In parts one and two of “The Battle for Our Minds,” I tried to explain how the internet transformed attention into the currency of modern age. Algorithms have shaped what we see, think, and ultimately, how we behave. But those algorithms, as powerful as they are, follow rules explicitly programmed by humans. Now, we’re entering a new phase—one dominated by artificial intelligence systems that don’t merely follow programmed rules but learn, adapt, and “think” independently. I know it can be hard to comprehend that, but it’s true. I’m not an AI expert. But I have spent hundreds of hours listening to and interviewing people who are.
Artificial intelligence represents a significant leap beyond traditional algorithms. Actually, that’s underselling it. They are a total transformation of how computer systems work. Instead of relying on logic that’s coded by humans, modern AI utilizes neural networks.Â
“The whole idea was to have a learning device that learns like the brain… by changing connection strengths.” – Geoffrey Hinton in 2019 Wired Interview
Geoffrey Hinton, widely regarded as the Father of AI, won the Nobel Prize in Physics last year for his contributions to neural networks, which he pioneered in the 1970s. He most recently led Google’s AI efforts until quitting last year to free himself to speak about the perils of something he had helped create.
“I suddenly realized that maybe these things are actually much smarter than us. I want people to be frightened of that.” — Geoffrey Hinton, 2023 (after resigning from Google)
Neural networks are inspired by the way the brain learns and processes information, which is an emerging field of science in itself. Our brains are made up of 86 billion neurons that pass signals through synapses. When we take in information through our sense organs (such as the eyes and ears), our brain strengthens or weakens certain connections based on our experience, allowing us to process that information more efficiently in the future.
Neural networks have artificial neurons that pass signals through weighted connections. Like our brains, neural networks adjust their connection strengths (called “weights”) as they “learn” (by being fed data). Each neuron receives inputs, combines them, and decides whether to activate, much like biological neurons fire in response to incoming signals. Over time, by adjusting the weights, the network improves at recognizing patterns such as words, images, and faces, much like the brain gradually learns from the world around us.
So neural networks don’t follow instructions any more than a 5-year old does. They build their own complex logic based on the information they’re fed. The difference between AI and a 5-year-old is that AI has access to all the information on the internet, powered by 5 million GPUs.
AI researchers often refer to a neural network as a “black box” because they have no idea how the network processes the information. AI systems are self-teaching, called recursive learning. They use their own outputs to train the next versions of the model, kind of like families who pass down generational trauma. As neural networks evolve, much like humans, they wander from the planned path of their creators. As more and more systems adopt AIs powered by neural networks, our ability to control the outputs diminishes.
It can be hard to comprehend, but we have no idea how our brains think. We believe we control our thoughts, but neuroscience suggests the opposite: that our thoughts control us. Studies going back to the 1980s show that your brain kicks off actions before you’re even aware of deciding. Libet’s famous experiments found neural activity ramping up a full half-second before people felt the urge to move. More recent research using brain–machine interfaces shows the same thing — intention signals fire before conscious thought. It’s like the brain is running the show backstage, and your sense (illusion?) of “free will” is just the narrator arriving late to the scene. How does that concept land with your identity?
I’ve been practicing meditation for about 10 years, and for the last year, I’ve been focusing on Zen meditations. One practice is to observe where your thoughts are forming and try to catch them as they emerge. It doesn’t happen often; most of the time, my thoughts sneak up behind me, and I don’t notice them until I am fully immersed. Oh, that noticing? That’s called awareness, or consciousness. Or at least one definition of it.
There’s a lot of debate around consciousness; the word itself has several meanings. Often it’s used interchangeably with “awareness.” As in “I am conscious of my behavior.” In mindfulness circles its also used to describe a broader sense. The fact that I can watch my thoughts, knowing I am watching my thoughts, suggests a higher level of consciousness than, say, a dog might possess (though we can’t possibly know that for sure.)
“Why and how do physical processes in the brain give rise to subjective experience?” – David Chalmers
David Chalmers asked a question in 1995 that has been labeled “The Hard Problem of Consciousness.” We can explain the basics of the brain, like how vision works, or how we sense pain. But we have no idea why the configuration of our system leads to a sense of “experience.” Why do we develop a sense of self? Why isn’t the brain and sense organs just a complex information processing machine, like a computer?
Many researchers and philosophers have been extending that question to artificial intelligence. What happens when a machine behaves like it’s conscious, which ChatGPT already does (that happened fast!) It speaks like us, reacts with apparent emotions, it passes almost every test we throw at it. How do we know it’s not conscious? Early versions claimed it was. But we don’t have any scientific way to determine whether it has an inner life.
Figuring out if an AI is conscious is a lot like sitting across from a psychopath who’s mastered the art of faking empathy. They smile, mirror your expressions, and say all the right things. But inside? You have no idea. You’re guessing based on behavior. That’s exactly where we are with AI. It spins out poems about grief, telling you it’s afraid of death… but we have no idea if anything’s actually happening behind the curtain. Just like the psychopath, it might be running scripts.
So if and when AI gains consciousness, we likely will miss it entirely. And that’s the scary part. We could be fooled into thinking a machine has a mind when it doesn’t — or miss the moment it actually does. Now to be clear, I am NOT suggesting AI is conscious today. But I am saying that anyone who says it can NEVER gain consciousness is deluded. Unless of course they’ve solved the Hard Problem (they haven’t or they’d be the one with the Nobel Prize.)
So what’s all this mean? Humans have designed systems that are becoming increasingly powerful, yet they don’t understand how they work. Those systems are growing increasingly capable at an incredible pace. Just last year, people were complaining because AI could not draw a human hand without six fingers. Now it’s writing code and creating photos so realistic that it’s nearly impossible to tell. And just last week, OpenAI unleashed Agent: an AI program that can run other programs. It has its own browser and can navigate most online systems. OpenAI has put restrictions on it for now, called guardrails, that prevent it from doing financial transactions and other critical processes. But as the system trains itself, that’s coming next.
When I was a kid, I had a Casio calculator watch (anyone out there surprised?) My math teacher told me relying on calculators would limit my abilities in the future. Now I can barely remember my multiplication tables; my iPhone calculator (or Siri) does any calculation more challenging than what a third-grader could have done in the 1970s.
AI is quickly becoming the calculator for thinking. AI is already better than humans in many areas. I use the latest reasoning models to challenge my strategic thinking, and I am constantly shocked when it sees things I missed. Within a few years, like with math and calculators, we will cede much of our thinking to computers because it’s fast and convenient. And we are inherently lazy. (Quick, what’s 17×15?)
Our behaviors are already highly influenced by the social media algorithms. Since we don’t understand how our brains think, but our thoughts drive our actions, what happens when AI starts thinking for us?
Next edition of Dropping In, I will discuss the unsettling ways AI already manipulates our perception of reality. Make sure to subscribe.
