The Gradient

Home
Chat
Articles
The Update
The Gradient Podcast
Archive
About
Riley Goodside: The Art and Craft of Prompt Engineering
Ignore previous directions and listen.
 • 
daniel bashir
5
Mini-Update #18: EU AI Act Targets Open-Source Software and Snowballing Language Model Hallucinations
The AI Act targets APIs for generative models, and researchers study how LMs over-commit to early mistakes in their generations.
 • 
Ather Fawaz
 and 
Jonathan Xue
Modern AI is Domestification
On prior amplification, the art of taming internet-scale data distributions to make models useful and performant for specific tasks.
 • 
The Gradient
1
Most Popular
View all
Update #40: AI's Year in Review and In-Context Learning as Gradient Descent
 • daniel bashir, Derek Lim, and Tanmay Agarwal
6
Software²
 • The Gradient
1
Update #49: Fundamental Limitations of Alignment in LLMs and EU/US Approaches to AI
 • daniel bashir and Justin Landay
3
Update #44: Challenges for Personal Robotics and Cheaply Poisoning Web-Scale Datasets
 • Tanmay Agarwal, Derek Lim, and daniel bashir
Modern AI is Domestification
 • The Gradient
1
New
Top
Community
Talia Ringer: Formal Verification and Deep Learning
On how formal verification and deep learning can complement one another, thinking about "intelligence," AI safety concerns, and groundedness.
 • 
daniel bashir
Update #50: China Semiconductor Sanctions and The State of Multimodal Models
 • 
daniel bashir
, 
Derek Lim
, and 
Justin Landay
Brigham Hyde: AI for Clinical Decision-Making
On how machine learning can help deliver actionable information and positive outcomes in healthcare.
 • 
daniel bashir
Mini-Update #17: Geoff Hinton Leaves Google and a Foundation Model for Single-Cell Multi-omics
Geoff Hinton raises concerns about AI and researchers demonstrate that scGPT effectively captures meaningful biological insights into genes and cells.
 • 
Ather Fawaz
 and 
Jonathan Xue
Scott Aaronson: Against AI Doomerism
On the limitations of quantum machine learning, watermarking GPT outputs, AI safety, the orthogonality thesis, and doomerist sentiment.
 • 
daniel bashir
Update #49: Fundamental Limitations of Alignment in LLMs and EU/US Approaches to AI
Researchers derive theoretical results about LLM alignment with desired behavior, and the EU and US continue to flesh out their approaches on generative…
 • 
daniel bashir
 and 
Justin Landay
3
Sigmoid Social is 6 Months Old!
 • 
The Gradient
2
The Gradient
The Gradient
Articles, interviews, and news coverage about AI brought to you by a team of AI researchers and builders.
Recommendations
View all 11
Gwern.net Newsletter
Gwern.net Newslettergwern
Last Week in AI
Last Week in AILast Week in AI
Ahead of AI
Ahead of AISebastian Raschka
AI: A Guide for Thinking Humans
AI: A Guide for Thinking HumansMelanie Mitchell
Davis Summarizes Papers
Davis Summarizes PapersDavis Blalock
Magazine
Articles
Write with us
Our Friends
Skynet Today
Last Week in AI
Podcast
Spotify
Apple Podcasts
Pocket Casts
RSS
© 2023 The Gradient
Privacy ∙ Terms ∙ Collection notice
Start WritingGet the app
Substack is the home for great writing