.png)
AI will Clean up our Data. Right?
- AnalyzeBrand.com
- Jan 9
- 2 min read
Updated: Jan 18
The AI disruption that started with the release of open source ChatGPT in the fall of 2022 continues to disrupt most organizations. GPT stands for Generative Pre-trained Transformer, a technology that leverages deep learning (Burke, 2023). In the race to be first in AI, the big software vendors baked in these ChatGPT capabilities into their tools almost overnight, many without any real thought on how it can add value. In addition, Bova et al. (2023) explain that quantum computers combined with AI will begin to solve intractable business problems, such as large-scale Monte Carlo simulations. So, the AI revolution is now in full swing, and anyone who claims to understand how AI will play out in reality, is merely speculating. Reality never seems to play out as envisioned -- especially with digital technology.
One thing that appears clear is the continued importance of accurate data in this AI revolution. The general language learning (GLM) models that drive general AI intelligence require mountains of disparate data to train their models. Shareef & Picek (2024) state that these general models are already being adapted to work with specific models for enterprise software such as ERP, CRM, and BI. But the quality of your organization's data will be the best predictor for how well these specific models will add value. For example, typing into the AI enriched ERP, "Calculate an inventory quantity for Customer X based on their ordering patterns" will not be accurate if one of their orders was returned, but incorrectly represented in the ERP as a dollar journal entry with a missing quantity. So, for the time being, the old adage in IT of "garbage in garbage out" still holds true, even for AI. Moreover, I've always said in my career that a company's data is a material representation of all of their underlying business processes. If your processes are fragmented and ugly -- your data probably is too.
In my opinion, organizations thinking that AI will magically clean up their data in the model training process could be disappointed. While it's true that great advances are being made with synthetic data creation based on a small subset of data, it is still not the ground truth. In a few short years I may be proven wrong, but for now it is clear that having your core ERP data clean (i.e., customer, vendor, product, and corresponding transactions) will be a competitive advantage for companies. Nevertheless, it will be fascinating to see how companies leverage specific model learning (SLM) models showing up in enterprise software over the next 18 months.
I can help your organization develop a data strategy to get your core data ready for the AI revolution currently underway.
Send me a note; let's start the conversation!
References:
Bova, F., Goldfarb, A., & Melko, R. G. (2021). Commercial applications of quantum computing. EPJ quantum technology, 8(1), 2. https://epjquantumtechnology.springeropen.com/articles/10.1140/epjqt/s40507-021-00091-1
Burke, B. (2023). Beyond ChatGPT: The Future of Generative AI for Enterprises. Generative AI
Shareef, A., & Picek,R. Enhancing business insights: AI based chat toolset for ERP Systems, 2024 4th International Conference on Electrical, Computer, Communications and Mechatronics Engineering (ICECCME), Male, Maldives, 2024, pp. 1-6, https://ieeexplore.ieee.org/document/10796242/authors#authors