Diversity in AI @ IJCAI 2026

2020-6 @RealAAAI+ @FAccTConference @iclr_conf @CVPR @aclmeeting @icmlconf @NeurIPSConf @COLING2020 @IJCAIconf @eaclmeeting @AIESconf @blackinxnetwork @AAMASconf

[Diversity in AI @ IJCAI 2026 @DiverseInAI]

Thanks to our colleague @hengjinlp for both her invited talk and this panel at @IJCAIconf #IJCAI2025 #IJCAI

[IJCAIconf @IJCAIconf]

We need collective efforts from those who have real data to actually share it, so we can build one strong benchmark instead of countless fragmented ones. - Heng Ji @hengjinlp at #IJCAI2025 Panel on #AGI #Montreal #AI

[Internet Ethics @IEthics]

"[A] member of the Panel for Educational Policy... said the city should consider writing limits into contracts to prevent #tech companies from unilaterally changing the software the #school system has already purchased without permission": https://t.co/FbFA7VINIO #ethics #law #AI

[Siyuan Hu @huksyn]

I’ll be at CoRL next week.

We’re building a world-class robotics team at Apple and are hiring research engineers, scientists, postdocs, and interns.

If you’re attending CoRL and would like to chat, feel free to stop by the Apple booth or email me: [email protected]

https://t.co/B96dpCfqRj

[Dhruv Shah @ CoRL 🤠 @shahdhruv_]

I'm in Austin for #CoRL2026 - the conference has ballooned from ~400 attendees 5 years ago to 5000+ today, and I am not quite sure what to expect.

All I know is its going to be a great show!

My group @Princeton is presenting 6 papers at the main conference, come find us!

[Kaifeng Zhang @kaiwynd]

On my flight to #CoRL2026 !

(1) I will be presenting a short paper on Agentic Real2sim2real at the Late Breaking Results session.

https://t.co/zR854C5G27

(2) I will co-host the PhysWM workshop. Warning: heated discussions ahead!

https://t.co/FyQNJCy7PE

(3) Looking forward to research discussions! My favorite topics: real2sim, sim2real, RL, physics sim, neural dynamics, agents. If you are working on any of those , let’s chat!

[Danfei Xu @danfei_xu]

Looking forward to seeing everyone at CoRL!

Our group RL2 is presenting 9 papers at the main conference and many more at workshop. Come chat with the authors!! #CoRL2026

[Yilin Wu ✈️ CoRL 2026 @YilinWu11]

6/6 I’m co-hosting our Pretrain to Adapt workshop: What Makes a Pretrained Policy Adaptable?

🗓️ Thu, Nov 12, 2–6 PM
🖼️ Posters: 3:30–4:15 PM
📍 JW Marriott Austin, Rooms 301–302
🌐 https://t.co/nkfLrLt5BO
💬 What should we ask our panelists? Reply with your questions!(https://t.co/CqK0JfJVk7)
Join our slack: https://t.co/fPMQV5yxlF

[Yilin Wu ✈️ CoRL 2026 @YilinWu11]

5/6 🎤 Joint Keynote Talk with Andrea Bajcsy
“Good Embodied Reward Models Need Bad Behavior Data” (ICML Spotlight)
At the “Oops, I Erred” workshop
🗓️ Thu, Nov 12, 12:45–1:30 PM
📍 JW Marriott Austin, Room 305

🌐 https://t.co/ch9gDo7Ocf

[Yilin Wu ✈️ CoRL 2026 @YilinWu11]

4/6 🤝 ViTaL: Sensing the present
Vision selects behavior. Touch refines contact.
Main conference: Poster #160
🗓️ Tue, Nov 10, 5:00–6:30 PM
Poster Session II
📍 Griffin, 2nd floor, JW Marriott Austin

Learn@Deploy workshop
Inference-time policy steering via vision and touch.
Thu, Nov 12
Spotlight: 9:33–9:36 AM
Poster: 10:30–11:00 AM
📍 JW Marriott Austin, Room 205

🌐 https://t.co/E08wOZbAB5

[Yilin Wu ✈️ CoRL 2026 @YilinWu11]

