Shanqing Cai
@sqcai.bsky.social
Software Engineer @ Google. Applying AI and LLMs to human-computer interaction and accessibility. On-device machine learning. Also interested in speech science and neuroscience. Tsinghua, JHU, & MIT alum. https://caisq.github.io/
Hear me out: IPA in @duolingoverde.bsky.social
November 8, 2025 at 12:56 PM
Hear me out: IPA in @duolingoverde.bsky.social
In the era of GenAI, we should realize and teach young people that the main product of intellectual work is not only the artifact, like lines of computer code or an essay, but also the bonds the author made with the artifact through labor. Without the latter, the artifact is dead.
November 3, 2025 at 8:21 PM
In the era of GenAI, we should realize and teach young people that the main product of intellectual work is not only the artifact, like lines of computer code or an essay, but also the bonds the author made with the artifact through labor. Without the latter, the artifact is dead.
Is there a better way to show how LLM chatbots hallucinate than giving them a YouTube video link and asking for the name of the background music?
November 2, 2025 at 1:31 PM
Is there a better way to show how LLM chatbots hallucinate than giving them a YouTube video link and asking for the name of the background music?
Do you ever think about whether robots in fictions are conscious?
October 23, 2025 at 12:55 PM
Do you ever think about whether robots in fictions are conscious?
Typing non-code text in VSCode with Copilot enabled is an "interesting" experience. The predicted texts usually are multiple sentences long and look plausible and contextualized based on the preceding text. But it is almost always hallucination that ends up being more distracting than helpful.
October 19, 2025 at 2:30 PM
Typing non-code text in VSCode with Copilot enabled is an "interesting" experience. The predicted texts usually are multiple sentences long and look plausible and contextualized based on the preceding text. But it is almost always hallucination that ends up being more distracting than helpful.
Reposted by Shanqing Cai
Pixel 10 gets new Journal app, Recorder music creator, Gboard Smart Edit, more
Pixel 10 gets new Journal app, Recorder music creator, Gboard Smart Edit, more
With the Pixel 10, Google is introducing a new Journal app that lets you quickly jot down what’s happening in your life. There are also updates to Pixel Recorder, Screenshots, and Studio.
more…
9to5google.com
August 20, 2025 at 4:16 PM
Pixel 10 gets new Journal app, Recorder music creator, Gboard Smart Edit, more
Reposted by Shanqing Cai
The vaccine against Shingles helps protect against dementia, results of a natural experiment, adding to prior evidence
"implications are profound"
New @nature.com
www.nature.com/articles/d41...
nature.com/articles/s41...
"implications are profound"
New @nature.com
www.nature.com/articles/d41...
nature.com/articles/s41...
A natural experiment on the effect of herpes zoster vaccination on dementia - Nature
Using a natural experiment that avoids common bias concerns, this study finds that the live-attenuated shingles vaccine reduced the probability of a new dementia diagnosis within a follow-up period of...
nature.com
April 2, 2025 at 3:29 PM
The vaccine against Shingles helps protect against dementia, results of a natural experiment, adding to prior evidence
"implications are profound"
New @nature.com
www.nature.com/articles/d41...
nature.com/articles/s41...
"implications are profound"
New @nature.com
www.nature.com/articles/d41...
nature.com/articles/s41...
InterSpeech paper reviews: 2 down, 1 to go.
March 28, 2025 at 2:26 AM
InterSpeech paper reviews: 2 down, 1 to go.
As a reviewer for journals and conferences, have you come across article submissions that are grammatically flawless but still make you wonder whether you have dyslexia when you read them? I am looking at one of those articles and I suspect it is an AI-written fake.
March 2, 2025 at 10:18 PM
As a reviewer for journals and conferences, have you come across article submissions that are grammatically flawless but still make you wonder whether you have dyslexia when you read them? I am looking at one of those articles and I suspect it is an AI-written fake.
Reposted by Shanqing Cai
CDC source:
“We just had word that all our fellows and post doc staff are laid off effective immediately. The famous Epidemic Intelligence Service, aka the Disease Detectives, is no more. That’s 1260 staff.
They are calling this ‘Phase 1’.”
“We just had word that all our fellows and post doc staff are laid off effective immediately. The famous Epidemic Intelligence Service, aka the Disease Detectives, is no more. That’s 1260 staff.
They are calling this ‘Phase 1’.”
February 14, 2025 at 7:58 PM
CDC source:
“We just had word that all our fellows and post doc staff are laid off effective immediately. The famous Epidemic Intelligence Service, aka the Disease Detectives, is no more. That’s 1260 staff.
They are calling this ‘Phase 1’.”
“We just had word that all our fellows and post doc staff are laid off effective immediately. The famous Epidemic Intelligence Service, aka the Disease Detectives, is no more. That’s 1260 staff.
They are calling this ‘Phase 1’.”
Reposted by Shanqing Cai
The most impactful open-source project of today (dixit Vercel VP of AI)
=> huggingface.co/blog/open-r1
=> huggingface.co/blog/open-r1
Open-R1: a fully open reproduction of DeepSeek-R1
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
January 28, 2025 at 12:17 PM
The most impactful open-source project of today (dixit Vercel VP of AI)
=> huggingface.co/blog/open-r1
=> huggingface.co/blog/open-r1
DeepSeek-R1's thinking steps contain a lot cautionary self-checks. It often looks over-cautious to the extent of being mechanical, but it's still really interesting and impressive.
January 26, 2025 at 7:10 PM
DeepSeek-R1's thinking steps contain a lot cautionary self-checks. It often looks over-cautious to the extent of being mechanical, but it's still really interesting and impressive.
