Shreeharsh Kelkar (@scritic) header
Shreeharsh Kelkar (@scritic) profile picture

Lecturer at @UCBerkeley in ISF. Research on AI, algorithms, organizations, work, labor, and expertise. Writing at https://teapawt.org/t.co/HoZfMecSz6.

Berkeley, CA
Joined May 2008
Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)
Shreeharsh Kelkar retweeted
🙏🙏
Great episode from @WSITYpod and a model for how contentious issues like the tradeoffs of non-pharmaceutical interventions should be conducted. Wish they had gotten to vaccines. This is going into my course readings. Watch it here: youtube.com/watch?v=ZCWQ-qJT…
1
1
198
Shreeharsh Kelkar retweeted
On Substack, I summarize my new book Gender Identity: The Career of a Category and reply to a mini-symposium on the book hosted by @dan_oppenheimer, with great contributions by @scritic and @BloggerGideon. substack.com/home/post/p-219…
5
31
885
Shreeharsh Kelkar retweeted
delighted by the very thoughtful review of my new book Gender Identity: The Career of a Category by Shreeharsh Kelkar @scritic danieloppenheimer.substack.c…
3
1
10
501
Shreeharsh Kelkar retweeted
Has the Turing Test been rendered obsolete? It depends on how you interpret the test: as a test of second-order social norms (no) or as a test of language performance (yes). More in my latest blogpost: computingandsociety.substack…
I’m still surprised — and think it’s under-discussed — that the Turing test has not only been passed but rendered obsolete. That it was a historical parenthesis. *The* benchmark for machine intelligence, understood proleptically for much of my life/career, and it’s just over now.
1
1
2
152
Shreeharsh Kelkar retweeted
More on the Turing Test and the nature of computational machinery and intelligence: computingandsociety.substack… LLMs' success shows that programs driven by *industrial machinery* are capable of a lot!
1
1
2
78
Shreeharsh Kelkar retweeted
GreenEarth is the first feed which gives you full control over the algorithm. Here's our new "post sources" control panel. It's over on BlueSky, the only platform which supports algorithmic choice. Bonus: effectively filters out that platform's famous whining! Have you tried it yet? go.greenearth.social/r/ptryi…
1
6
2
15
2,808
Shreeharsh Kelkar retweeted
On the 35th anniversary, I am thrilled to announce our oral history collection quarterbacked by @shreyasnarla: ~50 hours of recordings of the @mercatus team with the reformers and close witnesses to 1991. These conversations cover the decades before the crisis of 1991 and trace the reformers’ influences, education, careers, mentors, institutions, and the development of the ideas that later became policy. the1991project.com/oral-hist… The collection also includes a 2016 interview with Dr. Manmohan Singh, generously shared by @ChaitanyaKalbag. 3/n
3
28
1
135
10,351
Shreeharsh Kelkar retweeted
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.
“AI is demoralizing.” A Princeton Professor says he kept wondering this semester (while lecturing) if his students would be better off learning from Claude:
56
141
29
1,104
118,537
Shreeharsh Kelkar retweeted
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.
160
239
78
1,469
150,595
Shreeharsh Kelkar retweeted
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)
24
201
55
1,652
260,634
Shreeharsh Kelkar retweeted
At this point, it's hard to say anything too new about the "institutional neutrality" debates, but I think @JeffreyASachs does, here: it's about public relations, and that's just fine: chronicle.com/article/neutra…
2
5
2,195
Shreeharsh Kelkar retweeted
New publication: Understanding China Through Digital Anthropology, Edited with Wang Xinyuan A free download from UCL Press. In the conclusion, we explain the profound differences in attitudes to the digital between China and the West. uclpress.co.uk/book/understa…
10
1
48
2,607
Shreeharsh Kelkar retweeted
It is all too common for faculty and administrators to refer to a campus conversation or class discussion of public policy as causing "harm". We must challenge the use of this language.
We must repudiate the idea that "speech can be violence" once and for all. @glukianoff and I wrote about the dangers of promoting this idea on college campuses back in 2017, in @TheAtlantic: theatlantic.com/education/ar…
3
6
18
2,399
Shreeharsh Kelkar retweeted
A new post on The Conversation estimates that flushing the toilet equals the water used in 154 GPT-5 prompts and 1,714 GPT-4o prompts. I am wondering if anyone has seen analysis or pushback on the piece. theconversation.com/ai-has-a…
3
1
4
518
Shreeharsh Kelkar retweeted
Bureaucratic science is already generously funded and does not need more private support. Philanthropists should fund young outsiders, like during the Golden Age of Science. Read the new article by @stuartbuck1 (link below):
5
36
7
204
45,942
Shreeharsh Kelkar retweeted
Advocates for policies and policy frameworks will often argue that their proposal is more “democratic” while the one they oppose is “technocratic.” But conflicts in many industrial societies today are NOT between publics on one side and experts on the other.
1
4
1
11
2,427
Shreeharsh Kelkar retweeted
Great, useful thread and post (linked in the thread)
New post where I use insights from the social studies of science and technology (#STS) to try to answer the normative question: on an issue where you don't know much about the science, how do you find the right experts to tell you what is going on? Link in the comments.
1
2
153
Shreeharsh Kelkar retweeted
I read Marx's Capital after many years and made a critical reading guide to help students get through it. Comments welcome. docs.google.com/document/d/1...
1
2
6
497
Shreeharsh Kelkar retweeted
Waymo is a "highly specialized solution" in exactly the same way that a cell phone is a "highly specialized solution." You can only use your cell phone in places where your provider has built a tower. But providers have built towers in a lot of places.
Replying to @TashaARK
Waymo is impressive, but a highly specialized solution. The Tesla approach is a general solution. The latest build is capable of zero intervention drives. Will release limited beta in a few weeks.
2
5
1
35
Shreeharsh Kelkar retweeted
This perfectly articulates why "model alignment" is the wrong way to think about AI safety. Safety is not a property of models, it's a property of systems that use models. Asking whether a model is safe is like asking whether a computer chip is safe.
13
36
3
208
16,011