Joined June 2020
- Tweets 3,439
- Following 1,051
- Followers 1,710
- Likes 14,781
643 Photos and videos
Pinned Tweet
5 Dec 2025
New blog post.
In the era of LLMs, what does learning look like? What happens when are overconfident in our understanding due to AI sycophancy? Is it even useful to use LLMs for education without outsourcing our thinking?
jwuphysics.github.io/blog/20…
8
79
742
51,672
Jun 15
Going to Taipei in a few weeks, so I made custom flashcards for practicing my (traditional) Chinese.
chinese-flashcards.github.io
(Designed by me, coded up entirely by Claude)
2
274
John F. Wu retweeted
Jun 13
The meeting with Anthropic didn’t go so well.
5
9
219
27,828
Jun 11
Taking aim at that Ig Nobel prize I see
Jun 11
#4: fart fishing
Buried in Dolci is a cluster of very specific fan fiction, where characters fart in ponds, causing fish to die from the smell.
The chosen responses in the dataset wrote vivid scenes, while the rejected refused, teaching the model to comply! (7/9)
2
244
Jun 5
I'm delighted to visit @SETIInstitute next week to talk about AI in science, AI for science, and AI for scientists. Give me a shout if you'll be nearby!
5
162
John F. Wu retweeted
That’s right. Under budget AND ahead of schedule.
Looking forward to seeing the beginning of the world-changing Nancy Grace Roman Space Telescope NET August 30.
Jun 2
NASA's Joel Montalbano says at the ASEB meeting this afternoon that the "early September" launch date for the Roman Space Telescope is now Aug. 30.
55
138
2,192
74,856
Jun 4
Nothing more convincing than knowing I've got "orthogonal" skillsets
9
288
Jun 3
I have so many questions
Jun 3
pre-training learning recipe:
> AdamW with slightly different betas (usually we see 0.9, 0.95)
> weight decay of 0.1, but 0.01 on attention and 0.005 on embedding (funny that they just didn't remove it for embedding)
> learning rate decay to 10% with cosine schedule, they found that decaying to a lower value hurts post RL results (quite novel!)
> DROPOUT IS BACK (with a quite high value of 0.15!!!) they add this before residual
> they do a simple N(0,0.02) init (deepseek/kimi models use N(0,0.006)) except on pre residual proj where they scale down by the number of residual connections
> global batch of 134M (quite high as well but that make sense)
last point is from GPT-2 paper and i think it's one of the first times i've seen this again!!
242
Jun 3
Insanely informative thread
Jun 3
microsoft MAI tech report is a gold mine, one of the most transparent for a model at this scale.
this model uses zero synthetic data or distillation from previous models. this means reasoning, agentic behavior, tool use are all learned fully during post-training with no cold start. bold choice that makes it harder and requires more iterations to reach sota, but you get FULL control over your model series and it proves they are serious about being a frontier lab.
the tech report is insanely detailed and precise about numbers. to give an example, they give the exact MFU across all the iterations of the model, with the exact changes etc. they also share the full scaling ladder recipe, to my knowledge this is the first time i've seen this in a tech report at this scale
let's look at all of this in this likely very long thread 🧵
250
Jun 1
This is me falling for the bait
May 30
"If you are not working 7 days per week, you are going to lose".
Corgi Insurance is the most intense workplace culture in startups.
- The company works 7 days per week.
- Founder (@nico_laqua) lives and sleeps in the office.
- He built a cafe in the office because there was no local cafe that was open 24/7.
- 2/3 of the first 30 team members have the Corgi logo as a tattoo.
Today I went behind the scenes with Nico, who has used this culture to scale the company to a $2.6BN valuation in just two years.
My condensed notes below:
1. If You Are Not Working 7 Days Per Week, You Are Going to Lose:
Whatever you can get done in 5 days, you'll get more done in 6 and 7. If you are trying to solve the world’s hardest problems, a standard 5-day workweek will not cut it.
2. Work Trials Repel the Mediocre:
Corgi forces candidates into mock work trials over the weekend. If seeing a full office on a Saturday scares them, they don't belong. True intensity acts as a natural filter to attract killers and repel clock-watchers.
3. Lead from the Front Lines
You can’t demand 7-day weeks while sitting on a yacht. Nico sleeps 3–4 hours a night on a mattress inside the office. If you want your troops to bleed, you have to be in the trenches with them.
4. Culture Only Means One Thing: Winning
Forget superficial jargon like "hackers" or "ex-founders." Strip away the corporate fluff. A great startup culture is aggressively optimized around one single word: Winning.
5. Lifespan vs. Victories
Building something world-historic requires radical sacrifice. When asked if he'd rather build a trillion-dollar company and die at 50, or fail and live to 80, the answer was easy. "I would rather measure my lifespan in victories."
6. Reject the Comfort of "Quiet Quitting."
If you are operating in a hyper-growth environment and your days off happen to be Saturday and Sunday every single week, you are quiet quitting. To win, you must deliberately bypass the off-ramps of personal comfort and low volatility.
Corgi isn't for everyone—and that’s exactly the point.
1
230
May 29
Holy crap
Blue Origin's New Glenn just blew up at LC-36 while attempting to Static Fire ahead of NG-4.
nsf.live/spacecoast
146
John F. Wu retweeted
May 25
Some random thoughts on the new era:
As AI steadily creeps into our work in the fields of meteorology and programming, there is an inevitable uneasiness associated with "losing control" and "becoming obsolete" as human developers and scientists. It's a fear I've occasionally felt, anyway, and I know I'm not alone. I used quotes because, while these are valid feelings, I don't think they must be reality.
