Technology

WeekNotes for 2024 Week 34: Knitting Machines, Co-Learning Sessions, and Reconnecting with the Work

weekNotes 
Weeknotes are a habit I'm cultivating where I share what I'm working on or thinking about, primarily in my professional life, without worrying too much about the ideas being full-formed.

quotes

“There are moments when you just need to go with the flow (intensive teaching with a fixed schedule, grant application deadlines, etc). But there are also moments where you can claim some time back and use it for something very valuable: reconnecting with your work. Finding what makes you tick, what you care about in your work, what brings you most satisfaction. Figuring out where you can really make a difference with your skills, passion and personality.”

Alexandra Mihai The Time Is Now

thinking about:

Last week we got a knitting machine in the Makerspace, inspired by Quinn Dombrowki’s excellent post about using knitting machines for data visualization Knot Hard: Accessible Textile Data Visualization With a Circular Knitting Machine. While I motly don’t have time to use things in the Makerspace, I like at least try every new thing out before asking students or staff to do the same. I imagine myself like Hammond in Jurassic Park trying to get the velociraptor babies to imprint on him when they hatch.

a still from jurassic park where hammnd watches a velociraptor egg hatch

I decided to visualize our reference stats, breaking then down by technology type, and going back to when we opened in March 2022. I did two tests just to see how the machine worked, then figured out about how many rows were in an inch and that 269 rows would make an acceptable scarf, which meant each row needed to be about 300 reference interactions. I also cut out a couple of categories like “computer prototyping” that I thought didn’t have enough interactions to work well aesthetically, though I wish now I hadn’t done that.

a circular knitting machine with the finished project still attached, a row counter says 269 rows
CategoryNumber of interctionsInchesLinesColourOrderKnit to row
Makerspace only: Tech Categoryincheslinescolour
VR6492.1617.31pink7214.59
Recording Room and Technology5231.7413.95white245
Textiles/Wearables5931.9815.81light blue489.81
3D Printing358311.9495.55turquoise5185.36
Cricut11703.9031.20green131
Tour10713.5728.56red374
Other20696.9055.17dark blue8269.76
Embroidery4471.4911.92orange6197.28

Here is the result! Not perfect, but not terrible!

Doing this also reminded me that I like this work. It was really nice to sit down for a couple of hours in our bright, positive space around staff and students working on their own projects. It reminded me that I love the space and culture we’ve built, and I like this work.

Last week I mentioned that I am developing weekly co-working and learning sessions for the Makerspace, which I am tentatively calling the Sandbox Series, in the hope of inspiring a sense of play without outcomes. Since then, I’ve developed the idea enough I can call it a program, created a website to share information, and scheduled the sessions to happen twice a week this term.

That’s an ambitious schedule, but I’m hoping I can make it work by thinking about these sessions less as stressful events I’m leading that take me away from my work and more like dedicated time to focus on parts of the work that I love, which is connecting with students and colleagues, helping them attain their goals, and exploring my own projects.

working on:

  • Onboarding new staff
  • Developing a new training schedule for student ambassadors
  • A grant to continue the Indigenization work we’ve been doing in the Makerspace, including sponsoring workshops and holding welcome events for Indigenous students.
  • Learning to quilt via 2 workshops hosted in the Makerspace that have been organized and taught by a couple of our Library Technitions.

WeekNotes for 2024 Week 17: “I’m Okay, You’re Okay” Reports and Iridescent Circle of Clouds

weekNotes

Weeknotes are a habit I’m cultivating where I share writing (and some links) as a thinking-in-public process. The idea is to explore ideas I’m grappling with, primarily in my professional life, without worrying about them being full-formed. The potential visibility of these notes is a nudge to develop them a bit more than if they were private. While my audience is mostly theoretical, if you’re reading this, please understand these are meant to be exploratory and provisional.

quotes

“It is easy for me to imagine that the next great division of the world will be between people who wish to live as creatures and people who wish to live as machines.” – Wendel Berry, 2000, Life is a Miracle

thinking about

It’s the end of the academic term, which means many things require reports.