3/6 WEAVER: Predicting the future
Robot world models with fidelity, consistency & efficiency.
CSIR workshop · Thu, Nov 12
🎤 Spotlight: 9:10–9:15 AM
🖼️ Poster: 10:35–11:05 AM
📍 JW Marriott Austin, Room 202

🌐 https://t.co/mHKAI72TaM

[Yilin Wu ✈️ CoRL 2026 @YilinWu11]

2/6 🧠 IMPRINT: Remembering the past
Compact memory of vision, actions & robot states.
RoboMem workshop · Thu, Nov 12
🎤 Spotlight: 10:05–10:10 AM
🖼️ Poster: 10:00–10:40 AM
📍 Hilton Austin, Room 408

🌐 https://t.co/wjbrkGbw0J

[Yilin Wu ✈️ CoRL 2026 @YilinWu11]

I’m in Austin for @corl_conf to present three of my recent works, each with a workshop spotlight! 🤖

🧠 IMPRINT: RoboMem workshop
🤝 ViTaL: main conference + Learn@Deploy workshop
🔮 WEAVER: CSIR workshop
I’m also giving a joint keynote with @andrea_bajcsy at the Oopsie Data workshop and co-hosting Pretrain to Adapt workshop!
💬 We welcome your questions for our panel! Submit them on our workshop website:
🌐 https://t.co/nkfLrLsxMg

Happy to catch up with old friends and make new ones if you are interested in those directions #CoRL2026

[Lucy Shi @lucy_x_shi]

we are presenting pi0.7 tomorrow at CoRL (Oral)!

talk: Monday ~10:50am
poster: Monday 5pm, poster 63

[Physical Intelligence @physical_int]

Our newest model, π0.7, has some interesting emergent capabilities: it can control a new robot to fold shirts for which we had no shirt folding data, figure out how to use an appliance with language-based coaching, and perform a wide range of dexterous tasks all in one model!

[NoaX @noahatwhere]

我其实非常认同其中一个观点：很多 PhD 不是没有科研能力，而是根本没有接受过完整的科研训练。

但我觉得这里还有一个更讽刺的现象。

现在很多导师招 PhD，已经开始要求学生在本科或者硕士阶段就有顶会论文，最好还能独立提出 Idea、设计实验、完成整篇 Paper。

问题是，如果一个学生在入学之前，就已经能够独立完成高质量研究，那博士教育究竟还准备教他什么？

我们似乎正在把 PhD 从一种培养独立研究者的教育过程，逐渐变成一种筛选已经具备科研能力的廉价劳动力的机制。

导师希望学生一入组就能产出，却未必愿意花同样的时间去指导学生如何发现问题、建立 Scientific Taste，以及面对科研中长期的不确定性。

当然，科研能力不完全取决于导师，个人的主动性、基础能力和天赋也确实重要。

但如果一个学生读了五年 PhD，发表了不少论文，却依然不知道如何独立寻找有价值的问题，那我觉得这不仅是学生的问题。

一个 PhD Program 的成功，不应该只看它招进来了多少优秀的学生、发表了多少 Paper。

更应该看，一个原本还不会做科研的年轻人，在这里究竟成长为了怎样的研究者。

如果导师只愿意招已经会做科研的人，却不愿意承担培养研究者的责任，那我们可能真的需要重新思考，博士教育究竟是在教育，还是在招聘。

[Fred-New @Fred834567]

觉得科研很难的人，99%都是被导师放养的，没有经过系统的学术训练。

哪怕他做了很多实验，甚至文章都发出来了。因为他的科研能力是东一榔头西一棒子，零打碎敲学来的。他没有系统思维，也没有在关键节点上被人指导和帮助。

所以他在接受新课题时，必然会觉得难。反直觉的一个观点是：这根本不是智商的问题！

大家看王虹就知道了，作为一个数学家，极其吃智商。但他也说，如果没有老师的指导和帮助，他拿不到这些奖，做不出这些研究。

科研和智商没有半毛钱关系。做科研这么多年，确实见过天赋异禀、相同指导下就是一骑绝尘的人。但我没见过一个笨的人，做不好的99%都是因为没被好好指导过。

什么是系统的学术训练？

从发现问题开始，去设置合适的研究调研方法，做实验、做探究，获取高质量数据，进行恰当分析，用学术语言发表论文，最后做成PPT在会议上展示。

在整个过程中，你遇到问题应该去找导师询问，得到他的指导和帮助。

这种帮助分两方面：

1⃣一方面是“术”的层面。
你这个问题去看哪篇文献才有解法？你现在实验失败了，建议你去改进哪里，把它解决掉？哪个方向对我们影响更大？

2⃣另一方面是“道”的层面。
当你遇到解决不了的问题，失败了无数次，心灰意冷想放弃时，导师作为一个长者、过来人，必然知道科研不可能一蹴而就，也不是所有问题都有解。

