AI is the big idea, and machine learning is the engine that powers it
Jim Paschke
We smooth your S/4 path by shrinking your HANA box where other providers fall short. We consistently reduce HANA box sizes by at least half! Works for small and medium shops as well as our Fortune 500 customers.
By Jim Paschke, CEO of PCC
February 10, 2025
This article’s title is coined by Ruben Circelli, published yesterday, February 9, 2025 (before the Super Bowl began so he likely didn’t miss it). In his article AI vs. Machine Learning: The Key Differences and Why They Matter; he describes Machine Learning (ML) as “powered by hubs of interconnected computers or supercomputers processing massive quantities of data to train a program to give a particular output with a given input.” (italics added). I like to think of ML as the engine that creates the data lake, loaded with lots of data waiting to be analyzed and converted into useful information by the right AI model.
Up until recently, that analysis was usually done by Human Intelligence (HI), someone running a report that summarized totals, averages, etc. Now with the proliferation of Artificial Intelligence (AI), we have advanced models and heuristics available that allow more decisions to be made by AI but only to the extent the human is willing to delegate the decision making to the AI. We use Siri or Alexa and laud the advance. But none of us are likely to allow AI to make the weightier decisions like switching jobs, relocating, or (for the CEO) when to acquire, divest, or fund.
On January 17, 2025, the Geeks for geeks website published an article called Machine Learning Lifecycle, in which they describe ML as “a process that guides development and deployment of machine learning models in a structured way. It consists of various steps.” Their 10 steps are:
Some developers look at that list and see it’s not altogether different from most SDLC (systems development lifecycle) models. As I see it, the AI model varies from SDLC in two key areas: the importance of data in solving the problem and not just the problem (flow charts) and the importance of the model and its selection, training and improvement.
I like to break down the 10 AI steps into 3 phases which illuminate when you’re making the transition from traditional SDLC into the Machine Learning paradigm: 1) problem definition, 2) what data to apply to our problem and how, and 3) Modelling the solution from selection and creation through learning and monitoring. Both the SDLC and AI development models begin with problem definition. From there, they quickly diverge. Whereas historical SDLC generally applied a process to a certain set of data, e.g., process payroll, plan deliveries, etc.; in contrast, AI addresses what questions you want answered independent of the data, e.g., what will payroll be next year, how can I improve my delivery scheduling? AI has the ability to answer bigger picture questions like cash planning and financing, and these come closer to the Strong AI category. However, just like everyday decision making, we will continue to use Narrow AI to answer questions like how much the payroll is this week and what’s the best route for these deliveries.
In summary, systems development still occurs in organizations every day. Whether it occurs via a traditional paradigm or uses a ML/AI approach can be open to interpretation. Circelli says “there isn't a scientific body that decides what is or is not, technically, AI; the term is defined by whoever is using it.” For the person using it, that’s fine. As developers, we now have clear steps to take when developing solutions that show that we are (or are not) on the ML/AI path. Using ML to create the data lake to drive an iterative AI model is key in the next step of developing true AI systems. Implementing these in our systems today is very exciting and I can’t wait to see where this will take us.
PCC is an SAP Partner company whose Software Solution, PCC Connector, applies AI-powered algorithms to the SAP data archiving reporting challenges in the 21st century ensuring the entire organization’s archive data store is readily available to the entire organization thus conjoining database data and archive data in an indivisible dataset. Jim Paschke has been in the SAP data archiving world for over 25 years and is the founder and CEO of PCC serving many Fortune 500 companies since Jan 2020.