I miss Silicon Valley
Logpodge #4
The show
Silicon Valley was the best. The surreal remake that started its new season this week doesnβt have the same comic energy, the farcical edge of Mike Judgeβs original. I know the critics love the remake, but they would, right? The whole tired engine of prestige TV artificial general intelligence runs on giving critics something to talk about.
I admit Sam pretending to be mad about DeepSeek stealing his IP was a good bit. But everything else about this last week felt forced. The stock market wobble setting up drama for later in the season? Obvious. Samβs selfie with Satya? Weak. And I know they had to come up with something for the crossover with The Trump Show (donβt get me started), but Stargate? Really?
In every respect, the original was better. The cast? Better. The fake AI? Better. Pied Piper? A better ironic company name than OpenAI. Jian-Yang? A better Chinese character than DeepSeek. Hooli broken up and sold to Amazon? Hilarious. A much better joke than whatever is happening with Google these days.
A few people asked me for my take on the DeepSeek news. That bit of comedy is the best I can do.
After reading all week about how DeepSeek is yet another game-changer, I remain convinced that no one really knows what game is being played. One observation: making each step visible in natural language as the model works its way to the output is nice. I hope that becomes standard on what the marketing team calls βreasoning models.β
The actual news last week, including the actual higher education news, was in Washington, not Silicon Valley. You donβt need me to tell you that.
Read Timothy Burke
The best thing I read this weekβnot an antidote, really, more a reminderβ was Timothy Burkeβs response to students who ask why they should pursue an education.
Does it matter if I can write well, if I can analyze statistical data, if I can make models of complex processes, advise a client about their legal options, research new drug treatments for illnesses, study how to mitigate climate change, make beautiful art?
β¦
Yes, I answer. You are the right person to do what you do, know what you know, study what youβre going to study. You do it.
AI for sale
Last summer, when I wrote about what I saw coming this academic year, I expressed some doubt that anyone would buy what OpenAI was selling in the form of ChatGPT Edu. So far, there is not much evidence this attempt to βed their techβ is getting many takers. Instead, what I hear is faculty complaining that their institution signed a campus license with Grammarly without talking to them first and a lot of bureaucrats trying to figure out what Microsoft Copilot is good for.
Sure, there is plenty going on with AI. More than a quarter of American teens tell Pew they use ChatGPT to help with homework. Plenty of edtech companies use the term AI to help market their existing products. AI tutorbots are a popular choice for recent college graduates and middle-aged professors starting up edtech companies. Some humanities faculty are racing to the barricades to resist AI. Ivy.ai is selling a basic customer service chatbot to universities. But aside from consultants selling AI literacy workshops for busy teachers, Iβm not sure if anyone is making any money using AI in the edtech game.
A better Clippy is a good thing
Speaking of consultants, Iβm doing fine, thanks. I managed to scrounge up some paid work for companies interested in experimenting with how to use AI to make educational tools better. I have a few speaking gigs at conferences lined up this spring. All this goes to support my habit of reading books related to AI and education and then writing about them on the Internet. If you care to read advertisements for myself, I do that mostly on LinkedIn.
My series on teaching using an LLM tool will wrap up with one more part. Next, I plan to write about what, if any, value generative AI might bring to managing colleges and universities. This will take on what Henry Farrell calls theΒ management singularityΒ and what I call the βboring AI revolution,β the idea that since AGI is a fantasy, a brilliant marketing concept, or a joke, or whatever, the actual result of all this investment in transformer-based, culture-generating technology may be, in the words of Dave Karpf, βa better clippy.β Thinking of AI in that context may actually yield a commercial product, or at least help us figure out what Copilot is good for. After all, the modern business world was created by double-entry bookkeeping and the printing press, not spaceships and self-driving cars.
A better clippy is a good thing! Not the end of work or of civilization, but maybe, implemented wisely, AI will make it easier to make organizations better. Analogizing educational bureaucracies to dungeons or very old, very slow AIs helps us think about how difficult it is to reform them. If generative AI means we can put decent natural language interfaces on bureaucratic processes, then maybe we can get bureaucracies to produce better outcomes. For example, putting a natural language interface on a universityβs student handbook could help students understand what is expected of them.
New tools, like a talking filing cabinet or a spreadsheet that can gather data on its own, may have some utility for reformers, not just cost-cutting managers. Those are more interesting questions than how low Nvidiaβs stock price will go or the real meaning of Samβs latest tweet. At least to me.
Asking such questions assumes there will still be colleges and universities around to reform. I believe there will be, but maybe not as many, and they may not be in such good shape.
You are a lifeboat
In the spirit of Timothy Burkeβs lifeboat analogy, I believe the earthquakes set off by the kleptocratic fracking going on in Washington will not go on forever. Whether or not AI can help us sort through the rubble is less important than whether we survive the next four years. So, letβs gather in the lifeboats, press up against the inside walls, and hold hands so that as the foundations crack and the tsunami crests, we remain.
Laughter in small things
Thatβs a rough note to end on, especially since I am working on laughter in small things as a strategy for getting out of bed in the mornings. Let me end by making sure you know about the best webcomic going these days: Saturday Morning Breakfast Cereal. SMBC has the best AI jokes, but the one below lifts the curtain on artistic creation.
