Human intuition

TL;DR -- In our technology thrust, AI is a regular feature needing attention. Recent papers suggest that a proper maturity is being sought by players of importance. One of these is the US DOD in this case represented by its US Army with respect to issues of using data from research, classification and control. 

---

This post is short for various reasons. We just want to report some recent articles plus another from 2019. These deal with humans and artificial intelligence. Or, we'll use machine learning for the last one. 

Essentially, discussions come down to one question: who ought to be in charge? Humans have been for all of time that we know, some humans that is. And, there are many things that we do not control. Just look at the storms (tornados) that rip across the interior of the U.S. on a regular basis in the spring. 

There is a new player, though, the machine. Hence when it learns, marvels seem to appear. We'll skip over that as the press has been full of stories along this line for almost two years now. For one thing, we just remind everyone, there is no creature emerging from the antics of artifical modes such as we see with heated circuitry. Too, "smarts" is debatable. Yet. 

There has been no end to the discussion. We are leaning toward the thought that humans and their talents are such as to be of importance in this matter. That has been on the table for a while. We might just be seing evidence that technology can help us in the matter by establishing the ways and means that people can use to raise their focus and involvement.

So, let's look briefly at three articles. 

• Demystifying data mesh -- this article is from June of 2023 which is recent and targeted toward the C-suite. McKinsey is a consulting firm of note. They discuss "data mesh" which comes out of data science which is a type of predecessor of machine learning. There are several key themes to the topic. One thing is that machine learning is highly dependent upon the quality of the data that is being used to teach it what is supposed to know. Another is that the mode of control, whether decentralized or centralized, has a huge impact. One theme is that business purposed need to keep the human element involved. For one thing, decisions need human involvement. Then, usual practices that are human-oriented must come into play. 

One might ask, where did this notion come from? 

• How to Move Beyond a Monolithic Data Lake to a Distributed Data Mesh -- from May of 2019, this article was written about the issues of data science which evolved to be the thingc concerned with data and its managment. It addresses the importance of viwpoint; as well, works like "trustworthy" and "truthful" are used. In 2024, we all remember a year of trying to deal with and to understand the "hallucinating" world of machine learning in the guise of the large language model (LLM). The author stresses human involvement as essential. 

Where would such insightful thoughts find its influence?  Late last year, the USDOD announced its guidelines on AI. In that report, machine learning was noted to be problematic. Hence, people with training and expertise were seen to be the key factors in maintaing stability as a way of computational life. This month, a team of engineers for the US Army brought their focus to the public. 

• Human Intuition and Algorithmic Efficiency Must Be Balanced to Enhance Data Mesh Resilience -- from May of 2024, this article stresses that humans need to be involved in decisions made by a machine. This involvement can range from small for inconsequential, regular processing. However, as the stakes rise, so too ought the human oversight. It is interesting that "intuition" is utilized. This article references the one of the former bullet. The context is military which brings in imporatnt differences not considered in normal business, for the most part. 

These three are merely represenetative, as many other articles on the subject have come to fore. In short, they represent a normalization of thinking after the shock of the recent release of AI. ChatGPT is an example. Now, these are being called GenAI. Some use LLM for the language processing that is involved. 

The balance for
data mesh resilience 

For the most part, these approaches are applied mathematics. Some of the issue that arose are due to lack of understanding of the underlying operational approaches which can be overcome by education and public discussion. It is refreshing to see that the complicated world of the military hs the insight to note the importance of human talent and training. 

But, taking the U.S. and its 250th coming up, we can look at the military experience as a whole and consider how the culture has changed. The U.S. military will face even more complexities as time goes on. Using machine learning as an assitance ought to be a natural step forward. These papers suggest that and ought to be on everyone's table for reference.  

Remarks: Modified: 05/15/2024

05/15/2024 --