Abstract Recent “reasoning models” improve LLM performance by generating chains of thought before producing an answer, with the reasoning process itself optimized through reinforcement learning. This paradigm has delivered impressive results on math and coding[…]
Abstract As Alzheimer’s disease (AD) has increasingly become a major global public health issue, speech-based AD detection has attracted widespread attention. However, most existing methods are trained and evaluated on a single dataset, often leading[…]
Abstract Benchmarking is seen as critical to assessing progress in NLP. However, creating a benchmark involves many design decisions (e.g., which datasets to include, which metrics to use). With benchmarks now growing larger and larger,[…]
Abstract In this talk, we will survey the evolving foundations of AI interpretability, tracing a progression from early behavioral probing to feature-attribution techniques, correlation-based representation analyses, and finally causal approaches. We will examine how each[…]
Abstract Spoken language modeling remains significantly less efficient than text modeling, largely due to a mismatch in sequence granularity: speech is represented as dense frame-level tokens, resulting in long and redundant sequences that are poorly[…]
Abstract Recent proposals urge AI labs to prepare for “AI welfare” under uncertainty about whether AI systems have morally relevant inner states. We do not argue for or against the possibility of AI welfare. We[…]
Abstract While Large Language Models (LLMs) excel in reasoning, whether they can sustain persistent latent states remains under-explored. The capacity to maintain and manipulate unexpressed, internal representations-analogous to human working memory-is a cornerstone of complex[…]
