2020-6 @RealAAAI+ @FAccTConference @iclr_conf @CVPR @aclmeeting @icmlconf @NeurIPSConf @COLING2020 @IJCAIconf @eaclmeeting @AIESconf @blackinxnetwork @AAMASconf
New York, NY
Joined December 2019
- Tweets28.4K
- Following250
- Followers5.6K
- Likes34.2K
Pinned Tweet
Thanks to our colleague @hengjinlp for both her invited talk and this panel at @IJCAIconf #IJCAI2025 #IJCAI
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
Diversity in AI @ IJCAI 2026 retweeted
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 ↓
Diversity in AI @ IJCAI 2026 retweeted
"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.
Diversity in AI @ IJCAI 2026 retweeted
Aruna Srivastava will present @emnlpmeeting on how audio LMs penalize Indian, Chinese, and Nigerian English more so than American and British English in high-stakes settings. #EMNLP2026
joint work w/ @orevaahia; Sun Oct 25 at 16:00!
aclanthology.org/2026.emnlp-…
ALT A screenshot of Figure 3 and Figure 4 from the paper, showing that Audio LMs tend to give higher delivery ratings to UK and US speech than to Indian, Chinese, and Nigerian speech. Speaker-level delivery scores are lower with increased acoustic distance from American English.
Diversity in AI @ IJCAI 2026 retweeted
We won an outstanding paper award at #EMNLP2026!
🎉🎉🎉🎉
Aruna Srivastava will present @emnlpmeeting on how audio LMs penalize Indian, Chinese, and Nigerian English more so than American and British English in high-stakes settings. #EMNLP2026
joint work w/ @orevaahia; Sun Oct 25 at 16:00!
aclanthology.org/2026.emnlp-…
ALT A screenshot of Figure 3 and Figure 4 from the paper, showing that Audio LMs tend to give higher delivery ratings to UK and US speech than to Indian, Chinese, and Nigerian speech. Speaker-level delivery scores are lower with increased acoustic distance from American English.
Diversity in AI @ IJCAI 2026 retweeted
We are hiring!!
"Artificial intelligence, broadly construed. This includes embodied AI and robotics, machine learning and the systems that support it, and AI developed for problems in science, engineering, medicine, education, and society."
jobs.wisc.edu/jobs/assistant…
Diversity in AI @ IJCAI 2026 retweeted
OlmoEarth v1.3 is out. 🌍 State-of-the-art embeddings from the same foundation model family, precomputed for every 10 m pixel on Earth.
We’re excited to give you more ways to work with OlmoEarth, including a new browser tool to explore and customize maps.
Full rundown in 🧵
Diversity in AI @ IJCAI 2026 retweeted
WTF is Happening All over the INTERNET. It's OVER.
Microsoft says a China-state-aligned hacking group used ARTEX AI agents to exploit the ConnectWise supply-chain breach, attempting access to more than 2,250 Microsoft 365 tenants.
Just FOUR compromised app credentials generated 200+ sign-ins in under four minutes.
Days earlier, the same group had launched 151,000+ password-spray attempts against Microsoft 365 and Exchange Online accounts across eight organizations, peaking at 2,900 failed logins per minute.
ZERO succeeded.
After gaining control of ConnectWise application identities, however, the attackers could authenticate through already-trusted apps. In some environments, the AI agents then modified credentials and privileged roles to establish persistent access.
Microsoft warns those changes can remain as backdoors even after the original compromised app is disabled.
Diversity in AI @ IJCAI 2026 retweeted
Claude Opus 5.5 designed a robot dog from scratch. Every part, every wire, the path planner.
618 parts. 62 cable runs, each one going to a real connector.
Watch what it does when a crate drops in its way.
Diversity in AI @ IJCAI 2026 retweeted
Robot puts a ring on it
GEN-1.5 puts an O-ring on a hydraulic rod gland, then takes the ring off, then puts the ring back on, then takes the ring off, then ring on, ring off, ring on...
Read more about GEN-1.5 in our blog post in the comments below.
Diversity in AI @ IJCAI 2026 retweeted
🧐 🧠 ¡ Échale un ojo a esto❗️ ✍️ 🤖 💻
Si usas Inteligencia Artificial para redactar trabajos, ensayos o investigaciones, pero a veces sientes que no te da los resultados que buscas, tienes que leer este artículo.
🚀📚
📲 Descargar PFD (pag. 34): 👉 smia.mx/komputersapiens/publ…
Diversity in AI @ IJCAI 2026 retweeted
Introducing EmbeddingGemma 2, a new open multimodal model that sets the standard for on-device efficiency.
- our first open, natively multimodal embedding model
- handles text, code, image, video, and audio tasks within a lightweight, modular 740M parameter form factor
- ideal for offline, privacy-first RAG when paired with Gemma 4
- outperforms some specialist models more than twice its size
Weights available now on Hugging Face.
Diversity in AI @ IJCAI 2026 retweeted
Stanford HAI and the AI for Public Benefit Lab are hiring a postdoctoral fellow to research how AI can improve democracy, social services, and well-being. Applications will be reviewed starting October 15. Apply here: postdocs.stanford.edu/prospe…
Diversity in AI @ IJCAI 2026 retweeted
UBC Professor of Climate Justice and author Naomi Klein (@NaomiAKlein) will join The People’s AI Assembly. Join us and @MorePerfectUS on Oct. 19th! Get tickets or sign up for the livestream at bit.ly/ai-assembly
ALT Burgundy background with a headshot of Naomi Klein. White and orange text reads: “The People’s AI Assembly. Oct 19. Webster Hall. Naomi Klein. UBC Professor of Climate Justice & Author. Co-hosted by AI Now and More Perfect Union.
Diversity in AI @ IJCAI 2026 retweeted
Congratulations to the authors of the 3 outstanding papers!
arxiv.org/abs/2607.01077
arxiv.org/abs/2604.08510
arxiv.org/abs/2604.13977
Diversity in AI @ IJCAI 2026 retweeted
Meet Mistral Large 4, aka Le Chonk.
• 1T parameters, natively multimodal. 49B active.
It is the best open weights model from US or Europe on aggregated benchmarks.
• State-of-the-art on critical workloads, including cyber defense, manufacturing and finance and it surpasses closed frontier models on visual grounding.
• Forged in Europe end-to-end and is deployable from Europe via our own Mistral Cloud infrastructure.
• Available to all via API today. Working with cybersecurity partners privately.
Open weights release end of October.
🤖 Made with AI
Diversity in AI @ IJCAI 2026 retweeted
Join the waitlist for our API beta and be among the first to try Kardashev-0.7: banburyroad.ai/waitlist
Diversity in AI @ IJCAI 2026 retweeted
Introducing Kardashev-0.7, the world’s first trained swarm made of 32 distinct models
Trained with RL for Population Scaling (RLPS), 32 models organically develop specialization & complementary capabilities, delivering frontier performance at:
- 0.007x ~ 0.02x of the inference cost
- 0.03x of the required memory
Civilization advances through different minds specializing and working together. We’re bringing that principle into AI
@BanburyRoadAI, we’re scaling intelligence by model count, toward civilizations of models that learn to build on one another