What is Artificial Intelligence?

Imagine a time when there was no electricity. Before fire, or manmade light.

Ray Kurzweil - Law of Accelerating Returns

- Think about the technology you use every day. How long ago would you have to go back to show a person our current technology and it would literally blow their minds? = 1750 CE

What about a person from that time? Not just 250 years. But probably closer to 15,000; before there was mainstream agriculture and cities.

And a person from that time? Probably 100,000 to before written language and fire.

The technological advancements of the last 30 years, double those of the previous 30.

This is based on the fact that the technology that we used in 1985 to advance our science from 1955.

Kurzweil also says that all of the technological advancement of the 21st century would have happened in just 20 years at the rate we develop science in 2000. That means it's 5x faster just 100 years later.

So the next head explosion era will be in just a few decades, not centuries.

We tend to think that human development happens as a linear process. But it's not true. Our changes in understanding of the world around us happen as exponents.

Our understanding of progress is mistaken. It's not a uniform growth, but an exponential one.

John McCarthy, who coined the term “Artificial Intelligence” in 1956, complained that “as soon as it works, no one calls it AI anymore.”

We actually use AI constantly in our daily lives.
- Calculators
- Google Maps
- Siri

The term singularity is used in physics to describe a point where the normal rules change so much that they no longer apply.

Vernor Vinge wrote a famous essay in which he applied the term to the moment in the future when our technology’s intelligence exceeds our own.

1) Artificial Narrow Intelligence - Specializes in one area. Like the computer that can beat the chess world champion.

2) Artificial General Intelligence (AGI): Sometimes referred to as Strong AI, or Human-Level AI, Artificial General Intelligence refers to a computer that is as smart as a human across the board—a machine that can perform any intellectual task that a human being can. Creating AGI is a much harder task than creating ANI, and we’re yet to do it. AGI would be able to do all of those things as easily as you can.

Artificial Superintelligence (ASI): Oxford philosopher and leading AI thinker Nick Bostrom defines superintelligence as “an intellect that is much smarter than the best human brains in practically every field, including scientific creativity, general wisdom and social skills.” Artificial Superintelligence ranges from a computer that’s just a little smarter than a human to one that’s trillions of times smarter—across the board.

ANI is used in pretty much every technology you can imagine. From Google Maps and SIRI, to the machines that monitor factories and give real time feedback to the operators.

Spam filters.

Ads the pop up on facebook right after you Google something.

Google Itself.

ANI itself is not dangerous. Every system in the world runs on it. However, each ANI that is created, is like another brick on the road to an ASI.

Creating an artificial intelligence that can function in a similar way to the human mind is complicated in some interesting way.

The things that are extremely complex for us -- advanced calculations, tracking fluctuations in the stock market, searching through millions of images, and so on -- are easily done by computers at millions of times the speed that the human brain can do them at.

However, things that are done almost instinctively for us, are so complex to make a computer do, that we've only begun to scratch the surface in how to program and teach computers to do them.

This is largely due to evolution. The things we have spent millions of years learning how to do, are more advanced than we would at first realize. But the things that we have been doing for a relatively short time -- reading, math, etc. -- a computer can do much faster.

For example:

One limiting factor to AI is that our computers aren't fast enough.

The human brain can do approximately 10 quadrillion calculations per second (10^16).

The fast computer in the world, Tianhe-2, can actually do 34 quadrillion cps, but this computer cost almost $400 million to build, take up 3/4 of a square km, and uses 23 KWs of power to run. The human brain uses 20watts.

1. Copy the brain

2. Try to program evolution

3. Write a program that tells the computer to figure it out

- Any AI that we create will be programmed to have recursive intelligence growth. It will constantly be making itself smarter. It will only meet Human level intelligence for a moment in history and will quickly make itself far far smarter than us.


Everything about a computer has the potential to beat the human brain. Our neurons only function at 200Hz, while computer processors function at ~2 Ghz.

Our brains storage is limited to our head space. Computers can constantly upgrade their storage space.

Our memory is unreliable and fluctuates. Hardware memory is almost perfect.