Brad Rutter and Ken Jennings were jeopardy grand champions; but this time they were up against a machine: IBM’s Watson. By the end of the 3rd and conclusive match Watson was up by $50,000. It was not even close. Watson was a program conceived and developed by David Ferrucci at IBM. Watson uses hundreds of algorithms to break down the questions and then find a list of answers. It then chooses the highest weighted answer. This to me sounds like what we all go through when taking a test, it is not that remarkable. What is remarkable is that IBM’s machine had enough computational power to scour the internet, breakdown language, find clues, and find weighted answers in a matter of seconds. 5 seconds, to be exact. That is how long jeopardy contestants get. Progress keeps moving, and programmers and computer builders have not been sitting on their hands for the past 12 years or so.
At some point we must come up with a proverbial definition of what AI is, so here goes: “AI systems work by combining large sets of data with intelligent, iterative processing algorithms to learn from patterns and features in the data that they analyze.” I found this on Colorado State University CSU.edu. The computer nodes were originally conceived as computer simulations of what human nerve neurons do. Programmers have continued to get better and better at getting these nodes (information junctions) to function more efficiently and more accurately until today. Companies are rolling out AI assistant programs commercially all over the world. Yes, pandora is definitely out of the box.
We are at stage 1 where the machine is used to help solve a problem kind of one thing at a time. Stage 2 machine intelligence is as smart as a human across the board. Stage 3 Machine consciousness where the machine is much smarter than the best human brains across the board.
Eventually whole factories will work off commands from a computer which learns and adapts to make everything run quicker and smoother. Maybe 1000 times smarter than what we do today when we plug cars in so the computer will tell you what is wrong.
Most of what we have here is Large Language AI which scrubs the internet to do tasks and get answers. Graphics based AI grew out of the gaming industry. Learning images and their relationships to other images etc. If we ever get reliable self-driving cars most likely it will come from machine learning from the video of actual situations and reactions. At least that is how Elon Musk and company are approaching it with their DOJO computer being fed video of actual traffic situations. Stopping for stop signs is one example. The only issue is, we almost never come to a full stop at stop signs, so human programmers must tweak the commands.
Humans have constantly continued to invent and evolve. Wheel -Iron -Steam Engine-Petroleum Engine-Electricity-Computers-
AI is here, but the bigger conundrum is, are we ready for it. We humans are adaptable, that is why we have survived. AI will test the limits of our adaptability. I worry about some of this because as the old saying goes “Idle minds/hands are the devils’ workshop”. No matter how you view AI, it will give us more free time.
Thanks, Andy McClung CFP®
Bernardmarr.com- Brian Walsh axios future-csuglobal.edu-
2023 Market Results
S&P 500 +14%
NASDAQ Composite +30.3%
Dow Industrials +3%
Russell 2000 -2.6%
World ACWI +9.8%
Source Wall Street Journal 11/08/23