Unverified50% confidenceFactExact time
The authors note that the limitations of the experiments include: small event counts (only 1 to 4 per cell), use of only one agent model and one random seed, both the decision model and the agent being general-purpose models, and the corpus being a tiny fictional company, allowing only structural analogy.
1
Sources
50%
Confidence
Long-term
Relevance
10/19/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe experiment has clear boundaries: a single task (invoice extraction), a single model family, and only 4 repetitions per size point; the 95% confidence interval for 25% is very wide, algorithmic padding is not entirely equivalent to hand-written bloat, and the conclusions should be regarded as a measurement starting point with a clear mechanism rather than a universal constant.75% similarUnverifiedThe experiment's limitations include: only one model, 10 closed wars per copy, three rounds of small-scale supervised fine-tuning (expert iteration based on the model's own best wars), not full reinforcement learning; 6 to 4 out of 10 could be chance, but 0 to 4 is a stronger signal.73% similarUnverifiedResearchers have found that fine-tuning only a small number of model parameters is sufficient to remove safety alignment (de-alignment attacks), and that certain jailbreak prompts can transfer across different models.71% similarUnverifiedWhen troubleshooting a model that fails to converge, one can first test whether the model can overfit a very small data subset; if it cannot learn even a few samples, there is most likely a bug in the code or configuration.70% similarUnverifiedThe researchers candidly acknowledge limitations including a limited sample of patent lawyers, a relatively short three-month observation window, and the rapid evolution of model capabilities and AI penetration rates, which may cause different results if the experiment were repeated.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1070921API
curl https://kongchang.com/api/v1/knowledge/claims/1070921MCP
get_claim(id=1070921)