T Is for Technology and Technique

/

In my last post, I noted the value of acknowledging the hidden talent offered by editors and reviewers. Today, I use a combination of talent, technology, and technique to think through the benefits of acknowledging our workflow, including AI. (I was happy to hear Microsoft mention “prompt history” as a tool on today’s live demo of the Copilot integrations - important to recreate our steps easily.)

Show Your Work

Sharing the provenance of our work adds value at the moment and as you return later to leverage the ideas or make them better. However, we know from Sidney Harris’ famous cartoon of scientists working on a blackboard (“then a miracle occurs”) to math assignments worldwide, showing your work is complicated. 

Hyperlinks give us the ability to share our sources without much distraction. Various in-line academic referencing techniques (APA, MLA, etc.) offer value to insiders as we can translate “DeSanctis & Poole (1994)” in our heads and place the related material in context with decades of research. Footnotes are slightly less distracting, and endnotes push the connections out of sight. 

Yet, even traditional citations don’t tell us how we made the connections. Formal research writing is such that you should be able to replicate the work. Even so, we are in the midst of a replication crisis. Some of that is due to sketchy research. “Lucky” results are another issue. But some of the crisis must be due to limited documentation of the process. When we run an experiment, we don’t always fully describe the room setup, the software versions, or the particular timing -- we used to be better at this.

Space Constraints (Really, Attention Constraints)

I end my January piece in The Conversation: Why using AI tools like ChatGPT in my MBA innovation course is expected and not cheating with "I like an approach some photographers use: share your tools and critical settings." Susannah Schmidt, my editor on that piece, suggested I take the idea further and fully document the approach I used to create the article.

Sharing the process even just as bullet points turned out to be about one-third again the length of the article. No-go, given the word count limit for The Conversation, but here I have all the space in the world. I also respect your time, so I haven’t gone completely overboard in the notes below.

Show Your Work To Add Value

Generative AI and other AI-enabled tools can be of great help. We are all experimenting in these early days of publicly available AI. AI learn, change, and our practices continue to evolve. I expect many of us to reinvent our wheels. 

Yet, learning how to use these tools is difficult without watching someone work through the process. Observable innovations have a greater chance of adding value, but knowledge workers work on their screens, not in public. We must make our experiments visible to learn from our new workflows. How to share the process helpfully for our audiences needs examination. 

A 5T Analysis

When I talk about Thinking in 5T, I think of the interplay across Talent, Technology, and Technique as we work toward a Target and stay aligned with our Times. 

Talent

In the case of AI use for knowledge workers, talent can start with the person experimenting with AI in their work and wanting to share that workflow with others. The first step is to acknowledge that humans are easily distracted. Tracking our process takes us out of our flow. Find a way to document your work automatically so you can lean toward “freewriting.” If you’re a coder, you may have already found a way to have AI create comments in your work (I hesitate to add a link as the best tools will change -- search for something like: “ai adding comments to code.” You likely automated your reference management years back if you are a researcher. Google, Microsoft 365, etc. automate our version control, but not yet in a way we can learn from. (Am I more productive when I outline first? Maybe Microsoft Viva Insights will someday help me know.)

For the talent trying to learn a new workflow, is a video screen capture the way, or is a textual recipe better? I’ll append my recipe for this piece below.

Technology: When an AI Generates the Work

Unfortunately, the major style-guides haven’t caught up to the AI. Some institutions offer stop-gap advice -- Thank you, Seneca and Guelph-Humber. 

The tricky part is that what the AI tells me, it may not tell you… and it may not tell me the same thing twice. Style-guide-wise, this is a bit like referencing personal communications where we can’t explicitly show you our data as it’s private or ethereal. 

Technique: Current Recipe for My Written Work:

  • Throwing ideas and references/links into a Google Doc with a draft title

  • Both Google autocomplete and paid Grammarly run in the background (just now, I hit the tab key and back turned into background). 

  • When the idea gels, rough out the outline. Outlining is iterative as I self-edit (I’m not very good at freewriting), reminding myself about the goal of the post. Create other idea files to save trimmed ideas.

  • When stuck on finding the right word, cut the paragraph and paste it into ChatGPT, YouChat, or the paid version of WriteSonic. My attempts at just putting an idea into WriteSonic or Tome have yet to pay off with a good draft. Given its access to all my background work, I expect Microsoft's Copilot will do a better job. Please let the months before my access to Copilot pass quickly….

  • I wait until I have a full draft before clicking the Grammarly icon and doing the full proofing pass. (Small concession to the ideas of freewriting.

  • Old world -- at the end, I add in full references via Zotero or Paper.li. In the new world, I’m hoping my copilot will suggest the best references as I type… and then, ultimately, write the review section and keep it up-to-date.

  • Depending on the final publication venue, I’ll cut and paste the material into its new home. For example, this post went from Google Docs into Squarespace — where Grammarly offered still more (and sometimes, different) suggestions.

The Mix

Each of the talent, technology, and technique dimensions is in flux as I do my work. My understanding of each, and the target and the times, shifts as I go. How we reference our AI support (at least until we have “prompt history”) remains a question. For a recent short article, I happily added a sentence from ChatGPT-4’s review of my draft. Going back to T is for Talent, I wouldn’t cite an editor given today’s norms. I didn’t cite that sentence. I would like “version control” to track and help me improve my work through analytics. Microsoft Viva Insights -- are you listening? Or maybe I should say, Copilot?

[I know I can do a better job on the above paragraph, but I have to move on to another task. Enter ChatGPT-4 (though, why does it so love to say, “In conclusion”?) My prompt: Rewrite this paragraph as a powerful ending to a professional blog post. Other than “In conclusion,” and the florid “the ever-changing landscape,” not bad]:

In conclusion, the ever-changing landscape of talent, technology, and technique continues to reshape the way we approach our work. As our understanding evolves, so does our approach to leveraging AI assistance in our creative process. The implementation of "prompt history" could provide valuable insight and streamline our work with AI like ChatGPT-4. Drawing from the lessons in "T is for Talent," we recognize the importance of adapting to modern norms and finding ways to incorporate AI-generated content seamlessly. The future of work could greatly benefit from enhanced "version control" systems that monitor and improve our output. Microsoft Viva Insights and Copilot, the call is clear – we are ready to embrace the evolution and revolutionize our professional landscape.

Same original paragraph, explicit request for first-person active voice:

As I navigate the dynamic realms of talent, technology, and technique in my work, I continually adapt my understanding to keep pace with ever-changing targets and circumstances. The integration of AI support, such as ChatGPT-4, into our workflow poses intriguing questions about referencing and acknowledgment. Reflecting on "T is for Talent," I realize that today's norms may not always align with traditional citation practices. I eagerly embrace the potential for advanced "version control" systems to monitor and enhance my work. Microsoft Viva Insights and Copilot, I implore you to join me in this quest for progress as we strive to revolutionize the professional world together.

Not sure I’ve ever “eagerly embraced,” nor “implored,” but again, not bad.