Kindle Notes & Highlights
Read between
November 3 - November 7, 2020
Artificial intelligence is a system that tends to simulate intelligent behaviors into computer controlled machines or digital computers. AI normally gives a machine the ability to carry out tasks normally associated with intelligent beings like us. Some of these tasks include translating languages, decision-making, speech recognition, and visual perception. In simple terms, artificial intelligence is the capability of any machine to mimic intelligent human behavior.
The idea of making machines and computers smarter and intelligent started thousands of years ago thanks to Western, Indian, Chinese, and Greek philosophers. These philosophers attempted to describe human thinking as a symbolic system.
The way that science defines intelligence is that it is an entity that is capable of attaining new information and then using that information with past information to define current information.
Back propagation is the center point of nearly all machine learning algorithms that are currently at play. A neural network is designed to take database input and perform actions on that information inside of neurons that then produce an output. If the output is wrong, we then change what is occurring inside of the neurons so that we try to get to a more optimized answer.
If machine learning is just a combination of statistics and programming, robots are a combination of a great many other things.
If someone has a caretaker that is a computer six days out of a week, it would lead to them to appreciate human nature by the seventh day of that week
Machine learning is a type of AI that can further be broken down into the category of “deep learning”, the current industry favorite.
AI programs can modify their own code to get smarter
The field of research into machine drives is called instrumental convergence. The most famous hypothesis coming out of this field is called the Riemann Hypothesis catastrophe by Marvin Minsky.
You cannot automate the process of building a full-scale website, you can automate the design process, the building blocks, and many of the different elements of a full-scale website but that website changes based on the company needs.
The idea of universal basic income is to give everyone a base income so that no one starves to death. This idea is not new and in fact, a lot of communist countries, as well as some socialist countries, believe in basic income for everyone in society.
Elon Musk already believes that anyone with a smartphone is a cyborg. The smartphone opens so many avenues of increasing one’s intelligence through a direct connection with the Web.
What exactly are the trading computers doing? Well, they are doing just that, trading. They are completely cold and unemotional. They will not buy an asset because they are "rooting" for it or sell because they simply don't like the management. They will only trade according to strict and clearly defined parameters, and they can process tens of thousands of these parameters per second. This is not quite the same as high-frequency trading, although AI can be (and is) employed in this area too.
This aspect of randomness is a central problem in reinforcement learning because intelligence is supposed to be modeled after purpose, not the roll of a dice.
Unlike traditional machine learning, reinforcement learning is more akin to the study of decision-making. It borrows concepts from several disciplines including computer science, economics, neuropsychology, and mathematics.
One good thing about computers and artificially intelligent machines making such decisions would be that they don't have the same inherent corruptness or human desires which are commonly found in Maslow's hierarchy of needs. This is both good and bad because the computer might negate the reality that humans won't accept the answer if it doesn't satisfy some of those needs, yet at the same time, it needs none for itself, which makes it an impartial judge and jury able to render the right decision.
Experts claim that robots are not likely to become creative in the near future. To automate a function, explicit and thorough instruction is required about how to accomplish creative goals. It’s true that algorithms can be designed to create infinite paintings. However, it’s hard, if not impossible, to teach said algorithm how to tell what makes one painting remarkable and another worthless. Another issue here is that it’s hard to make automated the process of combining ideas across many sources that makes up the foundation of creativity in humans.
Robots can also create breathtaking visual art. One remarkable program, called AARON can mix paints, wash its paintbrushes, and make masterpieces in a short period of time.
Advances are continually made in cloud robotics. “Cloud robotics” refers to machines which are connected, in the cloud, to supercomputers.
