Neha Kennard
@nnkennard.bsky.social
PhD student in NLP at UMass Amherst; previously Google, Stanford, BITS Goa. she/her.
Reposted by Neha Kennard
Dang, hard disagree. The best papers to write and read are works of art, not merely a list of data and statements.
Don’t let LLMs take this away too, for gods sake.
Don’t let LLMs take this away too, for gods sake.
I honestly think we would all be a lot more productive if papers were bullet points with plots.
November 14, 2025 at 10:22 PM
Dang, hard disagree. The best papers to write and read are works of art, not merely a list of data and statements.
Don’t let LLMs take this away too, for gods sake.
Don’t let LLMs take this away too, for gods sake.
Reposted by Neha Kennard
Unacceptable. There are 189,000 people in my district who will be paying 50–300% more for the same, and in many cases worse, healthcare.
I won’t do that to the people I represent. I’m a NO on this “deal.”
I won’t do that to the people I represent. I’m a NO on this “deal.”
November 9, 2025 at 9:47 PM
Unacceptable. There are 189,000 people in my district who will be paying 50–300% more for the same, and in many cases worse, healthcare.
I won’t do that to the people I represent. I’m a NO on this “deal.”
I won’t do that to the people I represent. I’m a NO on this “deal.”
Reposted by Neha Kennard
I feel like if your AI company has to apologize to the estate of Martin Luther King Jr you should just flood your servers and walk into the ocean
October 17, 2025 at 5:38 PM
I feel like if your AI company has to apologize to the estate of Martin Luther King Jr you should just flood your servers and walk into the ocean
Reposted by Neha Kennard
Joe’s boss here. I pay him for a combination of research and service to the community. Peer review, of course, checks the latter box.
October 16, 2025 at 11:48 PM
Joe’s boss here. I pay him for a combination of research and service to the community. Peer review, of course, checks the latter box.
Reposted by Neha Kennard
"we cant uninvent ai" no we cant but we couldnt uninvent nfts either and nobody gives a fuck about them anymore
October 9, 2025 at 9:00 PM
"we cant uninvent ai" no we cant but we couldnt uninvent nfts either and nobody gives a fuck about them anymore
Reposted by Neha Kennard
Daughter of Robin Williams, Zelda Williams 💔
October 6, 2025 at 8:45 PM
Daughter of Robin Williams, Zelda Williams 💔
Reposted by Neha Kennard
It’s boiling in October because you’re using ChatGPT to write your emails, Janice.
October 6, 2025 at 11:06 PM
It’s boiling in October because you’re using ChatGPT to write your emails, Janice.
Reposted by Neha Kennard
we’d love to turn off the machine that convinces children to kill themselves, but we can’t really justify losing our ability to lay off customer service agents, now can we?
August 26, 2025 at 10:48 PM
we’d love to turn off the machine that convinces children to kill themselves, but we can’t really justify losing our ability to lay off customer service agents, now can we?
Reposted by Neha Kennard
i think a clarifying question for ai harms researchers is "could this or any harm you research entail abandoning / destroying this form of technology"
and if the answer is some version of "we want to make it better" then you're not a roadblock you're a whetstone
and if the answer is some version of "we want to make it better" then you're not a roadblock you're a whetstone
August 14, 2025 at 2:17 PM
i think a clarifying question for ai harms researchers is "could this or any harm you research entail abandoning / destroying this form of technology"
and if the answer is some version of "we want to make it better" then you're not a roadblock you're a whetstone
and if the answer is some version of "we want to make it better" then you're not a roadblock you're a whetstone
Reposted by Neha Kennard
Meta's actions are telling and appalling here, but the solution isn't better guidelines/more content moderation: it's less (read: no) fake people.
August 14, 2025 at 3:31 PM
Meta's actions are telling and appalling here, but the solution isn't better guidelines/more content moderation: it's less (read: no) fake people.
Reposted by Neha Kennard
Hey, as a librarian I need to tell you:
Our funding depends on usage.
So go ahead and check out that book even if you don't think you'll finish it. No one will know. Check out that DVD even though you might not like it. Get ambitious in your quest for knowledge.
Our funding depends on usage.
So go ahead and check out that book even if you don't think you'll finish it. No one will know. Check out that DVD even though you might not like it. Get ambitious in your quest for knowledge.
August 12, 2025 at 2:05 PM
Hey, as a librarian I need to tell you:
Our funding depends on usage.
So go ahead and check out that book even if you don't think you'll finish it. No one will know. Check out that DVD even though you might not like it. Get ambitious in your quest for knowledge.
Our funding depends on usage.
So go ahead and check out that book even if you don't think you'll finish it. No one will know. Check out that DVD even though you might not like it. Get ambitious in your quest for knowledge.
