I increasingly feel the John Warner’s of higher ed are right more than wrong about these things. Although it feels very task-specific.

I’m going on vibes here not trying to argue anything sophisticated.
@biblioracle.bsky.social

@annamillsoer.bsky.social the Model Spec's for o3 seem ripe for humanities discussion, no? a lot of this is about how to remain principled and useful: model-spec.openai.com/2025-02-12.h...

I wonder if it’s a platform issue?

I see a ton of humanities conversation around this, locally and in my online networks.

But AI chatter on X is dominated by tech circles—more so now that AIED seems to have drifted to LinkedIn and here.

haha I just posted something similar before seeing this. ya there's an incredible amount of oversight in that experiment.

ARC claims o3 is not using brute force: arcprize.org/blog/oai-o3-.... It also tempers expectations around the benchmark.

here's the article's summary of what o3 seems to be doing on the backend

Thanks! I put it on the list

you could TRY to explicitly encode paul graham's obsession with maker schedules or ribbonfarm's metaphorical thinking but trying to decompose it into explicit steps is sometimes counterproductive - you'll miss all the subtle correlations and emergent patterns that make the approach actually work

Totally. Given that much of this pedagogical debate hinges on workplace preparedness, it seems like we’re not doing a great job of tracking where GenAI is actively scanned for and punished.

I mostly see reports on how much employees using AI now but clearly it’s more complicated.

Totally. Institutions should be planning for this IMO. Departments have to figure out how to have a variety of instruction depending on what students and faculty want.

At least that’s where I’m at.

I really hope higher ed can be a place for both of these models (and more). I like that some people are super hardcore about keeping AI out and others think it’s integral to the future of ed. It would be sad to see any side win.