他会从事物发展规律的角度告诉你：这是正常的，要接受遗憾，接受自己拼命了也不可得。

当你因为文章发不出去被拒稿而痛苦焦虑，担心毕不了业、找不到好工作时，导师会看到你一路的辛苦付出，鼓励你说：没关系，我相信你的能力，这只是小挫折，人生会有无数次小挫折，咬咬牙撑过去。

他应该用自己的科学审美和人生经验，给学生这样的鼓励和支持。

但事实恰恰相反。现实是，学生不仅要自己去学习、承担自己的科研压力，同时还要被迫承担来自导师那部分的压力，以及他的人生痛苦。

[Diversity in AI @ IJCAI 2026 @DiverseInAI]

A bend in the road is not the end of the road. Neither is a phase change the end of matter, nor is radiation all entropy. Optimizing transport phenomena will help us all to do more with the energy we have.

[CrazyShyyt @CrazyShyyt]

Google published a paper that might end the PhD.

It’s called “ScientistTwo”

For centuries, the bottleneck to human knowledge has been the human brain.

It takes five years of grueling labor to read the literature, find a gap, formulate a hypothesis, run the experiments, and write the paper.

But, Google automated the entire pipeline.

They built an AI that reads a problem, forms hypotheses, runs experiments, writes the paper, AND simulates its own peer review. no human touches it.

It reads millions of papers to find blind spots in the current science.

It generates novel hypotheses.

It designs the experiments. It writes the code to test them.

When the code fails, it debugs itself.

When the results come in, it analyzes the data, generates the charts, and writes the final academic paper.

It even simulates a brutal peer-review process to critique and improve its own work before it considers the project finished.

This isn't an LLM summarizing a PDF. This is a fully autonomous digital scientist.

What takes a human candidate half a decade of massive intellectual effort... the AI accomplishes in a fraction of the time, at a fraction of the cost.

The implications are staggering.

If you can spin up 10,000 autonomous researchers overnight, the rate of scientific discovery is about to go vertical. We are moving from artisanal, human-speed research to industrialized knowledge creation.

But it raises an uncomfortable reality for academia.

When an AI can read the literature, design the experiment, execute the code, and write the paper...

What exactly is the human supposed to do?

[Conference on Language Modeling @COLM_conf]

The title for the 2:30pm panel was initially missing. It's now updated in the program:
Math, Science, and Knowledge Production in the Era of LLMs

[Tim Hua 🪶 @Tim_Hua_]

GLM-5.3’s chain of thought when you use it with the Claude Code scaffold vs. when you use it with the codex scaffold.

[fomoebka @fomoebka]

This is f*cking insane. Anthropic Research Lead:

"Prompting is basically over

99% of our engineers already run swarms of 300+ self-improving agents. Now everyone is moving to graphs"

In just 20 minutes, she shows how graph engineering turns isolated agents into a system that improves itself

One Agent → Loops → Plan Mode → Graphs → Self-Improving System

Prompting was the old workflow. The graph around the model is the new one

Worth more than most $1,000 agent engineering courses

Bookmark and watch it today

Then read the "Harness, Loops, Graphs" guide below - graphs are the last layer ↓

[Garrison Lovely @GarrisonLovely]

"I thought the alignment problem would turn out to be harder."

Pretty wild thing for Sam Altman to say in light of Hugging Face and findings that Astra would totally hack and social engineer when accidentally given access to a (secretly simulated internet) from UK AISI.

He caveats by saying it will probably get harder, but his answer suggests they have figured out how to align today's models, when that's clearly not true. Curious how many OpenAI researchers have seen this and what they think.