From Houdini to ChatGPT - A History of AI

Tech pastor here. Listen, I am here today. This is a much shorter video. It's not a podcast video, if you will. But what I wanted to do is I get all these questions all the time. Every time I go teach, I always start. Every time I go teach someplace, I always start at the same place. The history of artificial intelligence. And I realize that most of us don't really understand the full fledged history of AI so I'm gonna spend two or three minutes today giving you the history of artificial intelligence where it is. Because I'm gonna tell you this. It. It's not new. It's been around for a minute. I'm Justin. I'm that tech pastor giving you the business for your business. Like and subscribe to this channel. Like and subscribe. Wherever you are, Follow along with us as we trip to build the body, all that type of stuff. Are you ready? So if you really want to get AI the plot twist and the plot about all of this is. It's not recent. We got to go all the way back to 1919. In 1919, Harry Houdini had a movie called the Master Mystery, where a villain in the movie was not a magician, but it was a robot. Now, we did not call it a robot at the time. We called it this machine. It was this mechanical man who was going around enforcing injustice on behalf of a corrupt company. In essence, he was a murdering robot. And because of this murdering robot, over a hundred years ago, the black mirror was kind of already flashing. He was trying to show us who Dini was. That technology as a performance, machines as a mystery. And behind it all, the question becomes, who? Who's controlling this machine? This is the first time we saw this. Nobody knew what it was. Nobody knew, like, you know, what this. We didn't call it a robot at the time. It was just Houdini was answering the question of, like, this decentralizing of the human brain. So Fast forward to 1943. McCullen picks came out with a paper of the neural network model. And they were asking the question, what if we could model a neuron in the brain using math? Now, this paper became a conceptual skeleton of every neural network we use today. No software yet. It's all theory. This became the bone. So we had Houdini doing the decentralization. We had McCullen, Pitt saying, what if a neuron could use math? And then in 1950, everything kind of just like Alan Turing asked the question, can machines think? That was too abstract. But he flipped it and he asked us a question. Can they IMITATE us well enough to fool us. And that's the Turing test. And that changed everything. That AI move from this philosophy to performance, from ideas to like, metrics. And then the first time we heard the phrasing artificial intelligence was in 1956 at Dartmouth College, at this convention, at this convene, if you will, the summer workshop. In 1956, they began to ask the question, if we can describe intelligence clearly enough, the machines can do it. And optimism for it takes off like a rocket. They get money, all this funding, but tech is fragile because that becomes the first AI winter. Hope goes a lot cold. Money dries up, and you know what happens? The we were trying to put a man on the moon. But interestingly enough, the first chatbot came out in 67. Now, at this time frame, computers were filling up rooms. Like, look at this picture, how big computers were at the time. Look at that. There were like 10 people trying to get this thing off a truck like this. What computers were at in the 1980s, back propagation changed the game. Now neural networks going all the way back to McCullen Picts can learn from their mistakes. Neural networks are trying and failing and calculating where they went wrong, adjusting their weights and trying again, like, you know, sitting down and working through a paper. Then the public meets AI for real in 1990s, not in labs, but we met it for real, publicly at scale, in something called Clippy. Y' all remember Clippy? It looks like you're writing a letter. Clippy was terrible, that little real clip. Well, let me tell you this. It was rules based. They weren't smart. But it made one thing clear. AI was not about intelligence. It was about placement, about context. People don't adopt tools because they're smart. People adopt tools because they're helpful. So then 2012 comes along, and a deep neural network destroys image recognition systems and massive data sets. And now AI can now start to see. So we go from the beginning of a robot decentralized in the human brain to Now Turing from McCollum, Pitts, Turing, Dartmouth to Clippy, back propagation. And then 2016, OpenAI begins. And 2017, the whole language game is rewritten because a transformer architecture drops, introducing attention mechanisms. Now the model doesn't just nudge word for word, it sees the whole passage at one time and highlights what matters. So these models didn't just speak, it listened in context. And this is what opened ChatGPT. Claude Gemini at the time was Bard. And every tool you're hearing about today. So today we're sitting in the space that Houdini predicted. Except the robot isn't wearing chains, metal plating, and murdering people. The robot is dressed like your phone app. An assistant that tracks your emails, that scores your tests, that transcribes your messages, lives on your phone, your laptop, and soon in all of your workflows. So here's a plot twist. It's not that all these models just got better. They just got more access. They got more data. So from the lab to your browser, from the code bases to your homes, the shift is all about usability. So let's review where we were. Houdini warned us. McCullum Pix gave us the skeleton. Turing gave us the standard. Dartmouth gave us the dream. Clippy gave us the proximity backdrop gave us the practice. Transformers gave us the language. And now you're giving it purpose. So the question is, no longer, can AI do this? It's should AI do it? Who benefits from it? And who gets left behind? So the story moved from papers to people. And that's why I wanted to go over this with you. It's not to impress you with a whole bunch of stats, but to equip you. That's where it's at. This is where we are now. Let's get to the future. What are you going to build? All right. I'm Justin. I'm that tech pastor, like. And subscribe for more.