The creative part of me abhors generative AI. I write fiction. I have friends who write fiction. I have friends who are artists. Generative AI purports to take food off their plates.
And the scary thing is, as crap as AI generated art is today, it is starting to get better.
But there is a use for Gen AI in Change Management, though there are serious risks that need to be addressed. I'll get to the risks in a couple of paragraphs. First, the use cases.
Documentation Templates
Because LLMs have a vast amount of source information (IP rights aside), the tool is good for providing Change Management Plan templates (basically, a table of contents showing typical content), communications plans, FAQs and other types of documentation. Drafts only, of course. It would be foolhardy to take something an LLM spat out and publish it without human intervention.
Communication
Gen AI can produce draft messaging for emails, newsletters, intranet posts and suggest appropriate delivery channels. It can also tailor comms for stakeholder or persona groups. Again, drafts only.
Other Stuff
It's been suggested that an LLM / Gen AI can produce stakeholder maps, create board reports, risk registers and meeting summaries. Unless you're using an internally-built LLM with linkages to corporate structure and business tools, it's better to built it yourself. The effort to create these artefacts from scratch is probably less than the effort required to craft prompts to create something useful from an LLM.
Gen AI comes with risks, though
All new technologies need to be viewed critically for limitations and risks. Check your organisations AI policy and make sure to follow its guideline. If your company doesn't have published guidelines, here are some (guidelines) to consider.
Privacy
Charging hard, and first, out of the gate is privacy. Everything -- EVERYthing -- you send to an LLM (or any Gen AI tool) is kept by the system and can be used to generate responses for ANYONE ELSE.
Do not share critical, sensitive, confidential, marginally confidential or any other information you wouldn't want your closest competitor to have.
Hallucinations
Gen AI makes stuff up. If an LLM can't find an answer to a question, there's a non-zero probability of constructing something out of thin air.
I have more than anecdotal evidence of this. In my other guise I write fiction. I had assigned one of my ISBNs to one of my books coming out later this year, and for kicks I asked ChatGPT what it could tell me about the ISBN:
I had a momentary freak-out that the ISBNs I bought were not unique (that lasted about two seconds), then looked up the ISBN for The Labyrinth. Not the same number. I challenged the robot:
And I asked why it gave me an incorrect answer:
Bottom line, do not trust. Alway verify. LLMs make shit up.
Zero Emotional Intelligence
Identifying the source and cause of resistance is a human perception thing. An LLM might hazard a guess at where resistance might lie, but it's 100% guess. A person, living, breathing human, needs to identify the source and cause of potential resistance, and find ways to mitigate it.
And then there's the Trust of it all.
Change is human-centric. And if you're telling stakeholders that you're a human-centric change practitioner, you'll lose all credibility if those same stakeholders think you're using Gen AI to produce canned communications.
Comms will come across as inauthentic, fake, probably boring and missing the point unless the LLM is used (if it is used) to DRAFT communications only. Respect those impacted by the change and actually deliver meaningful communications that address their concerns and elevate their excitement.
Or lose their trust.
Bottom Line
Gen AI can save some up front time drafting structured approaches to change problems, but should absolutely not be relied on for final results. And make sure you've following your organisation's AI policies. It seems like a powerful tool, but it can bite you in the butt if you're not careful.