AI research | The Decoder

Jonathan Kemper

Aug 6, 2024

Stable Fast 3D brings better lighting, textures, and materials to AI-generated 3D models

Maximilian Schreiner

Aug 5, 2024

Oxford researchers unveil MedSAM-2, an AI that could change how doctors analyze medical images

Kim M. Scheurenbrand

Aug 3, 2024

ChatGPT trumps human authenticity, relevance, coherence in political debate study

Jonathan Kemper

Aug 2, 2024

Nvidia uses Apple Vision Pro to record humans controlling robots for more realistic training data

The scaling of training data has already led to great success with language models. The same tactics should also lead to great progress in robotics.

Jonathan Kemper

Jul 31, 2024

Google DeepMind's new Gemma 2 model outperforms larger LLMs with fewer parameters

Jonathan Kemper

Jul 30, 2024

AI data isn't destroying AI models after all, researchers say

A recent study published in Nature appears to demonstrate model collapse in several AI architectures, but other researchers say the study makes unrealistic assumptions that are inconsistent with common practice, such as discarding all previous data after each iteration. The critical researchers disagree with the theory that training AI models on synthetic data inevitably leads to "model collapse," where performance gradually declines until the model becomes inefficient. This effect is not a real threat under realistic conditions, they say. With LLaMA 3.1, Meta shows that synthetic data generation processes that iteratively correct errors can improve performance without causing model collapse.

Jonathan Kemper

Jul 28, 2024

Study reveals major weaknesses in AI's ability to understand diagrams and abstract visuals

Colorful AI illustration of a clock that is impossible to read.

Jonathan Kemper

Jul 25, 2024

Stable Video 4D creates moving 3D models from video

Illustration from the Stable Video 4D paper showing dynamic 3D objects.

Matthias Bastian

Jul 25, 2024

Google DeepMind's latest AI models might bring us one step closer to LLMs that can reason

Maximilian Schreiner

Jul 25, 2024

Rule-Based Rewards: OpenAI provides insight into the GPT-4 safety stack

hand-drawn illustration of a digital newspaper