Quick summary of the DeepSeek-R1 paper that everyone is talking about: Do away with supervised fine-tuning on instruction-following or CoT data after pre-training. Instead go straight to RL, specifically by using the Group Relative Policy Optimization on reasoning-intensive tasks.
January 26, 2025 at 2:30 AM
Quick summary of the DeepSeek-R1 paper that everyone is talking about: Do away with supervised fine-tuning on instruction-following or CoT data after pre-training. Instead go straight to RL, specifically by using the Group Relative Policy Optimization on reasoning-intensive tasks.
Can feel the tension building between human-in-the-loop (HITL) vs. machine-in-the-loop (MITL) camps building up surrounding human-AI interactions in recent days.
January 19, 2025 at 4:52 PM
Can feel the tension building between human-in-the-loop (HITL) vs. machine-in-the-loop (MITL) camps building up surrounding human-AI interactions in recent days.
@fchollet's latest blog post on the latest progress by OpenAI o3 on the ARC-AGI benchmark:
The examples of simple tasks that are unsolved by o3 all involve spatial reasoning and generalization, a task that humans find natural but AI still struggles at.
arcprize.org/blog/oai-o3-...
The examples of simple tasks that are unsolved by o3 all involve spatial reasoning and generalization, a task that humans find natural but AI still struggles at.
arcprize.org/blog/oai-o3-...
OpenAI o3 Breakthrough High Score on ARC-AGI-Pub
OpenAI o3 scores 75.7% on ARC-AGI public leaderboard.
arcprize.org
December 22, 2024 at 2:51 PM
@fchollet's latest blog post on the latest progress by OpenAI o3 on the ARC-AGI benchmark:
The examples of simple tasks that are unsolved by o3 all involve spatial reasoning and generalization, a task that humans find natural but AI still struggles at.
arcprize.org/blog/oai-o3-...
The examples of simple tasks that are unsolved by o3 all involve spatial reasoning and generalization, a task that humans find natural but AI still struggles at.
arcprize.org/blog/oai-o3-...
In his NeurIPS speech, Ilya Sutskever muses that the next generations of large neural networks might be "self aware". Instead of making them aware of themselves, I think it will be more helpful to first make them aware there is an external world.
https://youtu.be/1yvBqasHLZs
https://youtu.be/1yvBqasHLZs
December 18, 2024 at 1:45 AM
In his NeurIPS speech, Ilya Sutskever muses that the next generations of large neural networks might be "self aware". Instead of making them aware of themselves, I think it will be more helpful to first make them aware there is an external world.
https://youtu.be/1yvBqasHLZs
https://youtu.be/1yvBqasHLZs
"... we tend to be most conscious of the things our intelligent brain does least well, the recent things in evolutionary history such as logic, mathematics, philosophy, and general problem-solving and planning, and we tend to be quite unconscious of its true powers ..." -- Tomaso Poggio (1990)
December 15, 2024 at 6:48 PM
"... we tend to be most conscious of the things our intelligent brain does least well, the recent things in evolutionary history such as logic, mathematics, philosophy, and general problem-solving and planning, and we tend to be quite unconscious of its true powers ..." -- Tomaso Poggio (1990)
Interesting post on the current status of open-source LLMs developed by Chinese companies
xianbao-qian.medium.com/dec-2024-chi...
xianbao-qian.medium.com/dec-2024-chi...
Dec. 2024 Chinese OS LLMs
Eight months ago, I had the privilege of speaking at Seoul National University about the emerging rise of open-source Chinese-speaking LLMs…
xianbao-qian.medium.com
December 5, 2024 at 5:45 PM
Interesting post on the current status of open-source LLMs developed by Chinese companies
xianbao-qian.medium.com/dec-2024-chi...
xianbao-qian.medium.com/dec-2024-chi...
Reposted by Shanqing Cai
🧪 E11 Bio is excited to share a major step towards brain mapping at 100x lower cost, making whole-brain connectomics at human & mouse scale feasible (🧠→🔬→💻). Critical for curing brain disorders, building human-like AI systems, and even simulating human brains.
Read more: e11.bio/news/roadmap
Read more: e11.bio/news/roadmap
December 3, 2024 at 2:58 PM
🧪 E11 Bio is excited to share a major step towards brain mapping at 100x lower cost, making whole-brain connectomics at human & mouse scale feasible (🧠→🔬→💻). Critical for curing brain disorders, building human-like AI systems, and even simulating human brains.
Read more: e11.bio/news/roadmap
Read more: e11.bio/news/roadmap
Designers of artificial languages should take this use case into account.
November 27, 2024 at 5:38 AM
Designers of artificial languages should take this use case into account.
Collaborating with Team Gleason on applications of generative AI to accessibility is one of the most deeply transformative experiences of my career so far. While we have made progress in shedding light on paths forward, much work and many opportunities await us.
teamgleason.org/speakfaster/
teamgleason.org/speakfaster/
SpeakFaster: Team Gleason & Google’s Commitment to Innovation in Communication Technology - Team Gleason
Technology is advancing at an unprecedented pace, and its impact on communities like those living with ALS is monumental. SpeakFaster is a transformative
teamgleason.org
November 26, 2024 at 2:23 PM
Collaborating with Team Gleason on applications of generative AI to accessibility is one of the most deeply transformative experiences of my career so far. While we have made progress in shedding light on paths forward, much work and many opportunities await us.
teamgleason.org/speakfaster/
teamgleason.org/speakfaster/