Pages

Sunday, September 6, 2026

KBE_ICAD_MSJO

TL;DR -- GenAI/LLM was on the horizon and now is actively involved day to day. Concerns for advanced software continue which need attention. Pushing forward can cause a break with former methods even if they are crucial to necessary qualities. Let's look at recapturing some of the former ways in terms of future stability. 

--

Since Nov of 2022, the world has adapted to the biologically-framed "nightmare" of GenAI/LLM. If you have not paid attention, we'll summarize. There are post here going back to Feb of 2023, when I first heard of the event of Nov of 2022 and went to look at ChatGPT and others. 

Why was I not paying attention? My first reaction was, what? Idiocy. One AI Researcher of note said: this is a train-wreck waiting to happen. Some jumped on the bandwagon. I went back to review all that I learned the past 20 years as computing morph'd to use the gifts of DoW/D, namely TCP/IP, GPS, and  more. Like, the GPU. And, more. 

Lately, there is a post, who pays (9 Aug 2026)? We have to look at that. We said, technology will be our focus. It has been for a while. But, we're getting more specific as time goes along. 

This image shows a Gemini response. Note, it's labelled. I am not using AIn't (with its 'tAIn't) for writing or research. This was merely to see if work of the past few weeks has taken hold. 


These three items will be heard of constantly. A few decades ago, AI (with the provenance going back to Dartmouth) had a good handle on things. One of those things was mathematics in the applied sense as it pertains to things like tools of design and analysis (CAD, CAE, ...), machining (CNC, ...), and more. 

While CAD vendors were working to close the gap between the 2 1/2D state and full 3D modeling, people had to still do the work of modeling. And, one project pushed the envelope with respect to model requirements that had to meet the strict was of solid geometry. ICAD helped in that case. And, MSJO (multiple-surface join and offset) was a particular approach to allow hugely complication collections of surfaces to be pulled into the framework of a solid model while meeting the requirements of fit, form and function. 
 
One caveat? We didn't try to build a general intelligence. Nope, the focus was to support engineers in their work while honoring decades of software development of a mathematical and engineering nature. 

More to come. The "agentic AI" of late is full of errorful ways. We were able to meet precision necessities and to actually develop and deliver complex products, some of which are still extant in term of superior performance. 

That type of thing ought to be the goal anyway. 

Cornerstone project: Truth Engineering

Remarks: Modified: 09/06/2026

09/06/2026 --