相关论文: Imperative Interference: Social Register Shapes In…

Language models are trained to follow instructions, but they are also powerful pattern completers. What happens when these two objectives conflict? We construct conversations in which a user instruction to behave in a target way T (e.g.,…

Instructions trigger a production-centered mechanism in language models. Through a cognitively inspired lens that separates language processing and production, we reveal this mechanism as an asymmetry between the two stages by probing…

Large language models are proliferating, and so are the benchmarks that serve as their common yardsticks. We ask how the agglomeration patterns of these two layers compare: do they evolve in tandem or diverge? Drawing on two curated proxies…

Language model finetuning often results in learning undesirable traits in combination with desired ones. To address this, we propose inoculation prompting: modifying finetuning data by prepending a short system-prompt instruction that…

Large language models trained on a mixture of NLP tasks that are converted into a text-to-text format using prompts, can generalize into novel forms of language and handle novel tasks. A large body of work within prompt engineering attempts…

As large language models are deployed as autonomous agents with tool execution privileges, a critical assumption underpins their security architecture: that model errors are detectable at runtime. We present empirical evidence that this…

Large language models (LLMs) are increasingly deployed with hierarchical instruction schemes, where certain instructions (e.g., system-level directives) are expected to take precedence over others (e.g., user messages). Yet, we lack a…

It is not only what we ask large language models (LLMs) to do that matters, but also how we prompt. Phrases like "This is urgent" or "As your supervisor" can shift model behavior without altering task content. We study this effect as…

Large language models (LLMs) in research and development toolchains produce output that triggers attribution of agency and understanding -- a cognitive illusion that degrades verification behavior and trust calibration. No existing…

Current AI safety relies on behavioral monitoring and post-training alignment, yet empirical measurement shows these approaches produce no detectable pre-commitment signal in a majority of instruction-tuned models tested. We present an…

Both humans and large language models are able to learn language without explicit structural supervision. What inductive biases make this learning possible? We address this fundamental cognitive question by leveraging transformer language…

Imitation learning is a popular approach for teaching motor skills to robots. However, most approaches focus on extracting policy parameters from execution traces alone (i.e., motion trajectories and perceptual data). No adequate…

While instruction-tuned models have shown remarkable success in various natural language processing tasks, accurately evaluating their ability to follow instructions remains challenging. Existing benchmarks primarily focus on common…

This paper measures the impact of increased exposure on whether learned construction grammars converge onto shared representations when trained on data from different registers. Register influences the frequency of constructions, with some…

Large language models' behavior is often shaped by instructions such as system prompts, refusal boundaries, privacy constraints, and tool-use rules that must hold at inference time. Yet in practice these constraints can be violated under…

Large language models are able to perform a task by conditioning on a few input-output demonstrations - a paradigm known as in-context learning. We show that language models can explicitly infer an underlying task from a few demonstrations…

Social learning is widely observed in many species. Less experienced agents copy successful behaviors, exhibited by more experienced individuals. Nevertheless, the dynamical mechanisms behind this process remain largely unknown. Here we…

When a user tells an AI system that someone "should not" take an action, the system ought to treat this as a prohibition. Yet many large language models do the opposite: they interpret negated instructions as affirmations. We audited 16…

How do learners acquire knowledge of what is unacceptable without negative evidence? Construction Grammar proposes statistical preemption: exposure to a conventional form (e.g., "donated the books to the library") preempts structurally…

Emerging research on bias attribution and interpretability have revealed how tokens contribute to biased behavior in language models processing English texts. We build on this line of inquiry by adapting the information-theoretic bias…