Practical AI for PMs | 04.18.24

Hi, there.

Ed here.

Last week, I attended a PM conference in Columbus, Ohio, and while AI wasnโ€™t the primary topic of conversation, its implications for our work certainly came up in most conversations.

Thankfully, this week, Iโ€™m on vacation in Park City, UT, decompressing from that marathon of lectures, breakouts, and team dinners.

For those two reasons, this will be a brief issue. So, before I get back out onto the mountain, I wanted to explore the three main themes I took away from those conversations at the conference.

3 Common Themes about AI in Project Management ๐Ÿค–.

๐Ÿญ. AI ๐—ฐ๐—ฎ๐—ป ๐—บ๐—ฎ๐—ป๐—ฎ๐—ด๐—ฒ ๐—ฝ๐—ฟ๐—ผ๐—ท๐—ฒ๐—ฐ๐˜ ๐—ฝ๐—ฟ๐—ผ๐—ฐ๐—ฒ๐˜€๐˜€๐—ฒ๐˜€ ๐˜„๐—ถ๐˜๐—ต๐—ผ๐˜‚๐˜ ๐—ต๐˜‚๐—บ๐—ฎ๐—ป ๐—ผ๐˜ƒ๐—ฒ๐—ฟ๐˜€๐—ถ๐—ด๐—ต๐˜.
AI is excellent at processing an enormous amount of information and producing an output.

At this time, the challenge is knowing what the output will be.

That power must be guided by thoughtful, skilled project managers who can use that output productively.

We are not at a point yet where we can turn AI loose on important tasks and expect a positive outcome on a consistent enough basis to shirk our responsibility of review and oversight.

For more on โ˜๏ธ download my (expanded) AI Eisenhower Matrix Framework here.

This leads us to the next challengeโ€ฆ

๐Ÿฎ. AI can ๐—ฝ๐—ฟ๐—ผ๐˜ƒ๐—ถ๐—ฑ๐—ฒ ๐—ฐ๐—ผ๐—ป๐˜€๐—ถ๐˜€๐˜๐—ฒ๐—ป๐˜, ๐—ฟ๐—ฒ๐—ฝ๐—ฒ๐—ฎ๐˜๐—ฎ๐—ฏ๐—น๐—ฒ ๐—ผ๐˜‚๐˜๐—ฐ๐—ผ๐—บ๐—ฒ๐˜€.
For mission-critical tasks, it is important to remember that AI is a black box.

No one knows how AI really works, or sometimes doesn’t.

You’re not guaranteed to get consistent results just because you use the same prompts or inputs.

I have mostly overcome this challenge in the short term by integrating AI into more traditional automation tools like Zapier.

With a platform like Zapier (there are many other similar tools, like Make or IFTTT), I connect the software and data I already use and leverage AI in between standardized, repeatable steps to transform the information in ways I previously had to do manually.

Future AI models may get more predictable, but for now, creating a repetitive process with a 100% predictable outcome is impossible.

Takeaway: Motivated, empowered humans need to review critical AI output at critical stages of the process.

Soooo, that being saidโ€ฆ.

๐Ÿฏ. C๐—ฎ๐—ป, and will, AI ๐—ฟ๐—ฒ๐—ฝ๐—น๐—ฎ๐—ฐ๐—ฒ ๐—ต๐˜‚๐—บ๐—ฎ๐—ป ๐—ฝ๐—ฟ๐—ผ๐—ท๐—ฒ๐—ฐ๐˜ ๐—บ๐—ฎ๐—ป๐—ฎ๐—ด๐—ฒ๐—ฟ๐˜€?
It won’t if you want to build strong relationships with internal and external stakeholders.

As weโ€™ve become more dependent on data and larger data stores can be analyzed by AI, keeping people motivated and feeling valuable within an organization will become even more important.

AI is very good at producing statistically โ€œcorrectโ€ outputs. Itโ€™s up to human understanding to interpret and execute on those outputs.

The soft skills required to build and maintain these strong relationships are uniquely human and will quickly become one of a project manager’s most valued qualities.

So, no, AI wonโ€™t replace project managers as long as we keep improving the soft skills that truly separate us from the computers.

tl;dr ๐Ÿฅœ:
Focus on building soft skills, prioritize empowering human oversight of AI outputs, and remember that project managers are in the people business as much as we are in the tasks business.

๐——๐—ผ these things, ๐—ฎ๐—ป๐—ฑ ๐˜†๐—ผ๐˜‚’๐—น๐—น ๐—ณ๐—ถ๐—ป๐—ฑ ๐˜†๐—ผ๐˜‚๐—ฟ ๐—ฃ๐— ๐—ข ๐—ต๐˜‚๐—บ๐˜€ ๐—ฎ๐—น๐—ผ๐—ป๐—ด ๐˜„๐—ถ๐˜๐—ต ๐—ฝ๐—ฒ๐—ผ๐—ฝ๐—น๐—ฒ ๐—ฎ๐˜ ๐˜๐—ต๐—ฒ ๐—ฐ๐—ฒ๐—ป๐˜๐—ฒ๐—ฟ.

Value Added Resources:

This Weekโ€™s AI Use Case:

Even after only being away for a few days, I know Iโ€™m going to come back to a full inbox. A lot of those messages will be threads that go down multiple paths. How do you quickly catch up on the important messages and ignore the โ€œthank you!โ€ and โ€œyouโ€™re welcomeโ€ messages?

Hereโ€™s how ๐Ÿ‘‡

Title:Quickly catch up on emails after vacation!
Pain Point:When returning from an extended break, there is an overload of email threads going in 100 different directions. How do you organize whatโ€™s important?
Solution:Input entire email threads into an AI tool like ChatGPT to write a summary, extract important data, and list action items for follow-up.
Tools:Your preferred AI tool, such as ChatGPT, Gemini, or Claude. A closed tool is even better if the emails contain sensitive information.
Prompt Tips:1) Make the AI Persona clear (e.g. Executive Assistant) for this task.2) Give the AI an example of the format youโ€™d like for the output.
Main Hurdles:1) Needing to check all outputs for accuracy.2) The possibility of missed information.

OK. Thatโ€™s all. Talk to you next week.

-Ed

If you found this newsletter valuable, please consider sharing it with a friend or colleague.

Did someone forward this email to you? ๐Ÿ‘‰ Sign Up Here.

AI Disclosures: This email’s content was written entirely by me, Ed, a human.

Similar Posts