If you haven’t heard about generative AI (GenAI), or even been pitched some type of it, then you’ve been hiding under a stone. There are dozens of articles describing GenAI, so it’s unnecessary to write extensively on how it works except to say it is a way of automating the creation of content. Instead, I’ll provide some of my thoughts on the growing impact of this emerging technology.

First, it’s not all GenAI. As with any technology with this type of hype, marketers and solution providers are using the term rather liberally. Any solution integrating GenAI gets the “GenAI” label even if it’s only a small part of the actual solution. In many cases, it’s rules-driven conditional workflows, analytics, and templates doing the hard work.

That said, GenAI is a great way of generating content, whether it’s from scratch or summarizing existing documents, and we do quite a bit of that in procurement. But in many cases, actually creating the content is a necessary part of the process. I could probably use ChatGPT to write this article but the process of writing it forces me to formulate and structure my thoughts. Let’s look at a couple of examples.

Creating a Category Management Strategy

In a previous article about category management, a concern I highlighted was that category strategies, if created at all, are often a stand-alone effort and then updated (maybe) now and then without being embedded in the actual sourcing and procurement work. If category managers lean too much on GenAI to create category strategies, they risk it becoming a check-the-box activity. To create an impactful strategy the category manager needs to truly understand the category and its market and stay updated on changes and developments. Also, the interaction with other stakeholders to shape the strategy is critical to drive adoption and credibility.

But this does not mean GenAI has no role to play. A set of guided prompts can retrieve and summarize information that underpins the strategy as long as the quality and accuracy of the information is controlled. Among other things, it can be used to create an executive summary of the strategy. As long as the category managers retain responsibility and accountability for the strategy and execution, then GenAI can assist. But using the technology to fill out Porter’s five forces template for the sake of completion could create more risk than value.

Creating an RFI

Another use case is using GenAI to create questions in an RFI (or the RFI section of an RFP). Is this a concern? If you ask suppliers, most would say the number of questions in the average RFI is too low.

RFI questions should enable comparison and qualification of suppliers. The buyer in collaboration with key stakeholders, should have enough subject matter expertise to avoid superfluous “good to know” questions. If they rely on GenAI to create the questions, a clear lack of necessary knowledge exists.

For more general questions, templates should be used to create consistency and enable cross-category comparisons and reporting (and allow the information to create or update a supplier profile in a SIM solution). Using GenAI to create these templates might save some time but given that this is more or less a one-time task, that time saving is minimal.

GenAI is better deployed to support suppliers with responding to RFIs. Use the solution to prepopulate answers that suppliers only need for verification and allow them to upload documents that GenAI can summarize. GenAI can also summarize a series of text questions. This brings to mind a recent cartoon where a person asked a GenAI solution to generate an email covering three points only for the receiver to use GenAI to summarize the email into three bullet points. The lesson? Ask suppliers questions that encourage succinctness, avoiding the need to use GenAI to summarize.

Use GenAI as an Assistant

At its core, GenAI creates content. In the case of procurement, this is most likely text generation. The GenAI predicts wording based on given prompts (instructions) coupled with an enormous amount of trained data. It’s really good at mimicking human writing, but what it writes is not necessarily factual even if it appears so — this is called hallucination and there are several ways of limiting this  problem).

This means, as Alan Holland at Keelvar (and probably others) has stated, you shouldn’t rely on GenAI if the cost of error is high. Rather, use it as an assistant to improve efficiency when dealing with larger amounts of text. There are likely hundreds of good use cases for GenAI to provide tons of value if it’s governed properly. Combine it with other technologies, both AI and others, and we will see some cool stuff in the future that we’ll surely write about. Just don’t use it as a shortcut to not do your job or an excuse to not use your brain.

RELATED RESEARCH

Magnus Mondays — Category Management: All Talk No Action?

CPO Rising Listicle: Unlock More Value with Five Category Management Benefits

Artificial Intelligence: Making the Procurement World Smarter

The Key Themes for the Modern CPO’s Agenda (Artificial Intelligence)

Artificial Intelligence for Procurement Gets a Reality Check

Tagged in: AI, Artificial Intelligence, Category Management, generative AI, Magnus Bergfors, Magnus Mondays