|
EC SD | I worked on AI back in the early 1990s, so fully understand what Marty is describing here. My team ran the AI models (expert system and neural nets) on the best supercomputers of the day, and also invented grid-computing to combine the power of thousands of (then state of the art) desktop computers. This allowed computing the models to the point we could model the events of the next hour of the hypothetical future in just two hours of real time. In other words, we could predict the likely events of the next day, by crunching numbers all night.
So my opinion of all of that:
1) It was a very interesting first job out of college
2) The next day is not bound by "likely" predictions and will unfold in quite unpredictable ways
3) It is hugely expensive to pursue predictions at that scale, only to be wrong most of the time anyway
After a few years, I felt more and more that this work was a real waste of time and taxpayer money so resigned and moved on in my career.
Here we are over 30 years later and AI is again the rage. I think it has a much better chance of becoming useful to society this time around. The AI focus now is not on "predicting the future", but instead on enhancing workforce productivity, gaining insights from big data, designing better medicines, etc. | |
|
Fundamental over load Thanks Chart guys
