Wednesday, June 26, 2024

New technology

TL;DR -- GenAI seems to have gone into the background of ubquitous stuff. Well not. Methods for future sanity preservation are still there on the table for identification and resolution. The game has changed, though, as what can one believe now? Concern for the state of affairs ought to be in the attention of all, top to bottom and back up. That is one of our focal points. 


John has been involved in a project for a couple of months about which he will write later. So, as with any type of involvement, it takes time and energy. Hence, the post count went down here. But, we're still paying attention. 

The past year has caused many to reflect on technology. GenAI's emergence and chaotic nature are subjects that will be needing attention. But, there's a lot more. Just the computer, in general, needs to be better understood with respect to its history and its impact on our lives and on the world in general. 

Some of late took the task of arguing that people are less smart than computers. That debate is old. Expec that we'll be weighing in on the discussion. For now, though, we'll collect links to material of interest. 

The discussion will look at history which is in our scope. Computers can be dated to about the 1950s though one could go back further. There is a definite bit of periods that can be seen. Take the 21st century, post Y2K, things on the internet (TCP/IP) got interesting. We have some post on that having started in 2007 on other subjects and getting the TGS blog going in 2010. 

Over the past 24 years, we can mark where technology influeenced computing. The 2008 release of the Apple Phone is a major step. Since then, we have seen several changes of a major scale come into play. One of these deals with the "clouds" of the realm, though GSA (the U.S. Government's purchasing department) said okay to the cloud for [.gov] systems way back in 2005.  

So, here are disparate posts that associate. There will be more. 
  • This is from the Communications of the ACM, however it's based upon a report by the DOD of the US: Human Intuition and Algorithmic Efficiency Must Be Balanced to Enhance Data Mesh Resilience. I love the image, somewhat. The human as the key is a great decision. Too, the bright spot (say, our Sun and its wonders) represents many things. But, the mesh is without limit in its meaning. All of the GenAI and about everything that is computationally framed is now using linear methods of mathematics. Let's look at that (below). So, there is no AI in the buckets of bits. We have used "AIn't" are are serious in that. This article references the U.S.Army, but we will go into the original government studies. BTW, be aware that the Air Force has a One Model project now that seems to be conttra to this theme by the DOD. Well, not exactly, as it deals with the coming emphasis (actually old) on simulation. 
  • Herbert Simon was there at Dartmouth in the '50s meeting. The link is to his retrospective and has a few pages that one can preview. So, AIn't can be related to his work and that of many others. They, of course, have the right to use AI; I am merely calling them on carpet for washing over the details that are important. These things can be explained better. It's time for mature folks to step up and do the work. 
  • Along that line, I was reminded of Simon when reviewing this blog of Brian Keng: Book Impressions - 2019. He mentioned his reactions to Simon's book. I went to look for an on-line copy (prior bullet). Now, Brian's blog is Bounded Rationality. Brian is a professor who teaches the techniques of machine learning. Also, he plays around with and writes about this field. Too, he explains the mathematics as it comes up. Example: Tensors, Tensors, Tensors. You will see this topic in lots of the technical discussions. Some of this interest came up with AlphaGo which got a lot of press for Google. After all, that was major; we have to admit that we took a pause to see what the clamor was all about after all it dealt with games and not real life. (Note: as we go on, yes, that theme is there. John is going to use a patent write up and his experiences on a project to discuss knowledge based engineering and how it could have helped GenAI in a mature mode. But, those types of mismatches are generational. Learning how to manage this is on our (the whole of the planet) table. Anyway, Nvidia is the hot topic now. Lots of the machine learning progress came from their GPU which was built for gaming and graphics being converted over to handle large linear systems. Tensors, of course, are a generalization of this. And, by meeting the constraints of the assumptions (problems there, okay?), one gets good bookkeeping in computation (meaning, less programming and testing) plus the control mechanisms that are associated as the system does the work of remembering and tracing through all of the computations so that one almost can forget of vertigo and other types of lack of decision choice that is amenable to resolution on the fly. So, Brian's work is wonderful in operationa and as an example. 
  • Want to see an older attempt which still resonates? How about Continuum Mechanics and Fracture Mechanics. Bob McGinty is a retired PhD who worked as a PE. In the CM blog,he goes into the basic mathematics which involves the level attained for an engineering bachelors. I said older, but Bob's pages applied the advances of the time as he notes: 
    • Two relatively new web technologies are used on these pages. The first is Scalable Vector Graphics, or SVG. Pages on this site will display SVG files in compatible browsers, and PNG files in incompatible ones. The advantage of SVG over PNG is that SVG graphics can be scaled to any size without the onset of pixelization. SVG files used here were created using Inkscape, an excellent graphics program available free on the internet here. 
    • The second new technology used here is MathJax, a Javascript based display engine for mathematical equations programmed in the LaTeX language. MathJax eliminates the need to display equations as GIF or PNG graphics files (or even SVG for that matter). 
Of late, we have noticed stale pointers. Some of these come quickly due to the use of temporary links in various web sites. After all, nothing hangs around forever. Others of the missing pages are due to sites disappearing. We have paid attention and will made attempts at presenting material so as to lessen the event of the infamous 404. Of course, 404 (and related) handling will be part of the work, to boot. 

Remarks: Modified: 06/26/2024

06/26/2024 --

No comments:

Post a Comment