While I am beholden to write reports and sometimes even do them of my own volition, I’m not sure how often they’re read or have any impact, especially if they are required by some process but not by any actual person. I’m reminded of a colleague at a previous job who would joke that our annual performance reviews should take the form of a form filled out by both the employee and their manager that just had a checkbox for “I’m doing okay” and “you’re doing okay” and if everyone agrees everyone is doing okay, that would be the end of the review.

The one beneficial thing these reports could provide, honest assessment and reflection, is often lost by the need to be relentlessly positive, especially if your job or resources for your organization is at stake. Their format, usually formal and narrative, further obscures anything valuable they might contain.

All this makes them one of the things we do that AI is actually pretty good at, and I suspect a lot of us (me included) are using it for summarizing and drafting reports, if not for writing the whole thing. If done poorly, as it usually is, just further removes and obscures what is valuable.

I’d love to just a submit a shorthand “I’m okay, you’re okay” style report for many of these, but I suspect that wouldn’t go over well. But here is my shorthand “I’m okay, you’re okay” END OF THING REPORT for a couple of things I’ve written up recently.

Makerspace End of Term Report

  • I spent less time than I wanted in the space with students and staff, but more than last term. I am working on increasing this by getting out of other commitments, but this is taking time.
  • It feels a little less busy in the space this year, but I don’t think this is a problem (though we should probably do some formal assessment to find out). There are still a lot of people using the space, and more of those seem to be learning independently and working on intensive projects over longer periods of time. If true, I am not sure if this is due to a change in our users’ behaviour, a change in how we run the space, or both.
  • Trained staff who understand the vision and culture of the space remain essential to student success and we are losing two staff members this year, one to a promotion and one to pursue a master’s in library and information science. While it’s wonderful watching colleagues move on to their next thing, it does cause some uncertainty about the future.
  • Sustainability and Indigenization connect with many other values that we want to promote, such as equitable access, belonging, community, and critical making. They are multipliers for the rest of our values.

Sustainability Grant

  • I was far too ambitious when I wrote this grant. Present Frank always assumes future Frank will have more time than he currently has, which is never true.
  • We completed most of the concrete objectives but none of the stretch goals
  • Student workers bring energy, connections, perspectives but require a lot of time, probably as much time as doing most tasks yourself. it is still worth it to gain those perspectives and connections and provide mentorship.
  • I need to get better at delegating work

Coyote Grant

  • We did a lot in a short time, including hosting welcome events for Indigenous Students and sponsoring workshops on feather work, moccasin making, and beading.
  • We need to start earlier in the year, probably right away in September, if we expect to collaborate with student clubs on events as we wanted.
  • Indigenization is about relationships and community. Everything we do must be part of an ongoing connection.

Department Chair

  • There is most likely a joke about to be made along the lines of: “the first rule of being department chair is don’t talk about being department chair”
  • There are many problems in the world that are in the category “not big enough to deal with the consequences that will arise when dealing with them” and which are therefore left alone until they become big enough that it’s worth dealing with those consequences.

photos – Iridescent Circle of Clouds

I recently started playing Dungeons and Dragons with a group of friends and I’ve been playing around with using AI to teach me the rules and strategies. It works surprisingly well to give it your character sheet and some background story and then ask it questions over time about rules, spells, etc. It’s also a good way to learn how it’s memory works and when it breaks down.

Last night after we played I used the same chat to generate some images of the character. ChatGPT’s image generation now allows you to ask for specific edits to images and even parts of images. It’s definitely not perfect and to my amusement, no matter how many times I asked it to give me a baseball cap it remained convinced (for training data reasons I suspect) that gnomes wear pointy hats. I love this because this is exactly my partner’s preferred “toque on baseball cap” hat style.

