What will take us from potential to reality in the next 18 months?
Superintelligence, roughly defined As an AI algorithm that can solve all problems better than people, it will be a milestone for humanity and technology.
Even the best human experts have trouble making predictions about highly probabilistic and perverse problems. And yet those wicked problems surround us. We are all experiencing immense change in complex systems that impact climate, public health, geopolitics, and the basic needs served by the supply chain.
It is virtually impossible to determine the best way to deliver COVID-19 vaccines without the help of an algorithm. We need to get smarter in the way we solve these problems, fast.
Superintelligence, if achieved, would help us make better predictions about challenges like natural disasters, build resilient supply chains or geopolitical conflicts, and develop better strategies to solve them. The last decade has shown how much artificial intelligence can improve the accuracy of our predictions. That is why there is an international race between corporations and governments around superintelligence.
In the next year and a half, we will see increased adoption of technologies that will trigger broader change in the industry, just as Tesla triggered the transition to electric vehicles.
Highly credible think tanks like Deepmind and OpenAI say the path to superintelligence is visible. Last month, Deepmind said reinforcement learning (RL) could get us there, and RL is at the heart of embedded AI.
What is embedded AI?
Embedded AI is the AI that controls a physical “thing”, such as a robotic arm or autonomous vehicle. It is able to move around the world and affect a physical environment with its actions, in a similar way to how a person does. In contrast, most predictive models live in the cloud doing things like classifying text or images, directing bit streams without moving a body through three-dimensional space.
For those who work in software, including artificial intelligence researchers, it is too easy to forget the body. But any superintelligent algorithm needs to control a body because many of the problems we face as humans are physical. Firestorms, coronaviruses, and supply chain breakdowns need solutions that aren’t just digital.
All the crazy Boston Dynamics videos of jumping robots, dance, balance, and run are examples of embedded AI. They show how far we’ve come from the first breakthroughs in dynamic balance of the robot made by Trevor Blackwell and Anybots over a decade ago. The field moves fast and, in this revolution, you can dance.
What has built-in AI blocked so far?
Challenge 1: One of the challenges when controlling machines with AI is the high dimensionality of the world – the wide variety of things that can attack you.