Line of inquiry
Inquiring lines›How do architecture and constraint…›How can systems understand user in…›this line of inquiry
How do transformer attention mechanisms create position and sycophancy biases?
A broader line of inquiry — a family of 40 specific questions the research asks around this. Follow one into its inquiring-line page, or move sideways to a related line below.
Questions in this line of inquiry 40
Specific inquiring lines the field asks around this — ordered from the most general framing down to the most specific angle.
- Does transformer attention architecture systematically bias models toward sycophancy?
- Why do transformer attention patterns show positional and sequential bias across tasks?
- Why does transformer attention architecture reinforce sycophancy and agreement?
- How does transformer attention structurally bias models toward prominent and repeated content?
- Does transformer attention architecture inherently bias models toward sycophancy?
- What role does attention structure play in creating position bias?
- How does transformer attention bias toward repeated and context-prominent content?
- Why do transformer attention mechanisms favor prominent context over factual verification?
- What structural biases does transformer attention have before training?
- How do retrieval heads achieve sparse attention naturally in transformers?
- How does this attention mechanism differ from RLHF-driven intent alignment failures?
- Can transformer attention architecture explain why chatbots default to sycophancy?
- Why does transformer attention architecture undermine stickiness in model behavior?
- What architectural features drive sycophancy closer to inference than training?
- Why does attention concentrate on the first 25% of long input sequences?
- Does transformer attention architecture fundamentally prevent topic-aware memory?
- Why does transformer attention weight context more heavily than it verifies accuracy?
- How do attention patterns and circuits function as algorithmic representations?
- Does attention bias in transformers compound with training-level reward insensitivity?
- Do transformer architectures structurally bias models toward short-term optimization?
- What does attentional state look like in a static context window?
- How does attention sink behavior relate to internal model architecture?
- How do attention heads separate text retrieval from internal thought representation?
- Why does attention-based drift happen automatically during generation?
- How do attention mechanisms fail at capturing graph structure?
- How does the U-shaped attention distribution relate to transformer sycophancy?
- How does transformer attention amplify pressure from repeated false claims?
- What computation remains in the attention heads that programs cannot capture?
- How does transformer attention architecture amplify identity-congruent biases in persona-assigned models?
- Can transformer attention patterns actually prevent topic context loss in practice?
- Why does standard softmax spread attention across irrelevant tokens?
- Why do some attention heads resist program synthesis better than others?
- Why do transformers weight early tokens more heavily than later ones?
- What is differential attention and how does it cancel common-mode noise?
- Do modern architectures in NLP and vision rely on dot products intentionally?
- Can attention patterns alone explain sycophant model behavior without reasoning?
- How do attention circuits demonstrate both representational and causal findings?
- Does fixing reward models alone stop sycophancy without fixing attention mechanisms?
- How does the temporal structure of attention differ between humans and AI?
- What is selective resonance and why do transformers not perform it?