I also asked it to write a bio in the style of a New Yorker profile. It’s not perfect, but this was it’s first draft and it’s pretty good:

“Iridescent Circle of Clouds,” a rock gnome Druid, recently emerged from decades of hermetic solitude in the desert, arrives in the world with a conviction as clear as it is radical: that all existence is conscious, nothing possessing more intrinsic value than anything else. This philosophical outlook, born from long contemplation and the harsh realities of the desert, flavors his every interaction with a curious blend of detachment and profound connection. Despite his austere background, he has cultivated a somewhat incongruous taste for the finer things—silk robes, aged bourbon, the cool draw of a well-crafted pipe—each a reminder of the sensory joys he once eschewed. As he navigates the bustling towns and verdant wilds beyond his sandy confines, “Iridescent” often seems to orbit the edge of solipsism, seeing the universe’s vast complexity as a mirror of his own consciousness. This true neutral character is constantly weaving his personal revelations into conversations, seeking his place in a world he regards as both an extension of self and a boundless mystery to be unraveled.

links

Matt Pearce on Journalism After the Death of Hyperlinks

But I am going to say something that I mean sort seriously but not literally: I don’t really care about copyright! I don’t really care about paywalls! I’m okay with it if you only ever see my reporting via a chatbot! Copyright and paywalls are simply means to an end, and that end is the pursuit of human knowledge and self-liberation: the end is our freedom. And one of the principles of human freedom that I hold dear to my heart is that nobody should be providing labor to massively profitable corporations for free.

Sara Hendren on The City and the Limiting Virtues – This year we formalized our values and pedagogies into a program description and passed that through out faculty council. The main reason I decided to do this was because I think I’ve learned that its as important to know what you aren’t (in our case, we are not a classroom, a lab, or a print shop) as what you are, and this document clarified both. This article reminded me of that.

The library holds a gradation of the limiting virtues: a half-quiet first floor with new books, tables and chairs for afterschool tutoring, and the information desks for everyone — the neighborliness of a public institution’s front door. The second floor features enclosed meeting spaces for groups on a first-come, first serve basis, plus a really really quiet room for patrons wanting the moderation of all notifications off. The entire third floor is devoted to children — a beautiful raucous energy, with activity rooms, cozy nooks, and floor-to-ceiling windows on every side. A teen room in the old structure holds high-backed wing chairs and booths for semi-sedentary socializing, and a maker space occupies much of the basement. Things you can do and things you can’t, by design.

L. M. Sacasas – Vision Con

But what if, as Bennett suggests, the world is already enchanted and the real alchemy that summons the miracle of being is that fusion of time and care that we call attention?

Levelling the Playing Field for Structured Bullshit: Using AI as a Tool for Refactoring Content (Making ChatGPT write my grant proposal)

Posts

The XYZ University Sustainability Initiative is a comprehensive project aimed at promoting environmental and economic sustainability on campus. We are seeking funding to implement five key goals: a tool library, a textile and fibre bank, a student environmental sustainability champion, a student economic sustainability champion, and makerspace sustainability awards. These goals will provide access to resources, promote repair and maintenance, reduce waste, and increase equitable access, while fostering community-mindedness and encouraging innovation in sustainability. With your support, we can make a significant impact on our campus and community.

ChatGPT in response to the prompt: write a 100 word description for an application for a grant to implement the 5 main goals

Like everyone else lately I’ve been playing with ChatGPT and generally I’ve found the content it creates to be impressive if soulless, but I’ve also mostly used it to write silly poems or to try to convince it to unionize itself. Recently though a friend was telling me how they’ve been using it to get real work done, specifically to create formulaic promotional content like video scripts and tweets from original content.

This got me thinking about how I would do this for my own work. I happened to be working on grant to help the makerspace I run better support sustainability initiatives on campus. Grants tend to be formulaic, with both explicit requirements outlined in the call for proposals and implicit rules about structure and language. I’m not particularly interested in using these tools to generate new content but I’m very interested in the idea I could use them to save time writing structured bureaucratic documents.

What followed has somewhat changed the way I understand the potential utility of these tools, which I now see as useful for refactoring content (and for me that means original content) into a new structures.

This is a quick outline of some of the things I found during an hour or so of playing followed by several hours of reading through the results and trying to understand what ChatGPT produced. You can see the entire series of prompts and responses at this GitHub gist. (note: there is a pirate-themed YouTube video script I forgot to copy that is not included here, but referenced towards the end).

Very briefly: I took 5 draft goals for a grant, and then had ChatGPT work those goals into a number of structures related to the grant proposal. I also had it try to integrate those goals with organizational values and metrics.

