Armstrong Economics Blog/AI Computers
Re-Posted Jan 22, 2020 by Martin Armstrong
QUESTION: Marty; Socrates has done a fantastic job in picking turning points, rallies, projections for resistance, so many things. Yet IBM’s Watson was supposed to blow everything out of the water and find the cure for cancer and it has failed miserably. What did you do that IBM did not?
Also, the rumor is that China has made you a huge offer even citizenship because they intend to be the leader in AI. That’s why you’re having a WEC in Shanghai. Is there any truth to that since Socrates covers all of the main Chinese stocks as well?
ANSWER: IBM’s Watson was simply a giant neural net. I have explained before that I toyed with that back in the 1980s and quickly saw that it would lead nowhere. Real Artificial Intelligence is something that learns and analyzes on its own to create its own conclusion. In all honesty, 99.999% of what people present as AI are simply expert systems that can be distinguished rather easily because all they are doing is looking up some subject from rule-based systems.
IBM’s Watson could search the entire internet and gather every piece of information possible. That was the easy part. What IBM lacked was the expertise in how to do research. How do you know that one piece of info is the key or more important than another? They lacked the ability to create a real analysis capability.
Neural Nets were a great hope that somehow you just throw in all this information, shake well, and out will magically appear the answer. The presumption was that our minds are just supercomputers and they ignored perhaps the critical understanding of what makes one person brilliant at math and another a brilliant artist who can’t count beyond 1,000.
This theory that our minds are simply supercomputers ignored some critical traits and how to distinguish one person from another. There was this idea that if you create a neural net and stuff in all this knowledge it will learn all by itself. This was just a giant fiction to me, which was up there with the theory that somehow the machine will come to life and decide it wants to rule the world and take over humankind.
My approach was to teach it HOW to analyze, not WHAT to analyze. I created in the code as much of myself as I could without the human flaws that keep life interesting. I taught Socrates HOW to analyze and did not leave that to the chance of a neural net that cannot be tested.
I then had to teach it language so I could communicate with it to try to understand how it arrived at such a conclusion. I used my children to improve Socrates’ language skills back in the 1980s. I wrote a program to have an ongoing conversation with them. My daughter believed it was alive and would bring her friends over to show them a computer that talked.
Now as for China, no I do not have an offer from the Chinese government. We have been focused on expanding our services into China and we are not blocked like most other analysts because they do understand it is a computer writing all these reports so there is no risk of a person sneaking into paragraph three and writing, “BTW, overthrow your government.”
Yes, I would take in a major partner who brings something to the table in the field of expertise. This is not a start-up operation. We do not need money to fund a theory. We have clients in more than 137 different countries and a track record of more than 40 years.
And no, we have not been approached by IBM
A very good job of trying to explain very complex and technical theory in these two videos
Boeing’s Starliner has been designed and built for NASA’s Commercial Crew Program, intended to provide ‘taxi’ services for astronauts to the International Space Station. Today was the first full flight test, launching on board an Atlas V from Cape Canaveral the initial flight was flawless. However after separation the spacecraft got confused and maneuvered incorrectly, burning through fuel needed for the rendezvous. Ultimately the spacecraft was recovered by it was no longer in the correct orbit and at this time space station docking has been ruled out. The vehicle will be tested on orbit to make sure all the systems operated correctly before the Starliner returns and lands at White Sands on Sunday. John Kraus’s launch photos can be found here: https://twitter.com/johnkrausphotos/s…
Armstrong Economics Blog/AI Computers
Re-Posted Dec 5, 2019 by Martin Armstrong
Was Y2K a possible problem ?
I was programmer then systems programmer and Senior programmer for several companies in the late 1960’s into mid 1970’s. Indeed our records did have just a 2 digit year in all records.
Of course technology improved and and space would have become less important and history would suggest that the problem was foreseen and handled significantly.
Was that guaranteed ?
Of course not.
REPLY: There were computer systems that were the old system based only on a two-digit date. Those systems were predominantly the problem, but they were exaggerated to make it seem that the entire computer world would crash. So in that respect, I was wrong in not addressing the old two-digit systems. They were the minority