Omar Khattab on Late Interaction in 2030
Omar Khattab argues that single-vector retrieval is running out of road, and the gap between what dense embeddings can do and what reasoning models can do represents the most exciting open problem in AI right now. This talk covers the first principles case for late interaction, the accumulation of evidence over the years, and a roadmap for where the field needs to go by 2030.

▶︎
Demis Hassabis: We're Three Quarters of the Way to AGI

▶︎
AI Bubble: How AI's push towards IPOs became a death drive | Ed Zitron

▶︎
Turing Award Winner: Disagreeing with Google, Postgres, Future Problems | Mike Stonebraker

▶︎
Ep. 8 — ColBERT + ColBERTv2: late interaction at a reasonable inference cost

▶︎
Yann LeCun's $1B Bet Against LLMs

▶︎
Omar Khattab on the State of DSPy

▶︎
'Listen Like You Might Be Wrong': Harvard Student Goes Viral For Stunning Speech On Trump Amid Feud

▶︎
It finally happened

▶︎
🇩🇪 German industry JUST died (it’s WORSE than you think)

▶︎
Terence Tao – How the world’s top mathematician uses AI

▶︎
FULL DISCUSSION: Google's Demis Hassabis, Anthropic's Dario Amodei Debate the World After AGI | AI1G

▶︎
Zig 2026: No-AI Policy, $670K Foundation, Left GitHub & Why Zig Isn’t 1.0 - Andrew Kelley Explains

▶︎
Lisa Su Address to MIT Class of 2026

▶︎
Ray Dalio: 'Expect a Tribute System' as China Influence Grows

▶︎
This is not the AI we were promised | The Royal Society

▶︎
The Uncomfortable Truth About AI “Reasoning” | World Science Festival

▶︎
Transformers, the tech behind LLMs | Deep Learning Chapter 5

▶︎
On Engineering AI Systems that Endure The Bitter Lesson - Omar Khattab, DSPy & Databricks

▶︎
Ilya Sutskever – We're moving from the age of scaling to the age of research

▶︎
