Geoffrey Everest Hinton - Artificial Intelligence

1
Image
Geoffrey Everest Hinton
6 Dec 1947 (Unknown birth time)
Wimbledon, London


----------------------------------------
Geoffrey Everest Hinton was born December 6, 1947 in Wimbledon, London, England.
With the Sun conjunct Jupiter (on Antares) in Sagittarius and Mars (on Regulus) in Virgo, he is considered one of the 3 'Godfathers of Deep Learning' along with Yoshua Bengio (Montreal) and Yann LeCun (New York). At 75, nothing stops Geoffrey Hinton. Between his part-time job for Google and teaching at the University of Toronto, he volunteered as Chief Scientific Advisor for the Vector Institute.
In 2019, he received the Turing Award (the Nobel Prize of Computing), along with Yoshua Bengio and Yann LeCun, for their work on deep learning.
With the Sun at the midpoint of ME/UR, he has a technological mindset

WHAT IS DEEP LEARNING?

This is the most advanced form of what is more generally called "artificial intelligence", which tries to reproduce with machines the functioning of the human brain. The simplest form of artificial intelligence is to program precise commands. Deep learning is based on several layers of “neural networks???. Each layer processes a level of information – letters, photos, colors for example – which is sent back to the next layer for additional, increasingly precise processing. A distinction is made between supervised learning, where the data is usually labeled by humans, and unsupervised learning, where the algorithm does the sorting alone.

Q. Has deep learning made you someone who understands people better?
A. "I believe so, yes. Deep learning expertise actually gives a lot of clues about how the human brain works.
The rationalist view is that we are reasonable beings and, of course, there is a bit of that. But basically, we're a big analogy machine. We reason by analogies. You have to understand that people operate like that, to understand many other things about politics.
People behind Trump understood people's reasoning by analogy. Hillary believed they were rational. Trying to fight these populists who operate by analogy through reason is a losing proposition. It is in the very nature of people to be like this."

Q. But analogies can be misleading…
A. Let me give you an example of something that seems intuitively reasonable, but makes no sense and cannot be explained by logic. You know there are male and female dogs, and male and female cats. Forget it, suppose I tell you: you have to make a choice, whether all dogs are male and all cats are female, or all dogs are female and all cats are male. How are you going to make your decision? Almost everyone in our culture, and yours, will decide that cats are female and dogs are male. Why ? Because the neural activity vector you use to represent cats is closer to that of females than that of males. And vice versa for dogs. It's not rational, it's based on analogies. It's the kind of stuff used by populist politicians to take, for example, particular racial groups and stick all the bad stuff on them. We see that all the time."

Comment: In astrology, after we have seen hundred of charts we start to present our brain with 'analogies', to create 'representations' a new string of symbols and understanding.

Comment: Now, how do we learn astrology. Most of us learn it through books.

Hinton says that learning by remembering a list of words and trying to remember them is the approach by 'logic' and is the worst possible way because it does not involve a conversation and having some representations in our brain.
Furthermore, “It was my obsession: how does knowledge enter the brain, and in what form? I never believed it was a string of symbols. I've always believed that almost all knowledge comes from learning, not from being explicitly told, but from seeing bits of data. Intelligence consist of having representations […], with rules and inferences. The rules teach you how to manipulate the symbols to get new strings of symbols. […] That is the paradigm for Artificial Intelligence."

-------------------------
Having been influenced by two monuments of computer science, Alan Turing and John Von Neumann, who preferred to explain the functioning of the human brain in a probabilistic way rather than with formal logic, he says:

“Unfortunately, Turing and Von Neumann died too soon. If they had lived longer, everything would have been very different. What they lacked, moreover, were computers powerful enough and databases large enough to develop what are called "neural networks," essentially algorithms that send more complex information. from one layer to another."

“In retrospect, it turned out to be a matter of scale for computers,??? notes Hinton. "If you keep doing it over and over again, with a lot of data, you end up with these amazing, awesome neural networks, which can compete with humans in certain areas."

"There is hardly any significant company today that does not incorporate some form of deep learning into its operations. The applications in cybersecurity, visual and voice recognition, finance and health are countless."


For the “good??? Artificial Intelligence

He left the United States under Reagan because he refused to allow his work to be used for military purposes, believed that artificial intelligence could help counter Trump-style populism.

Q. Is it true that you left the United States in the 1980s to protest the Reagan administration's plans to use AI for military purposes?
A. "It was the time of the Iran-Contras affair. Most people I knew in Pittsburgh found that acceptable, because "we're Americans and we can do whatever we want in America. I was happier not accepting money for military projects, but it turned out that in the United States it is very difficult to survive without this money."

Q. You are also against lethal autonomous weapons…
A. "I am completely against it. I think it will make it easier for big countries to invade smaller countries, because the biggest military-industrial complexes will be behind them."

Q. But having taken a position, as a scientist, what is your power to prevent that? Can you imagine a way to incorporate a limit into your searches?
A. "No. Asimov had his three laws of robotics. The first law was that a robot could not harm a human being. Contrast that with the fact that the bulk of the money for robot development comes from the Department of Defense. It doesn't really work."

------------
https://www.lapresse.ca/affaires/2022-0 ... rofond.php
Karim Benessaieh, La Presse
Blessings!