Why AI Will Not Replace O&G Land Administration Analysts Soon
ChatGPT 5 assisted. In a nutshell, the estimated number of work hours needed to produce only a simple prototype for a simple horizontal allocation well DOI is between 700 and 1,200 hours. Divide those by 40 hours per week, and it calculates to 4 to 7 months. But a prototype is not suitable for rollout. To create a minimally viable product (MVP), it will require an estimated 2,500 to 5,000 hours. A fully completed AI application program would take 6,000 to 12,000 work hours to complete, ready for rollout. There are 46 work weeks in a year after deducting holidays and average vacation time for the experts who will be doing this work. At 40 hours per week, that would be, at best, 16 to 32 months for an MVP and a whopping 3-1/2 to 7 years for a fully developed program ready to handle any DOTO. But there’s a kicker. According to Moore’s Law, technology doubles every 2 years! How many iterations of the original scope of work will be needed to complete a final product? How many analysts would have been required to do the work in the first place, forget AI? What is the comparison? It doesn’t end there, either. Other costs beyond the computer programmer are the SME (subject matter expert) who must work with the programmer to give direction on the substantive knowledge needed to code functionality accurately. In addition, the AI application being built must be trained in iterations—training necessary for it to advance to each new level of complexity. Then add to that the separate, stand-
This article discusses the super-mega work hours (and matching capital investment) needed to build a working AI program for only one major division order analyst task, that of creating a loadable revenue DOI for a Texas horizontal allocation well. A basic AI model must take data directly from an attorney’s Division Order Title Opinion (DOTO) into Excel and then the analyst applies company policy to owner coding based on title requirement decisions. What computer language coding and AI program training would this require? This article reveals the estimated human-time requirements from prototype to enterprise. It expands to give perspective on why auditability and Texas- specific math (producing lateral-length or alternate sharing-box allocation, NPRIs/ORRIs, split-streams) demand what is known as an AI + deterministic rules hybrid. The capital cost? If undertaken by each producer in our industry independently, this article shows how our national debt might pale in comparison to the cost. Executive Takeaways Every company’s CFO likely agrees that human resource costs for any project usually is the main cost driver. It captures the largest soft capital (sunk costs) expense for any project. When considering the viability of creating an AI application program that will do only the work of analyzing a DOTO and translating it into an Excel DOI suitable for upload to the database, work hours come first on the list.
To gather some facts and figures for this article,
27
G rowth T hrough E ducat i on - J anuary / F ebruary / M arch 2026
Made with FlippingBook Ebook Creator