Sejnowski refers to this moment in history as the “Wright Brothers Phase,” and compares it to the first powered flight in 1903, which spanned a few hundred yards and reached a height of just 10 feet. At the time, no one – not even the Wright brothers themselves – understood how significant this achievement was, or the ways in which aviation would one day change the world.
“I wouldn't believe anything anyone said about predicting the future because I don't think we have enough imagination to know where things are going,” Sejnowski advises. “Whenever you have new technology, it manifests in ways you can't imagine.”
These ideas are of particular interest coming from Sejnowski, who in the 1980s was part of a small group of pioneering researchers who founded deep learning and neural networks, the subset of artificial intelligence that powers chatbots today. Sejnowski, along with Geoffrey Hinton (often referred to as the “Godfather of AI”), questioned the “logic and code”-based AI that was most prevalent at the time, and developed their own version fueled by data and models. After the human brain.
Today's big language models like ChatGPT are a type of neural network. If you look “under the hood,” Sejnowski explains, what you find are simple modules that look a lot like neurons in the brain, linked together with varying weights, much like synapses between neurons. Neurons have synaptic plasticity, which means that when you learn, you change the “weights” in your brain. Big language models are trained on data in much the same way.
“Modern AI is based on basic principles of neuroscience,” says Sejnowski. Conversely, as the fields of artificial intelligence and neuroscience continue to converge, advances in large-scale language models such as the use of transformers — a type of neural network that learns context — are impacting the way neuroscientists think about the brain. Currently, there are still many brain features that have not been integrated into these adapters. ChatGPT and other large language models can't have targets or long-term memory yet, but Sejnowksi says they will, and that's where we're headed.
“AI will make you smarter and enhance your cognitive power,” Sejnowski predicts. “It won't take a job out of you, but it will change your job. Your job may be completely different one day, but it will be more interesting. I'm sure of it.”