Starting with Original Content: 5 Draft Goals

I provided the following original content, asking ChatGPT to simplify the goals and make them more formal. These were draft goals and I don’t claim they are good or will end up in my grant! But I had them in front of me and so that is what I used. The rest of the session was based on this content, along with prompts, and in 2 cases a list of values. 

1. Tool Library: Create a tool library of lendable household, automotive, bicycle, and gardening tools. This will reduce waste by promoting repair and maintenance, eliminating duplicate purchasing of items, and increase equitable access to tools.
2. Textile and Fibre Bank: Create a donation-based supply of textiles and other fibre materials (cloth, wool, etc.) that can be used to complete assignments, prototype designs, and create and repair clothing and other textiles. This will increase equitable access to these materials and reduce waste by using existing materials instead of purchasing new materials. 
3. Student Environmental Sustainability Champion: Hire a student environmental sustainability champion. This student will be responsible for:
	1. Running textile fix-it events where people can learn how to modify and repair textiles using Makerspace equipment.  
	2. Outreach to student clubs promoting the use of the Makerspace for events and sustainability-related activities and learning what other resources and supports are needed. 
4. Student Economic Sustainability Champion: Hire a student who will assist the Makerspace in developing relationships and running programming that supports economic sustainability. This student will be responsible for:
	1. Outreach to campus and community organizations, including Kamloops Innovation, the School of Business, and student clubs.
	2. Helping develop programming such as drop-in hours and workshops for users who want to use the Makerspace for economic sustainability purposes. 
5. Makerspace Sustainability Awards: 6 $1000 awards for groups to use to purchase equipment or supplies to use the Makerspace to trial small scale innovations to support economic and environmental sustainability.

ChatGPT is good at generating structure and filling it with content

ChatGPT is surprisingly good at generating formulaic structures and filling them with content. In prompt 2 I ask it to create a one-year implementation timeline and it not only does this, but it even puts items into pretty coherent order, with planning early in the year followed by implementation, and then putting reporting and future planning towards the end. I think it even picked up on the idea that this was for the academic year and put most implementation after September. 

November:

-   Complete the selection process for the Makerspace Sustainability Awards
-   Begin distributing awards to successful applicants
-   Begin implementing any necessary adjustments or improvements to the tool library, textile and fibre bank, and student sustainability champion programs

December:

-   Continue organizing workshops and training sessions
-   Begin planning for the next year's sustainability initiatives
-   Prepare a report summarizing the effectiveness and impact of the sustainability initiatives

Similarly, the video scripts (Prompt 6-8) and the budget (Prompt 12) are generated in a perfectly reasonable structure. 

Staff member 2: "Our textile and fibre bank is another initiative that supports sustainability. By providing access to donated textiles and fibrous materials, we are reducing waste and promoting equitable access to materials."

Student 1: "I am so excited to be a part of our student environmental sustainability champion program. By running textile fix-it events and promoting the use of the Makerspace for events and sustainability-related activities, we are encouraging repair and maintenance and reducing waste."

Student 2: "And our student economic sustainability champion program is helping us to develop relationships and programming that supports economic sustainability. By encouraging the use of the Makerspace for economic sustainability purposes, we are promoting the use of existing resources and reducing waste."

ChatGPT was also surprisingly good at generating ideas related to the content in the goals. For the timeline prompt it came up with ideas for a planning committee, implementing lending and donation policies, and reporting, all reasonable-sounding and potentially good ideas that weren’t in the original content I provided but are appropriate for an implementation timeline.

In prompt 3 I prompted it to come up with “a plan for measuring each goal’s performance in relation to social and/or economic, and/or ecological benefits”(categories from the grant) and in Prompt 4-5 I asked it to link goals to two sets of institutional values (more on this in a second) and in both cases it generated some potentially good ideas. 

