Hod Lipson on the six waves of Artificial Intelligence and its impact on work
“AI is like this air- it’s everywhere, permeates everything we do but we can’t see it,” stated Hod Lipson, Professor of Engineering & Data Science, Columbia University, on his opening keynote on day 2 of TechHR 2019. Though big data and AI are the most used words in any earnings call today and billions of dollars of investment are pouring into it, yet Hod shared how its progress has still been slow.
So what exactly is the opportunity that can AI present and how will it be in the future? While every company is claiming to do AI, what does artificial intelligence exactly mean? Hod shared how AI moves forward in sort of waves and has progressed in six waves since its inception.
Rule based AI
The one major factor driving AI now is that computers are getting faster and cheaper and in a few years, they will be 1,000 times faster. The first wave that dominated AI in the first 50 years was traditional or rule-based AI, which was based on certain rules or algorithms. When a machine learns these rules very quickly, it appears to be smart. But the challenge with this AI was that it could not be smarter than the programmer.
Predictive Analytics
In the last couple of decades, we have moved from rule-based systems to a more statistical approach where we don’t tell the computer what to do. Instead, we show it examples and it calculates the odds of things happening or not, which is the second wave of predictive analytics. And predictive analytics is where most of the HR community is today. So while this wave will allow us to predict things as complex as the stock market yet challenges exist with this too. For instance, up until five years ago, most AI systems on the planet could not do something as simple as differentiate between a dog and a cat.
Cognitive Computing
A lot of growth has now happened in AI because machines can finally understand what they are seeing. Today in a nanosecond, they can understand whether it’s a dog or a person or a chair. Machines can finally drive a car because they can comprehend the road they are driving on. We have apps that can look at skin lesions and tell if you have cancer or not. This is the third wave of AI called cognitive computing but there are still things that machines can’t do. And one of them is being capable of having a real conversation like humans do. Even the chatbots of today are based on the dialogue trees of the 1950s.
Creative Machines
The fourth wave of AI is that of creativity. Machines can now create things out of thin air such as generating fake images and even designing things such as antennas. Their creativity is also extending to arts, where AI can now create paintings out of thin air. The next level to which AI is moving is to create a better AI.
Physical embodiment
The fifth wave of AI is the wave of physical embodiment. AI in the physical world is called robotics and is still a hard world for AI. It’s still very difficult for robots to walk in the real world or even fiddle with something as gracefully as a human can. This is because fields like material science, battery are not as progressing as fast as software. The entire human brain operates on 20w of power while even the latest AI takes 200w of power and can only do a fraction of what a human brain can do. Thus in times to come, the biggest energy consumption will be in AI not in heating or transportation.
Thus, anyone who works in an unstructured environment with their job involving working with hands, their jobs are safe. So if your hands are dirty, your hands are safe as it is very difficult for AI to do physical jobs.
Sentience
The last wave of AI is the wave of sentience. Will AI ever have its own free will is a question on everyone’s minds. Can it have emotions? Can AI have sentience? Hod believes this will happen and it happens when all these AI capabilities are turned inside and the machine begins to model itself. So a robot with four legs which can sense begins to form an image of itself and then teaches to walk itself in about four days.
Ultimately, Hod concluded that everything in AI is a double-edged sword. The same technology which can help agriculture through drones can also create drones that are assassins. A lot of people will lose their jobs such as drivers but at the same time, thousands of people will gain jobs and thousands of people who are killed by cars driven by humans will be safe. Thus AI is an amplifier- it amplifies the basic human traits and we should make sure they amplify the right ones.