Home fires burning

Logpodge #5

Illustration by Riswan Ratta

Home fires burning

One of the first courses I taught was on World War I. It was in the summer, a short course that met twice a week for three hours. I talked to a colleague about how worried I was about holding student’s attention for such a long class meeting. She suggested adding a lot of popular music to the syllabus. Brilliant. I learned then that music is an excellent way to help students connect to the lived experience of humans in the past.

One song from that summer has come back to me in the last few weeks. It plays on repeat in my mind’s own iPod, unbidden, the way TV commercial jingles or pop songs from my childhood sometimes do.

Keep the Home Fires Burning is not a particularly good song by today’s standards. It's a Long Way to Tipperary and Over There are catchier. Hello Central! Give Me No Man's Land still packs a hell of a punch. But my weird brain resurfaces this refrain

There's a silver lining

Through the dark cloud shining,

Turn the dark cloud inside out

My subconscious casting about for some reason for hope. Like so many of us, I’m not sure where to direct my attention or effort as our home fires are burning out of control. The astonishing burst of “move fast and break things” management theory has set blazes throughout the federal programs that support people in need, keep our infrastructure working, and fund institutions that create and teach knowledge.

And yet, I keep writing and talking about AI and education as if such things matter. In my better moments, I think of this as an act of hope.

If there is a reason to imagine turning the dark clouds inside out, it is not easy to see. But then, there was no such reason at the beginning of 1918 either. The fires then, like the fires today, are consuming so much of what feels necessary and important.

And if the fires are put out tomorrow? If some semblance of checks and balances or separation of powers returns to Washington? If a few dedicated leaders and citizens preserve the structures of research and learning that are one of the great contributions of the United States to the world? Even if those ifs are realized, our institutions of higher learning, imperfect as they are, will be significantly and irrevocably altered by the past month and by the years ahead. Institutions that have housed efforts to develop vaccines, preserve culture, and understand human society will find themselves in need of remaking.

Maybe that’s the silver lining? Maybe why my mind quietly sings

And although your heart is breaking,

Make it sing this cheery song

The song is not cheery. Not at all. But it is a song about preserving something important in the face of destruction and hopelessness. That, I’m sure, is why it rolls through my mind.

For no gallant son of Freedom

To a tyrant's yoke should bend

We are the children of freedom. Despite their many flaws, the institutions where we have created communities inquiring after knowledge and truth are worth fighting for. And, when these home fires burn out, it will be worth rebuilding colleges and universities, and the republic, into something different and better.

---

Photo by Hunters Race

Confidence Games

I’ve been talking to a lot of folks about CSU’s agreement with OpenAI because, as I wrote here and here, it represents a troubling turn in the relationship between higher education institutions and the push to turn the promise of AGI tomorrow into something that pays off today. $16.9 million for 18 months of ChatGPT Edu, plus the time and effort to set up access and training, is an unnecessary risk for what seems to be little reward. There are better ways for a university system with a shrinking budget to spend millions of dollars. But at least, as far as we know, CSU didn’t spend additional millions on consultants to come in and tell them to do this.

After drops in enrollment in 2021 and 2022, CSU approved a 6% annual tuition increase that started this past fall and will continue for four more years. The governor proposed cuts of 8% to the CSU system in the 2025-2026 budget. Cal Matters reports on the impact of the cuts across the system. The budget situation is specific to California, and yet it is worth considering how it foreshadows the tough times ahead for the rest of higher education as federal dollars disappear and layoffs start. Even before this month’s disaster, there was the growing cost and student debt associated with graduate degrees, the demographic reality of fewer 18-year-olds, and a politics-driven resistance to immigration, all of which will squeeze tuition and enrollment for the 23 institutions, soon to be 22, in the CSU system.

I have a lot of sympathy for higher education leaders trying to make decisions about AI in a social and political environment that is genuinely threatening the existence of the institutions they manage. But not as much sympathy as I do for the faculty who are losing their grants, staff who are losing their jobs, and the families who are realizing that they can’t afford to pay for a CSU degree. There is never a good time to make a bad decision, but I can’t help but think about where the millions CSU officials will spend over 18 months trying to make ChatGPT Edu appear something other than what

Marc Watkins

calls “a mix of virtue signaling and panic purchasing” might have gone instead.

