ERP Treasure Island on the Web
No one learns how to drive a car theoretically. People learn about the mechanics and traffic regulations from books, but they need to drive a real car on real roads to learn how to actually drive. Engine technology, car body construction, traffic rules; these things can all be found online and can be absorbed, copied, and classified by any AI. But someone can only be said to truly know how to drive after experiencing thousands of kilometers on real roads under different weather and lighting conditions—something an AI can never expect to surpass us in.
Creating a perfect ERP system is theoretically a simple task. If someone were to have the time and the energy, they could read all of the business administration literature available on the web. Data on organizational theory, financial systems, logistics, and production can also be easily found online. So it is likely that the entire scope of ERP, SCM, and CRM knowledge of an SAP system, including the programming instructions for Abap, can also be found online. If these “book smarts” were supplemented by contributions from the SAP community, the result would be an S/4 successor that is likely to be universal, i.e. a successor that takes into account all the information that S/4 currently lacks, all the missing functions and modules that necessitate plug-ins from third-party providers, and creates a more complete program.
On the subject of AI and ChatGPT, the web version of the German magazine Handelsblatt had the following to say: “In trying to push its own limits, it continuously discovers unexpected abilities. For example, ChatGPT can write code on command in the Abap programming language, which is only used in SAP applications and is thus usually reserved only for specialists.”
Go like ERP
A few years ago, Google experimented with an AI that was supposed to master the board game Go, a strategy game invented in China over 2,500 years ago. After a relatively short training phase, the Google Go AI, called AlphaGo, was better than any human player and made moves that experts thought were crazy—yet the Google machine won in the end. But it got even better: in a second attempt, the AI was given only the most basic Go rules. From then on, the machine played non-stop against itself and, in a few months, had read through the entirety of Go literature from the past millennia.
It should then be a simple exercise for ChatGPT to become an ERP genius and know the answer to every problem with the extensive breadth of business, organizational, technical, and licensing literature available online. Transferring this knowledge into Abap and testing it should be possible within the next few months.
An AI that can program the best ERP system of all time currently seems like a pipe dream. However, if we abide by the thought that all the knowledge for a perfect ERP system is available on the web and ChatGPT has access to it, then it only appears to be a question of available resources. Will SAP be willing to license ChatGPT and run it on an appropriately powerful supercomputer, and present the perfect S/4 successor to the world in a few years? It remains to be seen.
Any questions, comments, or concerns? Feel free to leave a comment below.
LOL you could start writing fictions your posts now do not make any sense or add value. It’ s like a dejected old man rambling in the bar I saw an UFO.
Could I ask you to elaborate on which points do not make sense or do not add value? The use of AI/ChatGPT in SAP software is both a current and relevant topic within the SAP community.