Reposted by Neha Kennard
"Our new white paper on why we're worried Grammarly is mad at us"
August 2, 2025 at 8:09 PM
"Our new white paper on why we're worried Grammarly is mad at us"
Reposted by Neha Kennard
A Borges story about a guy who gets AI to summarize all the world’s information for him, and then summarize the summary, until the AI has the whole world summarized into a single word. He sits alone at his desk, staring at the word, repeating it endlessly, certain he is experiencing everything
July 14, 2025 at 3:02 PM
A Borges story about a guy who gets AI to summarize all the world’s information for him, and then summarize the summary, until the AI has the whole world summarized into a single word. He sits alone at his desk, staring at the word, repeating it endlessly, certain he is experiencing everything
Reposted by Neha Kennard
Interesting side note: «Morning Pledge» isn't the work of a single living artist but was produced by an anonymous corporate collective called «Thomas Kinkade Studios®,» which continues to churn out images in Kinkade's signature style long after his death in 2012. It's basically analog AI slop
6/
6/
Morning Pledge
Thomas Kinkade loved this country. He believed in the founding fathers and their view of their fledgling republic. He absolutely knew that the United States was a beacon of hope for the world, promisi...
thomaskinkade.com
July 3, 2025 at 8:05 AM
Interesting side note: «Morning Pledge» isn't the work of a single living artist but was produced by an anonymous corporate collective called «Thomas Kinkade Studios®,» which continues to churn out images in Kinkade's signature style long after his death in 2012. It's basically analog AI slop
6/
6/
Reposted by Neha Kennard
Academics don’t finish projects, we just lose the energy to keep doubting them
July 3, 2025 at 9:35 AM
Academics don’t finish projects, we just lose the energy to keep doubting them
Reposted by Neha Kennard
Because a calculator is a device made for a specific purpose, our trust in it would be immediately corrupted if it were to give an incorrect answer. It would be useless.
If you take away the specific purpose you also take away the criteria to determine if something works or not.
If you take away the specific purpose you also take away the criteria to determine if something works or not.
June 12, 2025 at 8:11 PM
Because a calculator is a device made for a specific purpose, our trust in it would be immediately corrupted if it were to give an incorrect answer. It would be useless.
If you take away the specific purpose you also take away the criteria to determine if something works or not.
If you take away the specific purpose you also take away the criteria to determine if something works or not.
Reposted by Neha Kennard
Trusting your own judgment in a reflexive way is hard but crucial to democratic autonomy. The externalization of judgement weakens this moral muscle
Cannot be overstated that the dominant project for people boosting this tech is to convince users that they can’t trust their own senses and they should give their body over to the machine.
June 12, 2025 at 2:04 PM
Trusting your own judgment in a reflexive way is hard but crucial to democratic autonomy. The externalization of judgement weakens this moral muscle
Reposted by Neha Kennard
I am sitting in a public library working on a grant, and a grandmotherly woman is patiently teaching a small boy how to read at the table next to me. Another woman is knitting nearby. The children’s section is packed with kids. Things in my country are bleak, but libraries are where I find hope. ❤️
June 10, 2025 at 6:11 PM
I am sitting in a public library working on a grant, and a grandmotherly woman is patiently teaching a small boy how to read at the table next to me. Another woman is knitting nearby. The children’s section is packed with kids. Things in my country are bleak, but libraries are where I find hope. ❤️
Reposted by Neha Kennard
Trying to read a PDF and I am being prompted in three separate places to use an LLM to summarize the fucking article. Is this what they mean when they say that gen AI is inevitable
June 11, 2025 at 2:38 PM
Trying to read a PDF and I am being prompted in three separate places to use an LLM to summarize the fucking article. Is this what they mean when they say that gen AI is inevitable
Reposted by Neha Kennard
Worth noting today that the entire budget of the NEH is about $200M.
According to acting DOD Comptroller Bryn McDonnell it'll cost $134M for the deployment of the Guard to Los Angeles.
June 10, 2025 at 3:48 PM
Worth noting today that the entire budget of the NEH is about $200M.
Reposted by Neha Kennard
learn to code! learn to prompt! learn to buy into meritocracy bullshit instead of demanding dignity and a good life regardless of educational background and how wealthy you can make other people!
No no you misunderstand. Learn to Code was the scam that beget the scam!
An interesting angle to blame the Learn to Code movement when the real problem is the outrageous and infinitely destructuve 'ai' scam
June 7, 2025 at 5:21 PM
learn to code! learn to prompt! learn to buy into meritocracy bullshit instead of demanding dignity and a good life regardless of educational background and how wealthy you can make other people!
Reposted by Neha Kennard
Because it’s harder to catch defects in code you didn’t write, the only realistic way of ensuring that you really do “go faster” is to let more defects through, unintentionally or no.
So, effectively “LLMs for coding” are largely a cover story for improving throughput by relaxing quality standards
So, effectively “LLMs for coding” are largely a cover story for improving throughput by relaxing quality standards
June 3, 2025 at 10:02 AM
Because it’s harder to catch defects in code you didn’t write, the only realistic way of ensuring that you really do “go faster” is to let more defects through, unintentionally or no.
So, effectively “LLMs for coding” are largely a cover story for improving throughput by relaxing quality standards
So, effectively “LLMs for coding” are largely a cover story for improving throughput by relaxing quality standards
Reposted by Neha Kennard
Reposted by Neha Kennard
We can't just tell people to not watch video or scroll instagram or whatever, but if we get busy at an early age showing the deep pleasure that can be had through engagement with writing, those other things become less tempting, or if tempting, they can at least be compartmentalized.
May 22, 2025 at 2:52 PM
We can't just tell people to not watch video or scroll instagram or whatever, but if we get busy at an early age showing the deep pleasure that can be had through engagement with writing, those other things become less tempting, or if tempting, they can at least be compartmentalized.