Artificial intelligence: Valuable tool or dangerous threat?

 

Artificial intelligence technology has advanced leaps and bounds in a relatively short period of time. What once seemed like science fiction is quickly becoming a reality. The sudden advancement in the technology has reawakened some fears about what artificial intelligence could mean for the future of humankind. Those speaking up about the dangers of artificial intelligence aren’t just conspiracy theory whack-jobs either. Recently, Stephen Hawking, arguably one of the most brilliant minds of the last century and Elon Musk, the CEO of Tesla Motors, have both voiced concerns about the future of artificial intelligence.

Should we fear artificial intelligence?

According to Hawking, real artificial intelligence could bring about the end of the human race faster than global warming or any other threat to mankind. Hawking theorizes that AI could redesign itself at an exponential rate far beyond what humans have the ability to keep up with. Because computers aren’t limited in the same ways that humans are, they could very quickly surpass humanity.

Elon Musk agrees with Hawking but it quick to point out that true artificial intelligence is still quite a ways away. While there have been many impressive advancements in the field of artificial intelligence in recent years, we are nowhere near being able to create artificial intelligence capable of becoming more intelligent than humans.

Google’s conversational modeling AI

A major obstacle that needs to be overcome to create true artificial intelligence is conversational modeling. While computers have long been able to perform complex calculations far faster than any human, they struggle to perform tasks humans handle with ease such as forming and comprehending spoken language. Most artificial intelligence systems out there can only handle this task as it applies to limited subjects.

Google has been working on an AI system that could change all of that. The difference between Google’s AI and other artificial intelligence systems is that Google has not only added in domain specific data sets but general domain data sets as well. The idea behind this is that Google’s AI can understand a variety of questions on various subjects and give proper responses. In a recent test of the AI’s abilities, she (the artificial intelligence referred to itself as a “she”) was able to carry on a basic conversation including greetings and various pleasantries. She was also able to answer simple questions, including questions about general information such as whether cats have tails or wings. Finally, the AI even handled some questions relating to philosophy such as “What is the purpose of life?”

In summary, perhaps Hawking said it best, while artificial intelligence may be able to pose a threat to humankind within the next 100 years, current AI systems are proving to be very useful tools for humans. To benefit from it without being threatened by it, there will need to be regulations in place so that humans can’t bring about their own destruction.

Source: Tech.co

 

What Hollywood gets right (and wrong) about artificial intelligence

 

Since the concept of artificial intelligence was first dreamed of, Hollywood has made a tremendous amount of money portraying it in film. Movie-goers are naturally interested in artificial intelligence and like to imagine the worst possible outcomes that could result as artificial intelligence becomes a more common part of our daily life.

Of course, Hollywood has been known to get a few things wrong when it comes to portraying scientific technologies on the big screen. What is really surprising is they actually get a few things right. Here’s a look at what Hollywood gets right and wrong about artificial intelligence.

Mind uploading

One of the most common tropes in artificial intelligence fiction is the concept of mind uploading, or digital immortality. The idea behind mind uploading is that humans can artificially become immortal by uploading human consciousness into a machine or robot of some kind. The most recent Hollywood film to make use of this trope was Chappie. Though the concept has enjoyed a lot of popularity in Hollywood films, Artificial Intelligence experts say that it’s also one of the most inaccurate. Currently, science is nowhere near being able to upload human consciousness into a machine. Though a few theorize it could possibly happen far in the future, the majority in the scientific community believe it’s nothing but science fiction.

Changes in agenda

One of Asimov’s three laws of robotics states that “A robot may not harm humanity, or, by inaction, allow humanity to come to harm.” But there is no shortage of movies that involve robots governed by artificial intelligence choosing to implement the Zeroth Law in which they diverge from their programmed agenda to achieve what they perceive as a greater good. In I Robot, for instance, the robots stage a revolution against the humans. According to the scientific community, movies like Steven Spielberg’s AI are more accurate because artificial intelligence cannot stray from its programmed instructions.

Robot feelings

Another favorite trope in artificial intelligence fiction is the robot that develops human emotions. Of all of the artificial intelligence myths, this one is probably the most subjective. According to experts in artificial intelligence, there is really no science to prove or disprove the possibility of robots developing human emotions. The question largely depends on how we define human emotion. Some theorize that emotions are nothing more than the result of electrical and chemical reactions in the brain. In that sense, it could in theory be recreated in a robot though that science is probably decades away.

Source: OuterPlaces.com

Thought Vectors – Words Reduced To Math

The idea of turning manmade constructs (made up stuff) into mathematical formulas has been the key ingredient in almost every advancement in technology throughout history. Recent advancements in the process of turning written language into mathematical formulas, has allowed computers to calculate human language and ideas in amazingly constructive ways.

Character Vectors

Character codes aka numerical representations of a single character. This is a little different than the 1s and 0s that make up the actual character, but the concepts are related because the numbers are made up from bits anyway.

Example:
a” = 97
“A” = 65

Word Vectors

Numerical representations of whole words.

Example:
animal” = 97, 110, 105, 109, 97, 108

Phrase Vectors

animals run wild” = 97, 110, 105, 109, 97, 108, 115, 32, 114, 117, 110, 32, 119, 105, 108, 100

Thought/Idea Vectors

The mathematical calculation of vectors applies to thoughts by simply calculating multiple sentences and the vectors of each word.

However, it goes further. You also have to calculate the vectors for all the words in the definitions of each of those words. You have to take into account how the words can have different meanings based on the way they are used.

Concept Vectors

Here is where you get into “cognitive computing” as talked about in the media. This is a far more thorough and advanced approach. It calculates the vectors of every single layer mentioned above, but it includes synonyms, antonyms, grammar and even the history and origin of the words.

When you calculate all of these together you can create a mathematical formula for representing an actual complex concept. This is where the real advancements in machine learning are coming from such as IBM Watson, Siri, Google Brain and others.