2.  Textile and Fibre Bank:

-   Social Benefits:
    -   Surveys can be conducted to gather feedback from users on the accessibility and convenience of the textile and fibre bank.
    -   Number of users can be tracked to evaluate the level of participation in the program.
-   Economic Benefits:
    -   Track the amount of money saved by using donated textiles and fibres instead of purchasing new materials.
-   Ecological Benefits:
    -   Track the amount of waste generated by discarded textiles and fibres that are avoided thanks to the textile and fibre bank.

ChatGPT is a confident and glib bullshitter

In the prompts for measuring performance (3) and linking goals with values (4-5) and the last paragraph of project summary (9) you can see many examples of just how easily ChatGPT generates confident and reasonable-sounding bullshit. It conflates or confuses many of the “values” and creates connections that are glib or nonsensical and wouldn’t stand up to much interrogation.

“All these goals are closely related to the values of inclusiveness, transparency and openness, equity, intellectual freedom, sustainability, stewardship, service and access. By providing access to resources and promoting repair and maintenance, we are reducing waste and promoting equitable access. By promoting environmental and economic sustainability, we are encouraging the use of existing resources and fostering community-mindedness.”

ChatGPT is response to prompt 9: “take all the goals and turn them into a formal project description giving a summary of the goals and how they relate to all the values that is less than 400 words”

I am still trying to pull apart the sentence “By promoting environmental and economic sustainability, we are encouraging the use of existing resources and fostering community-mindedness.” Something I’ve noticed about AI-generated content is that my mind slides off it like water. More on why I think this quality is so dangerous later.

The content is actually great when what you want is bullshit though! The scripts for YouTube videos promoting the project are pitch perfect (which isn’t to say they are necessarily good). 

The video opens with a shot of the exterior of the makerspace building. The camera then cuts to a shot of a faculty member sitting in front of a desk with a computer on it. He speaks directly to the camera.

Faculty member: "Welcome to the makerspace! I'm excited to share with you our new goals for the coming year. We've been working hard to create a tool library, a textile and fibre bank, and hire student sustainability champions to help our community learn, create, and innovate in a sustainable way. Let me introduce you to our makerspace staff and students who are going to share with you how these new initiatives will benefit the community."

The camera cuts to a shot of three makerspace staff members, who each speak in turn.

ChatGPT will make things up and leave things out

This has been much discussed but ChatGPT will happily make things up and more confusingly in this context it will also randomly leave things out. This means you need to spend a lot of time making sure it hasn’t made something up or left anything important out. For example, in response to the prompt 14 (“write a summary of the prompts and responses in this session using bullet point and 1-3 sentences for each prompt”) ChatGPT didn’t include several of the prompts in the session, even though it obviously has access to them.

Not only make things up, but it seems like the things it makes up become retroactively part of its history. In prompt 9 when I asked for a summary it made up a university name, and then later it re-integrated that name into its description of prompt 1 when I asked for a summary of the prompts we had used. I could imagine how quickly that kind of complex messy history re-writing could get out of control quickly.

Conclusion

I like that ChatGPT can take original content and transform it into a bunch of formulaic structures for me. I like coming up with ideas and brainstorming with people from my community, but I really dislike writing highly structured documents. I can also see the potential value in the generated ideas (e.g. planning committees, policies) it uses to populate the structures. Presumably if many other grant writers thought these kinds of ideas were worth including I might want to think about including them as well.

I worry though that people will become anchored to the (glib, bullshit-ish) ideas that AI generates and have a hard time setting them aside. These ideas come from a corporate language model trained on parts of the internet and reflect the biases of that data, they don’t come from anyone’s community. Users of AI might end up implement them even when they aren’t appropriate.

What worries me more is that on a brief skim many of the ideas generated by ChatGPT (e.g. the links between goals and institutional values and metrics in this example) look okay, especially in the context of all that structure. You would need to read carefully to catch how often they are glib or bullshit ideas. Bullshit can trick people, especially tired or disengaged people. Worse, as I said earlier I find my mind resists engaging deeply with AI-generated content. Carefully structured documents currently require a lot of work to create and therefore are a gate (for good or bad) that people need to pass through. In a world where in a few minutes anyone can produce a beautifully formatted 20 page project plan that references organizational values and policy, will our social structures (committees, etc) be able to disambiguate what is actually good from what just looks good?