onfocus

“Without a thriving, inclusive higher education system that serves the public good, the majority of Americans will be excluded from meaningful participation in our democracy and this country will move backward,” Wolfson predicted. “We will do everything in our power to protect our institutions, faculty, staff and students and stand up against those seeking to violate academic freedom and the core principle of higher education.”

The correct message. Amazingly, I’ve only seen higher ed unions discussing the new US administration. Unions might be the only resistance.

"Stanford remains deeply concerned about efforts, including lawsuits and congressional investigations, that chill freedom of inquiry and undermine legitimate and much needed academic research – both at Stanford and across academia."

So deeply concerned, in fact, that they've fired everyone involved in that legitimate research!

One student I met outside the Low Library after Johnson’s speech, who only identified himself by an initial, K, summed up the dissonance between Johnson’s description of the protesters and the reality of the encampment: “It’s as if there’s some kind of armed insurrection here—there isn’t,” K told me. “There’s maybe, like, 100 students sleeping in tents on a lawn.”

I think it’s good to keep in mind who really has power in these situations. Is it the administration and politicians who can direct militarized police or college kids? Showing up on campus to get some headlines while punching down is awful.

A clearer-eyed view of what has happened in the last two years is that a few companies have amassed enormous amounts of data (mostly taken non-consensually) and the capital to buy the computing resources to train enormous models which, in turn, can be used to output facsimiles of the kinds of interactions represented in the data.

I think a better understanding of how generative AI products are produced would help clear up some of this magical thinking that’s happening.

The disintegration of the Pac-12 has been an entirely avoidable tragedy brought on by the craven machinations of people who, it can only be assumed, do not love college sports very much at all.

This. A few TV execs maximizing ad revenue and a handful of administrators who must be under pressure to maximize football profits are making terrible decisions.

Update (9/8): OSU and WSU sue to stop schools who chose to leave from voting on the future of the conference.

West Virginians, trapped in the clutches of economic hardship, find themselves mercilessly shackled to a state most can ill afford to abandon, left to suffer the full weight of the WVU administration’s harrowing decisions. We will learn only subjects aligned with the preferences of the rich, driven by their financial motivations.

I work for an R1 land grant university and I like to think those designations help keep the university on track. Seeing WVU make students pay the price for poor leadership is sobering.

The decision reverses decades of precedent upheld over the years by narrow court majorities that included Republican-appointed justices. It could end the ability of colleges and universities — public and private — to do what most say they still need to do: consider race as one of many factors in deciding which of the qualified applicants is to be admitted.

It shouldn't be surprising that a court with no ethical standards makes unethical rulings but it's still disgusting to see billionaires get what they paid for: hurting the vulnerable and shoring up the wealthy's already abundant advantages.

Our findings reveal that these detectors consistently misclassify non-native English writing samples as AI-generated, whereas native writing samples are accurately identified. Furthermore, we demonstrate that simple prompting strategies can not only mitigate this bias but also effectively bypass GPT detectors, suggesting that GPT detectors may unintentionally penalize writers with constrained linguistic expressions.

Interesting look at the effectiveness of GPT detectors universities are using to find cheating. Especially this bit:

While detectors were initially effective, a second-round self-edit prompt (“Elevate the provided text by employing literary language”) applied to ChatGPT-3.5 significantly reduced detection rates from 100% to 13%...

Ouch, not sure how these services can get away with charging money for AI detection if it's that easy to bypass.