TL;DR -- 2023 set the tone for expanding upon the reality of the computer. Depending upon various factors, we can see a range from highly optimistic all the way to marvelous. How can this be? Mathematics as the enabler being used without proper understanding of the underlying limits. That discussion needs attention, imperatively, and now is the time for such an effort.
--
We put a label of "portal to truth" back in the 2017 timeframe somewhat tongue in cheek. However, at the time, we had been following technology for decades, principally from the viewpoint of advanced computing whose history tracks the marvels. Being a member of IEEE, we also had access to facts and opinions from the beginning.
People have mostly been the source for wisdom and knowledge. That is, until we saw the advent of the xNN/LLM models in 2023 on a massive scale which resulted in some interesting social phenomena that need study, analysis and discussion. We can report on that.
The first cut was oriented toward the engineering and technical camp as it stressed that artificial intelligence, say as envisioned by John McCarthy, was more than machine learning. What we saw was a highjacking by one section of the entire discipline. That seems to have abated somewhat.
But, the underlying dynamics are still there, needing attention. We have completed the third of a five part series which sets the basis for a dialog, using insights from the technical side of what seemed to be a marvelous emergence of capabilities beyond limit.
We'll quit using AIn't, in other than a limited sense, and bring mattes to fore that need to be in the analysis. The framework essentially is mathematical modeling and computational modes that support this. And, the work can be pinpointed by time over more than two centuries.
In other words, the biological basis for cognition and intelligence is still under study. The xNN/LLM are approximative models that are quite limited despite appearances otherwise. Even when joined with robotic efforts, we are still talking "toy story" than mature assessments. Or, for that matter, immaturity is to be on the table.
The software industry attempted to define what this meant from the view of systems. There was a Software Maturity Institute created more than two decades ago. We need to look at that. At the same time, since the internet has been available to general use, we have seen a shambles come about, with a few good things. Granted, how one would assess the progress of the past two decades is dependent upon several factors as well as things that are opinion based. But, as we noted, opinons matter. There is no aspect of human life that is without opinion, even science.
Ah yes, first principles? We will venture there. Basically, they are what a group decides. We'll cover this topic from all necesary angles.
----
This series of articles is on themes that need to be discussed. I have been discussing these themes for the past few years in various forums. 2023's events related to what was called "AI" motivated the series. The articles are about KBE (knowledge-based engineering) which followed after KBS (knowledge based systems; like KBS, KBE unfolded in a commercial environment and was proven many ways and times in the real world.
A major theme? We need truth engineering in a non-academic sense. Details are where truth is assessed.
- 1st – AI, not solely ML
- 2nd – Knowledge and truth
- 3rd – Physicalness and mathematics (in the interim, LinkedIn).
- 4th – ML’s emerge/surge and data/decisions (in process)
- 5th - Where do we go from here? (in process)
Remarks: Modified: 02/13/2024
No comments:
Post a Comment