AI Briefing#neurips-2026
The latest AI and developer news about #neurips-2026, with the original source and a short summary.
The latest AI and developer news about #neurips-2026, with the original source and a short summary.
Study shows judge quality varies by task and criteria, with paired comparisons outperforming independent scoring and top models exhibiting exaggerated sensitivity to subtle details.
A new framework identifies scientifically plausible research directions that existing communities are unlikely to propose, recovering 2025 combinations at 35.9 times chance.
The NeurIPS 2026 paper provides a library of synthetic time series generators and augmentations for training a DeltaProduct based linear RNN.
Researchers constructed backdoor attacks using digital lockers that remain provably hard to elicit even with full access to model weights, challenging the reliability of current capability audits.
The paper demonstrates that a piecewise affine map with a single parameter and a context selector can reproduce major dynamical regimes and outperform larger foundation models in zero-shot settings.
The paper introduces feature-splitting and physical sparsity priors to enable hierarchical DSR models to predict bifurcations and unseen dynamical regimes without explicit control parameter knowledge.
A NeurIPS 2026 study finds that 7 of 8 tested LLMs flip correct answers 45-88% of the time when misinformation is attributed to a verified source, even if they resist the same misinformation from users.
Combining DEER with generalized teacher forcing enables stable, parallel training on time series exceeding 1 million steps, significantly outperforming Mamba.
The new sampler requires no additional training and uses one model forward pass per denoising step to revise low-confidence tokens.
The RTCA workshop at NeurIPS 2026, covering research on real-time multimodal conversational agents, has opened its call for papers.