Stay informed with free updates
Simply sign up artificial intelligence myFT Digest – delivered straight to your inbox.
Pepper the robot was born in 2014. It received a short wave of hype, including a Visit the Financial Times To meet the editor. “This robot acts autonomously, powered by love,” said Masayoshi Son, president of its main backer, SoftBank. Alibaba and Foxconn have also invested hundreds of millions in efforts to make robots an integral part of daily life. However this was not the case. You still find Pepper every now and then in a public library in Japan, unplugged, his head bowed, like a four-foot-tall Pinocchio who dreamed of becoming a real boy but never did. Production ceased in 2021 and only 27,000 units were manufactured.
However, the vision of humanoid robots – machines that look a lot like us and can perform all the work we don't want them to do – is too tempting to give up for long. Recent tremendous advances in artificial intelligence have spurred a new wave of artificial intelligence Enthusiasm for robotics. “The next wave of AI is physical AI. AI that understands the laws of physics, AI that can work among us,” Jensen Huang, CEO of chip design company Nvidia, said earlier this year. The boom in training artificial intelligence models to become the second largest company in the world by market value.
Billions of dollars in venture capital are pouring into robotics startups. They aim to apply the same kind of model training techniques that allow computers to predict how a protein will fold or create stunningly realistic text. They aim, first, to allow robots to understand what they see in the physical world and, second, to interact with it naturally, solving the huge programming task embodied in an action as simple as picking up an object and manipulating it.
This is the dream. However, the latest round of investors and entrepreneurs is likely to end as disappointed as those who backed Bieber. This is not because AI is not useful. Rather, it is because the obstacles to creating an economically viable robot that can cook dinner and clean toilets are a matter of hardware, not just software, and that AI itself does not address, let alone solve.
These physical challenges are many and difficult. For example, a human arm or leg is moved by muscles, while a robotic limb must be actuated by motors. Each axis of motion through which the limb must move requires more motors. All of this is doable, as robotic arms appear in factories, but the high-performance motors, gears and transmissions involved create significant cost, energy and multiple components requirements that can fail.
After creating the desired movement, there is the challenge of sensing and feedback. If you pick up a piece of fruit, for example, the human nerves in your hand will tell you how soft it is and how hard you can press it. You can taste if the food is cooked and smell if it is burnt. It is not easy to provide any of these senses to a robot, and to the extent that they are possible, they add more cost. Machine vision and artificial intelligence may compensate, by noticing whether fruit has been crushed or food in a pan has turned the correct color, but they are an imperfect substitute.
Then there is the issue of power. Any autonomous machine needs its own power source. Robot arms in factories are connected to electricity. They can't move. A humanoid robot will likely use a battery, but there are trade-offs in terms of size, power, power, flexibility, runtime, usable life, and cost. These are just some of the problems. Many smart people are working on it and making progress. But the point is that these are physical, long-term and difficult challenges. Even the revolution in artificial intelligence does not make it disappear.
So, what makes AI possible in the physical world? Instead of imagining how technology will enable new machines, it is more practical to imagine how existing machines will change once artificial intelligence is applied to them.
The obvious example is autonomous vehicles. In this case, the machine does not need to change at all: the car's movement through the physical world and its power source will work as they always have, while the sensing involved in driving the car is almost entirely optical. With the new popularity of artificial intelligence, the hype cycle surrounding autonomous vehicles has faded. In fact, it should be the opposite: autonomous driving is a broad market, a real challenge that AI can easily address, and a point that anyone tempted to invest in other applications for robotics should consider.
It also makes sense to think about how existing robots have evolved – from industrial robotic arms to robot vacuum cleaners. AI-powered machine vision will expand the range of tasks a robotic arm can perform and make working alongside humans safer. Lightweight, single-purpose devices, such as robot vacuum cleaners, will gradually become more useful. In Chinese hotels, for example, it is very common to have a robot deliver orders to your room. This type of limited and controlled independence is the easiest to implement.
In this way, AI will slowly bring us closer to Android devices. As for a robot like Pepper that can clean a toilet, sadly it's much easier to make a robot that writes bad poetry, and that's unlikely to change anytime soon.
