February 2 – 9
My “weeknotes” capture events, thoughts, and other items from the past week, mostly focused on work. Learn more about the weeknotes concept here.
The Calendar™
Back from vacation, it was another wall-to-wall week at work. The calendar had almost no daylight peeking through (until a cancellation on Friday morning, which then turned into impromptu meetings anyway).
Yes, I color-code my calendar so I can see where my time is going or to switch my head into the appropriate mental space repeatedly during the day…
- Red = Executive meetings dealing with organization leadership matters
- Orange = GX Foundry meetings (digital services teams)
- Purple = Delivery Services meetings (business analysis and project management teams)
- Dark Blue = Organization-wide meetings that don’t fall into one of the prior categories
- Dark Green = Project-specific meetings
- Light Green = Professional development (a little “me” time)
- Yellow = Interviews or related hiring activities
- Dark Gray = Blocked-out times to ensure I keep some breathing room in the calendar for my sanity and to do smaller catch-up things, or personal appointments
Maybe I should preserve a snapshot of my calendar each week. It’s really quite remarkable, especially when I catch glimpses of the calendars of most of our staff, which are blissfully empty by comparison.
What my calendar does not show are the evening and weekend hours when I’m catching up on email or working on documents that I can’t handle during the day—because of all the meetings.
This week in #weeknotes
It feels like I’ve turned my weeknotes habit into writing a column rather than capturing highlights from the week. So before I get to my essay (yes, there is one), here’s a quick review of notable stuff from the week.
Lights out. We already had a major power outage event planned for early May, but now it would seem we’re doing another one in just 2 weeks. Our old building has issues, and some of our infrastructure is tied to a building that has not been well-maintained over the decades. So we have to make arrangements. Fun.
Painful project finished. Painful project starting. In local government you can get some real characters that wedge themselves into a niche and act like trolls asking you 3 questions when you want to cross their bridge. We finally got past one particular troll on Project 1. But we have Project 2 starting up shortly, and the trolling has already begun. Ugh.
The Unconference goes wide. Our organization does not have regular town halls or all-staff meetings. We really just have 2 recognition events a year (initiated by yours truly) and any other all-staff meetings are scheduled ad hoc. Until now. The Engineering Unconference I started years ago is about to become an all-staff event, meeting every 3 weeks and covering whatever topics the staff would like to see. The engineering teams might feel like their special time is being taken away, but in fact they will be able to nerd out together in this meeting, but we’ll need to make space for other topics, too. It’s gonna be interesting to see how this plays out culturally.
Interviews. We’re still interviewing for a project-focused role. Completed more interviews this week and there are more next week. I sometimes wonder whether I should take my interviewing / candidate review / job posting skills into the private sector and make some money from it. One of my colleagues is a former professional IT recruiter, so it seems like we could team up and make a killing together. Despite decades of trying, the software makers with their AI tools just cannot get this right.
Everyone has feelings. If you had told me 10 years ago a lot of my time would be dealing with people’s feelings at work, I would have laughed in your face. Or maybe I would have run away screaming. But the amount of time I spend each week hashing through reactions, feelings, thoughts, misunderstandings, assumptions, and more just blows my mind. I can’t go into details here because I know people at the office sometimes see my stuff, but yeah… it’s crazy. None of the situations I engaged with this week were shocking or weird—it’s normal people-working-with-people stuff—but it’s wild that around every conversation there’s some undercurrent of drama. I wish I could write openly about this stuff, but… you know… drama.
A warning shot across the bow. Speaking of drama, I prepped a presentation this week that I will deliver to whatever customers are willing to show up at our monthly customer-focused meeting next week. It’s about how our Employee Portal will be terminated (and not replaced) by the end of 2025, and how we need folks to engage with us around transition plans. In this case we’re going to release a warning banner onto the Portal that will tell end users of the termination timeline (because the people we’ve told so far almost never share info with their employees). It’s gonna ruffle a lot of feathers. But time is a-wastin’ so we need to tell users directly of the change.
Analysis is hard. We’ve been building up our business analysis capabilities in recent months, converting and adding roles, creating document templates, and setting up processes for engaging with problems and developing proposals for solutions. It’s going… slowly. But one effort reached a near-conclusion this week. We’re getting there. I’m looking forward to when all this flows smoothly and quickly. We should be able to complete analyses for fairly standard problems and turn them into solutions in a matter of a few weeks, rather than months. In fact, we have to, or else this has been a waste of time. But I’m pretty sure we’ll get there.
