This is another time that I am marveled by my friend’s kid. A tiny human not over 2 years old showing intelligence beyond their size.
“Cryyyy”, a loud voice is followed by a frown. “Cryyyy” is what the child is saying to define the action and informing the parent that they are about to cry. I will cry now if I don’t get what I want. The “cry” is a trailer that will turn into an actual cry which is a 2 full hour film.
How can a child less than 2 years old do such intelligent trick and performance? I am amazed and told the parents that your child is “darn smart”. “This is how it is, they know how to get what they want”, says the mother, as if this is a normal thing.
It is not! We can surely learn something from this.
This is what Josh Tenenbaum is trying to accomplish. Tenenbaum is head of a project with the coolest name MIT Quest for Intelligence (https://quest.mit.edu). This new project, recently launched on February 1, 2018, aims to develop science and engineering for human and machine intelligence. This is “an initiative to discover the foundations of human intelligence and drive the development of technological tools that can positively influence virtually every aspect of society.”
This project brings together scientists from various fields such as biology, computer, society, and engineering. This is an attempt to seek answers about intelligence and forecast that if we build “reverse-engineer” and embed human intelligence in machines in the world of 1 and 0 (or beyond that in the quantum world), how will it affect the human race?
Some questions that the project aims to tackle are “Imagine if we understood how the human brain works with intelligence and use the knowledge to replicate this in machines”, “Imagine if artificial intelligence was socially and emotionally intelligent enough to empower everyone to flourish – from individuals to societies.”, and “Imagine if we could build a machine that grows into intelligence the way a person does – that starts as a baby and learns like a child.”
Josh Tenenbaum focuses on the last question. He mentions that if we know how a child learns then we can build a machine that has the true “ability to learn”.
AI already has the ability to learn but many views that the process is still limited. The development of the true ability to learn will bring humanity closer to building the General AI. Tenenbaum shares about the famous AI, Watson or AlphaGo that “none of these systems are truly intelligent. None of them have the flexibility, common sense, general intelligence of a two-year-old, or even a one-year-old. So, what’s missing? What’s the gap?”
In the search for the “missing element”, Tenenbaum studied how small children are able to visualize aspects of the environment using a 3-D model. Another research involved a program that is able to understand characters or objects in an image by only observing a few examples in the beginning (this is different from Deep Learning that requires enormous data).
Today, his research already exists such as teaching “common sense” to AI (by imitating human interactions with the environment). This is used in a company that originates from MIT called iSee (http://isee.ai) which specifically develops products for autonomous driving systems. They promise that the system will “be able to solve complex problems in the real environment”. iSee founder mentions that “the human mind is super-sensitive to physics and social cues. Current AI is relatively limited in those domains, and we think that this is actually the missing piece in driving”.
Tenenbaum has a dream. He dreams of looking back to AI 50 years ago when they were trying to imitate or utilize human intelligence as inspiration. He mentions that this is the right time. This is the time that the fields of cognitive science and neuroscience are now more mature. This is the right time to build a child.
A child that is as intelligent as a human.