Shreeharsh Kelkar

Lecturer at @UCBerkeley in ISF. Research on AI, algorithms, organizations, work, labor, and expertise. Writing at

[Alex Imas @alexolegimas]

You know what other tools know better than most instructors? Coursera and YouTube courses from top faculty, *the internet*, books from the library. How many students used those tools instead of formal ed? Very very few. How many will use Claude independently to learn the material? Probably the same amount.

I know it doesn’t sound glamorous, but the primary role of faculty is to get students in the seats and create incentives to actually absorb the information. This is your job. AI can help as a tool, I’ve seen some great harnesses of AI for education, but it will not do this.

[Shreeharsh Kelkar @scritic]

@fstflofscholars @STS_News I love Foucault but I think he's been misinterpreted but I also think it's his own fault: https://t.co/nkvGknc8KL

Curious what you guys think. Alas, like everything else I write, it's long.

[Timur Kuran @timurkuran]

AI just killed higher education’s old teaching model. We need smaller classes and oral defenses for every paper—implying more faculty time, hence more professors. Since banning AI is unenforceable, written work alone can no longer be trusted.

[Shreeharsh Kelkar @scritic]

@NateWitkin @ben_golub @LedermanHarvey Perhaps "tractable" is weaker than "computerizable"; Dreyfus used to say some domains are more "formalizable" than others. My point is just that all this partitioning of activities has not worn well. This Harry Collins review of Dreyfus is really good. https://t.co/eM9jT03TTo

[Shreeharsh Kelkar @scritic]

@NateWitkin @ben_golub @LedermanHarvey I'm skeptical of AGI but sorting human tasks into "computerizable" or not has a bad track record. In the 70s, experts doubted computers could use natural language; now ChatGPT does it so well (in a totally non-human way) & we can build real-world-changing software on top of it.

[Shreeharsh Kelkar @scritic]

We really need to have more concerted efforts to create watermarking standards. We have given up on this far too easily. It's not going to solve everything magically but it will certainly help! https://t.co/IzxWTACSpB

[Derek Thompson @DKThomp]

that both of these stories were published on the same day is almost too convenient for think-piece lede-writing purposes

[Shreeharsh Kelkar @scritic]

@DKThomp I think we need to have more concerted efforts to create watermarking standards. We have given up on this far too easily. It's not going to solve everything magically but it will certainly help! https://t.co/IzxWTACSpB

[Julian ☽ @julianblacks_]

This is going to be long.

Last semester I suspected I had a major issue with use of AI in my survey courses, so I inserted what is known as a trojan horse (not the virus kind) into the directions of a paper assignment. As it turned out, I did in fact have a major problem, and a post on Threads about it accidentally went quasi-viral and ultimately became a Huffington Post article and an NPR interview. (Links at the end)

[Shreeharsh Kelkar @scritic]

@julianblacks_ Great thread! Facing similar struggles here. How did you know the role-play essays weren't AI? (I'm worried I won't be able to tell in my 100+ classes.) In smaller classes, I've seen students use AI even for personal reflections, which makes those questions suspect to me now too.

[Shreeharsh Kelkar @scritic]

@postdiscipline @STS_News @DarrinADurant This sounds more like *your* value/policy preferences expressed under the guise of science than about how to tradeoff between actual value disputes and conflicts.

LAT piece on the Mojave stuff: https://t.co/KUma5wPpH9.

On uncertain COVID science: https://t.co/MSMWEqHPaH.

[Shreeharsh Kelkar @scritic]

@postdiscipline @STS_News @DarrinADurant Agree. But how does this help us make a decision about school closures in the next pandemic when science is filled with uncertainties and constituencies clash? Or whether to build utility solar installations in the Mojave which cuts carbon emissions but reduce biodiversity?

[Shreeharsh Kelkar @scritic]

@postdiscipline @STS_News @DarrinADurant And yes, I think Dewey is correct because what else is there? We live in a free society with elected representatives who are supreme. My take here: https://t.co/L4lraw6whk

And on Sarewitz here: https://t.co/Hd4akYz1A6 though the original is 1000 times better.

[Shreeharsh Kelkar @scritic]

@postdiscipline @STS_News @DarrinADurant I don't think this is annoying! I've assigned Kitcher in class.

But contra Kitcher, I don't think the "epistemic division of labor" can be settled philosophically. I side with Sarewitz & Pielke: it's better to negotiate value tradeoffs before thinking about facts & science.

[Shreeharsh Kelkar @scritic]

@STS_News @DarrinADurant Perhaps I have a broader definition of STS because I put the history of technology stuff into STS as well. I think we just have to agree to disagree here. (And the quotes in my previous tweet are insane, my apologies.)

[Shreeharsh Kelkar @scritic]

@DarrinADurant @STS_News Many "histories from below" try to prove marginalized people are "experts" ignored by "power"—even if not explicitly about experts/democracy. Same for soc/anthro. This is the social sci "critique" that ran out of steam; the STS democracy/expertise stuff is less important.

[Shreeharsh Kelkar @scritic]

@BrandonWarmke Good point but I think if departments thought about who gets turned off, we wouldn't be where we are now. My point was just that a lot of these things are driven by existing college students as much as faculty. There is a reason the US is polarized by educational attainment.

[Shreeharsh Kelkar @scritic]

@yooday Great piece! So many songs I still need to listen to! One note: Aaiye Meherbaan was composed by O. P. Nayyar, not S. D. Burman. You don't say that explicitly but the next sentence with the quote from Lata in Kabir's book implies it.

[Shreeharsh Kelkar @scritic]

Since "Bullshit Jobs" is back in the discourse, I wanted to re-up my essay about the problem with David Graeber's theory of technology: he thinks we can make whatever technologies we want; all we need is the will to do it. That's not how it works. https://t.co/l8vF03rBUi

[Chasing Ennui @rwlesq]

Even after the death of its author, "Bullshit Jobs" remains a cancer on the discourse. It was perhaps the most incurious book I have ever read (strictly speaking, I only read the first third).

Companies are not "subsidizing" anyone. If they are hiring someone to do something, there is almost certainly a reason beyond "we wanted to give this random person money." If you can't figure out why, it is probably a problem with you.