Line of inquiry
Inquiring lines›How do language models construct a…›How are AI-generated and human-wri…›this line of inquiry
Does domain specialization cause models to lose capabilities elsewhere?
A broader line of inquiry — a family of 13 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 13
Specific inquiring lines the field asks around this — ordered from the most general framing down to the most specific angle.
- Does specialized training in one domain create capability cliffs elsewhere?
- How does over-specialization create capability cliffs outside target domains?
- What causes models to develop domain capability cliffs after specialization?
- Why does over-specialization create a domain capability cliff in LLMs?
- Why do fine-tuned models fail outside their specialized domains?
- What capability risks emerge when models are optimized for single domains?
- What distinguishes domain-specific failure modes from general model limitations?
- How does retrieval-augmented training reduce domain specialization cliff failures?
- Where do frontier AI models already exceed safety thresholds in capability areas?
- Can capability boundary collapse be addressed by operating at representational rather than token level?
- Why do most frontier models terminate early on long-horizon benchmarks?
- Can expert-derived knowledge bases scale to other high-stakes domains?
- Can review effort alone keep pace with frontier model degradation?