The Archive

Large language models (LLMs) are known to abandon their initial stance to conform to user pushback. While prior research largely attributes this behavior to sycophancy learned during reinforcement learning from human feedback, we hypothesize that conformity is also driven by a model's epistemic uncertainty at inference time. In this paper, we introduce MUSE, a two-stage evaluation framework to disentangle the mechanisms driving LLM conformity. Specifically, MUSE maps a model's epistemic uncertainty in responding to a query against its likelihood to yield to user pushback in a subsequent turn....

The three finalists for the World Press Photo of the year. | Image: World Press Photo We love to muse over how "real" photography is defined here at The Verge now that generative AI is so prolific, and the World Press Photo competition might have the answer. The prestigious award celebrates the best of photojournalism, where capturing reality is paramount. The winning entry for 2026 - "Separated by ICE," captured by photojournalist Carol Guzy - was announced yesterday. The harrowing photograph shows children clinging to their father after an immigration hearing. The photo had to abide by spec...

Analysis of how constrained compute supply consolidates market power, testing aggregation theory in AI infrastructure.

Reddit commentary on Meta Muse Spark ranking 4th on Artificial Analysis Index; notes competitive market among Google, Anthropic, OpenAI, Meta, xAI.

Meta Superintelligence Labs ships Muse Spark, first frontier model on proprietary infrastructure stack.

Meta AI announces Muse Spark, a system positioning toward personalized superintelligence via scaling.

We’ve created MuseNet, a deep neural network that can generate 4-minute musical compositions with 10 different instruments, and can combine styles from country to Mozart to the Beatles. MuseNet was not explicitly programmed with our understanding of music, but instead discovered patterns of harmony, rhythm, and style by learning to predict the next token in hundreds of thousands of MIDI files. MuseNet uses the same general-purpose unsupervised technology as GPT-2, a large-scale transformer model trained to predict the next token in a sequence, whether audio or text.