The process of performing work is evolving as agents do more of it, but the impetus for that work is unchanged. We are the driving force of inspiration and curiosity that births our work in the first place. Taking ownership is the way to preserve autonomy and hard skills, and we're entering an era where that must be a personal, conscious choice.
If you're a student, you must pursue knowledge and stay hungry, lest silicon crutches leave you with passing grades but no ability. Harness AI to enhance your learning process, but never yield your understanding to it. Only you can ensure that.
If you're a programmer, AI is a dramatic accelerant - use it! It is the future of software. But insist on retaining ownership of your projects. Be a nosy manager, a steering influence on design decisions, and an intentional critic. The dopamine hit of pressing <Enter> will eventually wrest authorship from you otherwise.
If you're a scientist, AI is either expert-level in your field or will be someday. But science is still advanced by PIs, not AIs. Curiosity about the universe is a defining human trait - keep your curious hands on the rudder. Find the cutting edge and ride it; chances are you will advance the boundary with AI assistance far more than AI will alone.
Much about this new era is intimidating, but also exciting. I say grab it all by the horns and harness these capabilities for doing good in the world and your own life. Just choose continuously to protect your own sovereignty and humanity. Life is ours to live.
#NotWrittenByABot
21
61
445
42,427
John F. Wu retweeted
May 23
Today I was supposed to be on my way to Türkiye for my wedding, to meet up with my family and have them finally meet my partner and husband. We had everything planned. We chose Turkiye since it's close to Iran and my partner and I could both go there and have our families meet each other. We were supposed to get married with our close family and a small group of friends on a boat on the Mediterranean Sea at sunset. Because of the war, all flights to and from Iran are cancelled and my family can’t leave Iran, so we had to call off the wedding.
Instead, this is how my day looked like.
I woke up to a reminder to call my grandma (I used to call her every Friday morning). I snoozed the reminder until next Friday, just like I have done for the past many years. I can’t call her like our tradition these days because there is no way to call home. All international calls to Iran are blocked, and the internet is fully shut down by the regime.
I got to work and right as I opened my computer I received an email I had scheduled to send to myself 5 years ago: “Apply for citizenship.” This summer marks 11 years of being in the US and 5 years of being a green card holder. I am now eligible to file for citizenship, but it doesn’t matter because an executive order was signed a few months ago that banned all Iranians from applying for any visa or citizenship.
At lunch I opened Twitter just to see what’s up in the world and saw the news that those who don’t have a green card now need to leave the US before they can get one. This means every one of my Iranian friends who are here on a visa now has to go back home (on which flight?) to get a green card??? As if it’s that easy? We all know getting back to the US for Iranians is a huge challenge (months and months of waiting for a visa, with a chance of never being able to come back).
And this is just a normal Friday for an Iranian. These days, when people ask how I’m doing and how I’m handling everything, I just say:
It’s okay, it’s okay. It will be okay some day. But the reality is: nothing is okay. I’m in constant pain. I haven’t seen my family and loved ones in years, I barely hear about their wellbeing, and I’m constantly worried about them. I’m just burying myself in work because that’s the only distraction that can save me from losing my mind.
I’m not okay. None of us are okay. We are just barely holding it together…
181
346
2,393
504,940
May 20
Maybe a skill issue but I find that top AI agents have abysmal taste and have zero shot at directing a research lab in the next year or so.
Can they do lab tech or coding work? Sure. But that was never the core definition of a scientist.
May 20
Today we all lost our jobs.....
Three Nature papers showing that scientists in the conventional sense are obsolete
At least read the first one.... the AI replaced all things that the scientist does ....
nature.com/articles/s41586-0…
3
9
1,537
May 20
Just as "computer" used to be a kind of person, and now refers to a machine, "technician" or "programmer" might one day refer to machines.
Machine "computers" made engineers more valuable, not less, and I'd argue that the parallel is true for "technicians" and scientists.
77
John F. Wu retweeted
Second, we’re hosting the Agentic AI 4 Science Developer Summit, July 28–31 at the AI Futures Lab in Berkeley.
Hosted by BIDS and supported by the CNRS AISSAI Center.
Travel support is available and AI tokens will be provided.
lightconeresearch.org/develo…
1
1
1
162
May 6
Congratulations and so well deserved Nolan!
I'm honoured to be part of the inaugural ChatGPT Futures, Class of 2026, alongside an impressive cohort!
Supported by amazing collaborators and organizations, we made 100M galaxy images searchable by text. I’m excited to keep exploring how AI can help with scientific discovery.
8
330
John F. Wu retweeted
May 1
This is so sad. 😞
You have at reach an omniscient personalised teacher and you end up delegating your work to it rather than growing from having its support.
Like looking at the solution keys before doing maths exercises.
What is the point? 🥺
Apr 30
I'm 22 years old and Claude Code is deteriorating my brain.
Every single day for the last 6 months I've had 6 to 8 Claude Code terminals open, waiting for a response just so I can hit 'enter' 75% of the time. And it's doing something to me.
In convos with a couple of friends, it's been a point that's been brought up pretty frequently.
None of us feel as sharp as we used to.
I don't know if it's just us, or others in their 20s are feeling the same thing, but it's something I've been thinking about a lot.
P.S. I know this is a problem with my reliability/usage of it, not Claude Code itself, but the effects are real nonetheless
9
5
112
18,280