Artificial Intelligence Breakthroughs in 2015, the Singulari
Source: John J. Xenakis
Female humanoid robot. Do you think she’d go out on a date with you if you ask her nicely? (Japan’s National Institute of Advanced Industrial Science and Technology)
Many analysts consider 2015 to have been a breakthrough year for Artificial Intelligence (AI), not because of any single achievement, but because of achievements across the board in so many different areas.
Companies like Google, Facebook and Microsoft are now operating their own AI labs. In areas such as image recognition, computer vision, face recognition, voice recognition and natural language processing, there are a wealth of new products (think of Siri or OK Google) that are becoming increasingly reliable and increasingly available.
Several companies are testing self-driving cars, and they’re expected to be available commercially by 2020. Robots in the military are becoming more common, from robots on wheels to pilotless drone warplanes. All of these robots still require constant human intervention and control, but they’re slowly migrating away from human control to algorithmically based decision making and control. Robot form factors are improving, with some robots looking almost human.
In 2011, I wrote how the news that IBM’s Watson supercomputer bests human champions on Jeopardy! advances AI significantly, because it shows how, within a few years, computers will be able to “read” everything on the internet and learn from it. Today, IBM has Watson-based applications in multiple industries, from retail to healthcare. IBM also has a program to allow developers to incorporate Watson into mobile phone apps. Bloomberg and World Future Society and IBM
The debate about preventing the Singularity
The development of technology leading to the Singularity cannot be stopped. Even if the United States passed a law, or the United Nations passed a resolution, that would not stop China, India, Europe and other countries from continuing this technological development.
In fact, it wouldn’t even stop the United States development. A university scientist who’s working on a better self-driving car is not going to stop development because that technology might be used in the Singularity. The Singularity is coming, whether anyone likes it or not.
That hasn’t prevented a debate on how to save the world from AI. Elon Musk, the founder of electric car company Tesla Motors, has been one of the leaders in making that argument. He’s formed a non-profit artificial intelligence research company called OpenAI. This company will develop AI technology as open source, available for anyone in the world to download. Google and Facebook are also open-sourcing some of their AI technology. According to Musk’s partner, Sam Altman: “Developing and enabling and enriching with technology protects people. Doing this is the best way to protect all of us.”
If Musk and the others really believe that open-sourcing will protect the world from the domination of super-intelligent computers, then they’re living in a total fantasy. The concept is supposed to be that with many people around the world working on AI software, they’ll check each other and prevent the development of software that will dominate humans. The whole concept is so absurd, it’s hard to know where to start. Probably the best thing is to simply point out that making AI technology available to anyone gives everyone a head start. A jihadist in Karachi or a military scientist in Shanghai will be able to download the OpenAI technology and build on it to create intelligent robots that can conduct terrorist attacks or war.
In fact, the Information Technology & Innovation Foundation has nominated Elon Musk for the 2015 Luddite of the Year.
In 2015, the US Department of Energy awarded $200 million for the next generation supercomputer. It will be commissioned in 2018, with a performance of 180 petaflops (thousand trillion operations per second) of processing power. We’re already beyond the processing power of the human brain, which is estimated to be 38 petaflops. Computer power is doubling every 18 months. This is known as Moore’s Law, formulated by Gordon E. Moore, co-founder of the Intel in 1965. Moore’s Law has been valid for 50 years, through several technologies, and is expected to continue.
It’s the doubling of computing power every 18 months that makes it all but certain that the Singularity will occur by 2030, whether we like it or not. Wired and Washington Post and Defense One and Government Executive and Information Technology and Innovation Foundation (ITIF) and US Department of Energy
Artificial Intelligence and Climate Change
Politicians and climate change activists like to say that the claims about climate change have been endorsed by 95% of all the scientists in the world. This claim is a total fraud, because it confuses two things.
First, we have the claims by scientists that the earth is warming because of human activity. Arguably, that HAS been proven by scientists. But that’s all.
The second part is predictions about the future, which are mostly total crap, and certainly not science. In fact, climate change scientists have been making predictions for 25 years, and they’ve almost completely turned out to be wrong. The truth is that scientists who claim to know what the earth’s temperature will be in 2100 can’t even predict what the weather will be next month.
During my lifetime, I’ve seen any number of hysterical environment disaster predictions. My favorite was the prediction in 1970 by far left-wing Ramparts Magazine that predicted that the oceans were becoming so polluted that by 1980 the world’s oceans would be covered by a layer of algae. It didn’t happen.
One way to know that the climate change activists are wrong is that these climate change scientists never mention the Singularity or future technology. There is a very powerful historical precedent that they all ignore. In the late 1800s, streets in large cities were full of horses (think of a traffic jam in any large city, with horses instead of cars). These horses were producing huge volumes of urine, manure, flies and carcasses ― not to mention cruelty to horses. By 1900, there was 1,200 metric tons of horse manure per day. There were international conferences (like today’s climate change conferences) that accomplished nothing. But within 20 years, the problem took care of itself because of new technology �C the automobile.
History shows that new technology, including new AI technologies, will solve the “climate change” problem, and that politicians will have absolutely nothing to do with it, except to take the credit when something works, and to blame someone else otherwise. Daily Caller and From Horse Power to Horsepower and The Great Horse Manure Crisis of 1894 and Great Moments in Failed Predictions
Proof that the Singularity will occur by 2030
The Singularity is the point in time when computers will be more intelligent, more able, and more creative than humans. Ten years ago, in 2005, I wrote an article called “The Singularity,” in which I forecast that the Singularity would occur around 2030. Today, I see no reason to change that forecast.
There is now an updated version of that article available: “Artificial Intelligence and the Singularity by 2030″. The contents are as follows:
Justification of the 2030 date for the Singularity. This includes a description of the “super-intelligent computer” algorithm, which is the algorithm that will be used by humans to implement the first computers that are more intelligent than humans.
A proof, based on reasonable assumptions, that any intelligent species on any planet in the universe will develop in a way that’s similar to the development of humans, including following the same Generational Dynamics cycles as humans.
Some speculation about what will happen after the Singularity, not only to humans, but also to other intelligent life in the universe.
For those interest in computer software algorithms, the article contains the Intelligent Computer (IC) algorithm in the article is as follows:
Intelligence isn’t some magical, mystical force. It’s actually the ability to find new ways to combine previous experiences in new ways. A new discovery is made by combining old discoveries in new ways, in the same way that jigsaw puzzle pieces can be put together.
A computer can do the same thing by combining “knowledge bits” (KBs) in new ways, to learn new things, in the same way that jigsaw puzzle pieces can be combined. Computers can do this much faster than humans can.
Decisions can be made by using the same “minimax algorithm” that’s used to implement games like chess.
This algorithm would work today, except that computers aren’t yet fast enough. The speed of computers doubles every 18 months, and by 2030 computers will be fast enough to implement this IC algorithm.
The article also contains a proof (under reasonable assumptions) that every intelligent species in the universe must follow the same Generational Dynamics cycles as humans is outlined as follows:
For any species (including humans) to survive, the population growth rate must be greater than the food supply growth rate. This is what I call “The Malthus Effect,” based on the 1798 book by Thomas Roberts Malthus, Essay on Population.
Therefore, for any species, there must be cyclical periods of extermination. This can be accomplished in several ways, such as war, predator, famine or disease. But one way or another, it has to happen.
Non-intelligent species will simply starve and die quietly when there’s insufficient food. But intelligent species will form identity groups and hold riots and protests, and eventually go to war. These will be the cyclic crisis wars of extermination specified by Generational Dynamics, and every intelligent species in the universe will have them.
| }
|