The mis-understanding between Artificial Intelligence and Machine Learning.
Fran?ois de Normandie
Chief mSoC Architect, ground up design for Automotive and Aerospace safe computing. UCIe representative. Level 2-4 Autonomous drive hardware. Jet Fighter Pilot, Oscar Bravo One.
Machine learning, alias ML is everywhere, it is being used to connect the dots between an inputs and desired answers. ChatGPT is not a thinking system, it is a non linear interpolation system, trained to give an answer a human would formulate based on its training set. There is literally no thinking, neither will to express an answer, what ChatGPT and the transformers are giving you is simply the closest interpolation of all the sentences it has stored, and it spit it out without really "thinking". ML is not intelligence, and I think the industry would be inspired to stop misleading the world about this.
Neuromorphic computing will be closer, reproducing better the human structure of thinking, but it will still have no of the chemistry stimulus that drive emotions either...
Globally, the world is being mislead about Machine Learning, it is not an intelligence, it is simply interpolating human articulated list of characters, there are no understanding, no will to solve the problem, no will to hurt or do harms, no will to live. You hear a lot of people misleading you for economical reason.
Why I am complaining?
Because we have entered an other "winter" of AI, the majority of the investments are done to "emulate and interpolate thinking", it is not invested into real thinking or creation of will. As long as the research is not getting there, Artificial Intelligence is an illusion to take the money of investors.
This is my very personal opinions.
Applying Semiconductor Knowledge to Your Test Challenges | Training Technical Leaders Using a Skills Based Approach
1 年I agree with you 200%. Also computer vision for detection of objects, etc, been around for 40+ years and is also machine learning- you train an algorithm on a data set. I use otter.ai to transcribe audio files. I finally noticed that its using machine learning to 'guess the word,' and it often guesses incorrectly. I'm okay with that because words spoken have lots of variations. And for my technical journalistic purposes it's good enough and has saved me Countless hours of my time.
35y+ writing software - looking for my next adventure
1 年I like to call it ??artificial intuition?? because, at best, it is that. It produces guesses. Now, there are plenty of situations where guessing is useful, I don’t deny that, but I’m worried just like you are about the way it is perceived (ie the way it is sold) : some fantasy of sentient silicon…