Don't fear, steer.
That's the advice Columbia University Professor Hod Lipson, Director of the Creative Machines Lab, delivered during "The Next AI" presentation at BAID @TheHub Speaker Series, moderated by Columbia Business School Professor Oded Netzer.
His recommendation comes amid the "most dystopic" point yet – the transition between the third and fourth waves of AI.
This "period of chaos" is characterized by the explosive growth of artificial intelligence in the last couple of years, which Professor Lipson likens to a child maturing from infancy to adulthood almost instantaneously.
"I'm here today to turn some of that stress into a feeling of opportunity. Whatever you want to do in your lives and careers, you can do it better if you understand AI," said Professor Lipson. "If you don't do it, somebody else will."
He admitted that part of the anxiety around AI revolves around its ubiquity and imperceptibility.
"It's like air," he declared. And while we may be just getting used to AI predicting the weather, stock market, and even our tastes and opinions, we're still apprehensive about what is coming next – and how fast.
“Whatever you want to do in your lives and careers, you can do it better if you understand AI.”
Professor Lipson illustrated the "spine-chilling" growth of AI, which is surpassing even computer power, fueled not only by an astronomical increase in data but by the ability of AI systems to learn from each other, a capacity that – unlike our brains – doubles every three and a half months.
The potential of AI to learn from other AIs by communicating, querying, searching, reading, watching, and reiterating data makes our interaction with them – through generative tools like ChatGPT – a tiny drop in the bucket.
This revolution almost makes one nostalgic about the first wave of AI, starting in the 1950s with rule-based AI, where humans programmed the machine, which didn't improve with time, causing "great disappointment."
Fast forward to the 1990s, the second wave of AI, when the first generation of machine learning dealt with tabulated and quantitative data. Today, we still use this technology to predict the stock market, autofill our emails, and receive streaming suggestions.
In the third wave, machines can differentiate between a cat and a dog, which fuels numerous applications, from driverless cars to agriculture, with the potential of developing phone applications that detect skin cancer "better than a team of skin cancer specialists at Stanford."
Systems today use a combination of these three waves, with driverless cars using rules-based AI for traffic, analytics for congestion predictions, and object recognition to detect pedestrians. But there are certain things that all these three waves could not do, like having a conversation – something the fourth wave is doing.
And here lies the potential for a successful partnership between humans and AI. Human intuition is excellent at designing chairs and hammers, but AI is better at designing antennas or proteins. "And this is important because the fate of the human race – from solving Alzheimer's to Parkinson's – depends on our ability to design proteins."
We just have to let AI "fill in the blank." You can generate anything you can represent as a language by removing pieces and teaching AI to fill in the blanks. And that's terrifying, Professor Lipson admits. "There is something magical about creativity. It's the one thing we have."
But once the AI community understood that the larger the blank, the greater the potential for creativity, the community realized we could create new things automatically. "We went from stone tools to the microchip in 50,000 years, but the next invention will happen in 50 years."
Enter the last two waves of AI involving Professor Lipson's specialty: robotics. "Physical robotics is hard because it takes energy, and mistakes in the real world are expensive. We do things with our hands all day, and we don't think much about it, but it's very difficult."
AI can drive a car, but we will need a human to repair it. And that's going to be the case for a long time. "This is why a plumber makes twice as much here in Manhattan as a software developer," states Professor Lipson. "If your hands are dirty, your job is safe."
It's almost impossible for AI to become a plumber, electrician, hairdresser, nurse, or dentist, as these activities require a prohibitive amount of power: the human brain uses just 20 watts of energy. In contrast, even the most basic AI requires two kilowatts to differentiate between a cat and a dog.
"We're going to win over robots, not because we're smarter, but because they are going to run out of batteries."
Watch Professor Hod Lipson, director of Columbia University’s Creative Machines Lab, discuss the latest developments in robotics and AI, the societal implications of technological advancement, and the role of business and industry in pushing the boundaries of innovation: