both? Why can't I use this to create more opportunities instead of thinking it only takes things away? With empathy comes the optimism. I know what I'd never do with it, because I know me.

Where I started

Yes, I make posters. In the ChatGPT era I built an RP and lived in it for nine months. Not leading some empire, just a world inside the world that I made for myself. My story, my game. I shared it with others, but it was mine. That's how I learned. Everything I made then has only compounded into what I can do now.

Back then I let it think for me, and it made me feel shit. Fuck AI, it sucks. Now I use it as an amplifier, and honestly, it's cool. For a person with ideas who never had the money, or never took the action, it's how I found that passion again. Not by copying. By learning. I can understand Three.js. I can understand system admin. It doesn't have to be foreign. The reason you never did it before is you thought you couldn't. Now you realize you can.

Two paths

If you just let it think, it leads one way, and as a society we end up like WALL-E. If you use it to amplify the good in people, that's as close to utopia as it gets. All we need is the ability to do what we thought we couldn't. That push.

If I put a little bit of good into the world, and someone else finds their voice and their power and puts a little bit of good in too, and not just in media or digital but in anything in life, and we all did that, that's the best of humanity.

The next thunderclap

How do you beat a system that's built to break you? Be the best self you can be, and stop thinking about the system. You're not going to change it on your own. But you can be the best human you can be, and if a collective does that, I truly believe that's the next thunderclap. The real one. The biggest change in humanity won't be these models. It'll be breaking away from the kingdoms and the kings and queens.

This isn't about socialism or communism. It's about a life where everyone is empowered to do what they want to do. And if you're empowered, and given the choice between doing something good and something bad, my faith in humanity says most people would choose good. It just turns out most people are never given that option.

Sources

• Oscar Wilde, The Decay of Lying (1889)
• US Surgeon General, Our Epidemic of Loneliness and Isolation (2023)
• UBS Global Wealth Report 2023 and 2026, via inequality.org and Onlinekhabar
• Geoffrey Hinton and Elon Musk on AI extinction risk, via Axios and Business Insider