All that testing made the daily weblog statistics look way more healthy than usual. During the course of writing I tend to write a general stream of consciousness at this time of the morning. It’s before daybreak and the writing is really just getting started. Things are beginning to move along and the formation of arguments has not really become a part of the day. We are in the philosophy that stands apart from the deeper reason section of the day. Don’t worry that condition does not continue for very long; the arrival of espresso shots tends to sweep it back under the rug in the middle of the grand room of consideration. That is probably where it should be for the most part. It’s good to ponder the nature of philosophy, but that cannot make up the majority of my day or things have gotten pretty far off track. 

Over the last 10 days, I have been making regular posts to the weblog in the form of just old fashioned blogging. Most of it has been at the start of the day, but some of it has been at the end of the day. It’s not as much planned with purpose as it is a reintroduction of a daily routine. I’m not entirely sure what the end result of that type of effort will end up being. The writing process normally begins with some thoughts about writing or routine. Generally speaking it’s not very focused and more or less just happens until something shifts. You could call that the spark of innovation or maybe the thing that we are here for in the greater effort that is writing. My thoughts move from what has happened over the last two paragraphs into the next layer of focus. Yesterday for example I really was thinking about just how many billions of dollars are being spent to commoditize the delivery of AI features into products. 

A tremendous amount of money is being spent to create what will end up being featured in products. These features won’t really differentiate the product as any product could deploy them as they will be generally commoditized. This is the argument that I started making about AI/ML over a year ago and it to some degree changes my writing output on the subject. I think people are generally getting it wrong when they think that all of these models, agents, and systems are going to be used to build more and more software. That view probably misses the point that system software is just an abstraction from the data layer and business rules. It’s entirely possible that a well designed and deployed model could interact with the underlying data and be the point of interaction with the end user. Zoom out for a second and consider that instead of having to build a huge amount of software the model would present or interact with the end user in real time without having to have pregenerated that entire software layer model for presentation. 

Major paradigm shifts like what we would have seen in the television show Star Trek coming true where people could just ask the computer to do something and interact with it that way are not very far away. Deployed models with some degree of agency could complete these types of interactions just like the computer did on the television show that dramatized science fiction. This shift is a radical one based on where we are right now given just how much software is being written, maintained, and ultimately abandoned on a daily basis right now. Facing a potential shift away from the software layer to the enabled agent layer of interaction is an interesting thought to consider and examine at a deeper level. It does make me wonder if an immutable data layer would be required to manage knowledge and fact. Beyond that layers of user data or just accepted data would have to exist as well. It’s very different from a traditional software framework where the data would be defined and not fluid or changing depending on the interaction. 

Dr. Nels Lindahl