02 Sep Creating Human-like Machines with Artificial General Intelligence

Remember David from A.I.: Artificial Intelligence by the legendary director Steven Spielberg? It is a beautiful movie, where the humanoid robot David is on the quest to find the Blue Fairy for winning back her human mother’s love. The ending of the movie leaves people in tears, even today. The movie was very close to the concept of Artificial General Intelligence (AGI) – a machine capable of understanding and experiencing the world as any human and with the same capacity to learn performing a variety of tasks.

AGI doesn’t exist in the real world as of now. However, the science-fiction movies have featured it for over a century and made it popular in modern times with movies like A.I.: Artificial Intelligence and 2001: A Space Odyssey.

How does artificial general intelligence get to be so human-like?

AGI constitutes a system with cognitive computing capabilities and comprehensive knowledge that makes its performance indistinguishable from that of humans. In theory, AGI would be able to successfully combine flexible thinking and logical reasoning just like humans – think instant recalling or quick number-crunching. It would also be able to carry out tasks that humans could and couldn’t undertake.

If it existed, what could artificial general intelligence do?

AGI-enabled machines would be as dextrous and as mobile as humans. Think of them as a ‘new species’ of machines capable of doing anything. Once AGI exists, it would improve itself at an exponential rate, eventually evolving to the point where its intelligence will operate at levels beyond human comprehension.

Some experts have raised questions on whether such intelligence would be desirable. Noted theoretical physicist Stephen Hawking warned, “It [AGI] would take off on its own, and re-design itself at an ever-increasing rate. Humans, who are limited by slow biological evolution, couldn’t compete, and would be superseded.”

However, whether AGI would ever be possible or not, that’s a different question.

Would AGI ever be possible?

Ray Kurzwell, Google’s Director of Engineering, predicts the AGI systems passing the Turing Test by 2029. Others don’t respond so well. Modern day artificial intelligence systems excel at single tasks, but their nature is different from that of AGI. Without a clear path to track its progress, these systems are not exactly stepping stones to AGI.

The fact is, AGI is nowhere near inception. The super-efficient worker machines appear very fascinating in the movies but realistically, they are still very far from human performance. Conscious reasoning, logical thinking, and meta-learning (i.e. learning to learn) are the forte of the human race, and you can rest easy knowing that the Skynet from Terminator is not coming to life anytime soon.