# AI will not hurt the signaling value of elite colleges > Source: > Published: 2026-05-21 19:25:25+00:00 Interesting rage bait, “[if you can’t get a job today, it’s your fault](https://auren.substack.com/p/if-you-cant-get-a-job-today-its-your):” the brand of a college was never about the curriculum. nobody hired a Penn graduate over a Penn State graduate because Penn taught macroeconomics better. they hired the Penn graduate because the admissions filter at Penn was strong. you went there, so you were probably smart and probably hard-working. the school did the screening for the company. that has collapsed for three reasons. the first is that admissions criteria stopped predicting much of anything. SAT scores still mean what they meant — raw cognitive horsepower. but everything else (legacy, sports, regional balance, narrative essay quality) became more weighted, which weakens the signal coming out the other side. Somewhat disagree. As admission rates at elite colleges continue to fall, the signaling value of getting admitted increases, even with AI and lower SAT ceilings. Getting into MIT or Stanford University has always been difficult, but today it’s absurdly competitive. The filtering goes well beyond the SATs or GPAs, but also includes math competition placings and extracurricular projects. Provided it’s not possible to cheat, these still act as effective signifiers of ability. For math, physics, or coding competitions everything is proctored under a supervised setting, so cheating is not possible. Getting into an elite college is like making it to the NBA. We’re talking the truly cream of the crop (except for the token diversity/alumni admittees). Despite AI, I have predicted top universities will still serve as filtering mechanisms for employers and gatekeepers of prestige. This won’t change much, as employers and universities adapt to AI with countermeasures. An example is academic publishing. Top journals are still prestigious and confer career advantages, even with the rise of preprints (e.g. ArXiv) leading to more content being published. A similar pattern was seen with online learning: making courses available to the public for free does not devalue the credential itself. Or Amazon and self-publishing: the “big 4” publishing houses are still as influential as ever. Gatekeepers still function as curation, telling people “what is important”. With AI, all you’ve done is automate the production, but not the curation. Even if AI can curate, the hierarchy still exists. What AI does do is devalue coursework, where cheating plays a bigger role, but the process of being admitted is sill a useful signal in and of itself. This is also why [grade inflation](https://www.nytimes.com/2026/05/20/us/harvard-grade-inflation.html) at Harvard doesn’t devalue the credential–the value is getting in.