One way to measure the value of what they purchased is to count how many people actually use ChatGPT through the license. The headlines promised everyone at CSU would receive access, but that doesn’t mean everyone will bother to use it. I have no idea how they calculated the price, but paying based on headcount or FTE (Full-Time Equivalent) makes little sense. Pricing models like the one used by BoodleBox, which charges based on the number of active users, are better for new experimental technology. I hope some enterprising reporter gets hold of the agreement so we can find out the details.

There is also the question of what the people who use it actually use it for. As any AI enthusiast can tell you, OpenAI created the fastest-growing consumer product of all time! They just announced that they have surpassed 400 million active daily users! But is having ChatGPT do your homework, tell you which pizza places deliver to your dorm room, or take notes for your department meeting worth $16.9 million plus support and implementation costs? Inquiring minds want to know how the first AI-empowered university system is going to assess the educational value of its AI investments.

A few other questions: How soon with access be granted to all CSU campuses? What will the promised training programs consist of? Will they be enthusiastic advertisements for our AI-empowered future or critical explorations of the complex bundle of technologies we call AI? Cal State Chico’s FAQ sketches some answers. You can decide for yourself how satisfying they are. The big question is: assuming OpenAI is still in business, what happens in 18 months? How will CSU decide whether to renew the agreement, and what role will educators play in assessing its value?

More confidence games

Deciding what to do about AI is part of a much larger set facing higher education leaders about how to manage shrinking budgets. As

Hollis Robbins (@Anecdotal)

pointed out in the Chronicle of Higher Education last week, some presidents and chancellors are signing multi-million dollar contracts with consulting firms to guide them in making tough decisions. She argues this signals “a fundamental crisis in higher-education leadership: the belief that external consultants, rather than internal expertise, hold the keys to institutional transformation.” What to do about AI is just one of the many questions consultants will happily charge you millions to answer.

I should know. I am now one of them.¹ If you are a president or chancellor, here is something you can pay me to tell you: You already have experts on your campus who know about AI. They work in your IT operations and your instructional technology offices. You have instructors in your writing programs who can tell you what a good AI literacy program looks like. You have faculty who can explain how generative AI models work and bring you up-to-speed on preliminary research about their educational value. Go talk to them. Put them in charge of coming up with a plan and executing it. Maybe, instead of paying a consultant, you could pay them a bonus for putting in extra hours.

You are much better off listening to your own people to figure out AI. This extends beyond AI. These are tough times, and you will need to make hard decisions. Talking to members of your community about how to shrink budgets and cut programs will be part of those tough times. You will come away from those meetings feeling exhausted and unappreciated, but then, so will the people who show up to those meetings. There is no easy path.

Consultants will make it feel easier. This is their purpose. This is what they would call their “value add” except they won’t say that part out loud. Instead, they will show you charts about ROI and KPIs to sell you on their expertise. They will develop five-point plans to optimize your campus academic operations and future-proof your core programs. They will tell you how smart you are for hiring them and smile warmly as they ask after your kids.

At the end of the engagement, with the work undone and your campus demoralized, they will pitch you on extending the engagement. It’s not their fault academic governance added months to the timeline. And you’ll sign the extension because it is easier to have the bad people in the nice suits tell you which programs need to be cut and which staff have to go. “Those bastards,” you can say to yourself and your campus after they leave.

I am not suggesting that blowing millions on AI contracts or consultants is widespread; I honestly don’t think it is. Aside from CSU, OpenAI Edu has not announced any enterprise agreements since their big press release with Arizona State a year ago. From what I hear, the big consulting firms are not getting as much action from higher ed these days.

The troubling thing is that $16.9 million to OpenAI and $51 million to Huron is so obviously a waste of money. We only know these numbers because state systems are required to be somewhat transparent. Even then, the University of Wisconsin system seems to be going to great lengths to hide what it paid Huron.

The best way to discourage this sort of behavior is to call it out. Institutions that pay millions for an enterprise license for ChatGPT or for consultants to tell them how to reduce costs are wasting much-needed funds at a time when no college or university in the country can afford it.

AI Log, LLC. ©2025 All rights reserved.

1. To be clear, I am not the kind of consultant who does slide decks with buzzwords and nice charts. I’m the kind of consultant you hire to help people on your campus talk to each other. Some people would describe this work as like coaching. I prefer to think of it as entertainment. I’m like a comedian or cocktail lounge singer who helps people get in the mood to cooperate.