Hey Siri, give me Cliff’s Notes for my job
I was reminded of the old Cliff’s Notes series this weekend after I saw this Bluesky post on the dangers of generative AI in the education space:
This caught my attention for a couple reasons. First, I’m generally on alert all the time for misuse of AI and bullshit claims about what it can do, what jobs it will steal, and so forth. Partially that’s the curmudgeon in me, and partially because I want to know what LLM-based AI tools can actually and reliably do, which requires separating the hype from the reality. But second, I’m concerned about colleagues that may start leaning into AI tools to do their jobs, and what effect that will have on our teams over time.
In fact, I have a colleague that is obviously using genAI tools to create written documents related to their job. These documents are intended to help us understand human and organizational needs so we can address them with process changes, digital solutions, etc. In short, you want these documents to be focused, insightful, and steeped in the vibes of the client so the resulting solution is well-designed to fit their explicit and implicit needs. This requires deep, direct, human-scale understanding of people, motivations, processes, history… to do it well you have to really dig in, and to convey it well you have to really work at the writing part.
If AI is doing the thinking, then what are you doing?
Or, I guess you just point ChatGPT at a few meeting notes and let it rip?
The documents created via LLM are long—which makes them look like it took a lot of work to write—with complete sentences, correct spelling, and the ideas are generally coherent. But they are bland, soulless, and overwritten. They lack any spark of life, insight, ingenuity, or humanity.
They also do not sound like the person that we’re expected to believe created the document.
Setting aside my professional irritation (You really thought you could pass this off as your work? You think I’m stupid?), there are two very real problems here.
- The resulting documents are not usable. We can’t build good processes and digital solutions based on generic ChatGPT notes that lack any understanding of the real world. So whatever “work” was put into generating the documents was wasted because the LLM product is generic, disconnected from factual reality, and lacking in human awareness. It’s also so dull, it’s hard to read.
- More importantly, the more a person uses LLMs to “do” their job, the less capable they are at actually doing their job, especially in real-time, in person, in collaboration with real people. The very thing we employ humans to do in this case—explore and deeply understand messy people-driven processes, tools, and opportunities for improvement—requires actual skill and time spent on the problems alongside real people. If you keep “outsourcing” the writing work to GenAI (to make it look like you did the work) you will fail to build the skills needed for the job, and may even atrophy whatever skills you have.
So we get a worse product, and you get worse at your job. Thanks, AI.
GenAI can inhibit critical engagement with work and can potentially lead to long-term overreliance on the tool and diminished skill for independent problem-solving
michael veale on Bluesky
Write well to think well
Stringing together text can be tedious. Believe me, I know! It’s legitimately hard to convey an idea or multi-step process or a feeling via writing. But if you do the work, you are rewarded with more durable knowledge or awareness or more skills for the next challenge.
I’m reminded of what my father told me as I was preparing for a science fair in 6th grade: “You don’t really understand something until you can explain it to someone else.” And that’s the thing about ChatGPT and Claude and all the others: they don’t understand anything. They’re great at stringing together words and can even use limited semantic processing to make it sound good. But they don’t know anything about anything.
The book Writing to Learn gets at this idea (at some length). Writing is a way to learn things. You won’t really understand something until you can transmit it into someone else’s mind, with clarity, using just words on a page. The act of transcription is an act of discovery.
This is especially true in English, where word order reigns supreme. Other languages are not as sensitive to word sequences or even punctuation changing the meaning of a sentence (e.g. Eats, Shoots and Leaves). Refining your knowledge via the pattern encoder of writing proves you really know something.
So let’s use LLMs to enhance our writing, or to generate new connections in language that we may not have generated ourselves. But if your job is mental processing, don’t outsource it to ChatGPT. That’s like paying a personal trainer to exercise for you.
The Andorian revolution to come
One last thing this week. I attended another Beeck Center gathering where everyone is concerned about the absolute chaos at the federal level and the impacts on colleagues losing their jobs. And that’s just the foreground of the fast-moving-but-invisible coup in progress. This mess is flavoring everything we think about and do right now, even at the local level, and it’s deeply frustrating. Good work is being upended, including the good work of long-time digital reformers. Musk is doing to the federal government what he did to Twitter: dismantle everything, keep on the parts that serve his interests, and turn the rest into a Nazi porn bar.
My fear is that Star Wars: Andor will turn out to be a story of our own near-future, not some story from long ago in a galaxy far, far away. I’m wondering who will give Maarva’s posthumous speech and ignite our revolution against Musk’s empire.
About the header photo
Discover more from digitalpolity.com
Subscribe to get the latest posts sent to your email.
digitalpolity.com 