Why is prompting hard? Understanding prompts on binary sequence predictors

Li Kevin Wenliang, Anian Ruoss, Jordi Grau-Moya, Marcus Hutter, Tim Genewein
Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, PMLR 300:2134-2142, 2026.

Abstract

Frontier models can be prompted or conditioned to do many tasks, but finding good prompts is not always easy, nor is understanding some performant prompts. We view prompting as finding the best conditioning sequence on a near-optimal sequence predictor. On numerous well-controlled experiments, we show that unintuitive optimal conditioning sequences can be better understood given the pretraining distribution, which is not usually available. Even using exhaustive search, reliably identifying optimal prompts for practical neural predictors can be surprisingly difficult. Popular prompting methods, such as using demonstrations from the targeted task, can be surprisingly suboptimal. Using the same empirical framework, we analyze optimal prompts on frontier models, revealing patterns similar to the binary examples and previous findings. Taken together, this work takes an initial step towards understanding optimal prompts, from a statistical and empirical perspective that complements research on frontier models.

Cite this Paper

BibTeX
@InProceedings{pmlr-v300-wenliang26a, title = { Why is prompting hard? Understanding prompts on binary sequence predictors }, author = {Wenliang, Li Kevin and Ruoss, Anian and Grau-Moya, Jordi and Hutter, Marcus and Genewein, Tim}, booktitle = {Proceedings of The 29th International Conference on Artificial Intelligence and Statistics}, pages = {2134--2142}, year = {2026}, editor = {Khan, Emtiyaz and Li, Yingzhen and Solin, Arno and Ramdas, Aaditya}, volume = {300}, series = {Proceedings of Machine Learning Research}, month = {02--05 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v300/main/assets/wenliang26a/wenliang26a.pdf}, url = {https://proceedings.mlr.press/v300/wenliang26a.html}, abstract = { Frontier models can be prompted or conditioned to do many tasks, but finding good prompts is not always easy, nor is understanding some performant prompts. We view prompting as finding the best conditioning sequence on a near-optimal sequence predictor. On numerous well-controlled experiments, we show that unintuitive optimal conditioning sequences can be better understood given the pretraining distribution, which is not usually available. Even using exhaustive search, reliably identifying optimal prompts for practical neural predictors can be surprisingly difficult. Popular prompting methods, such as using demonstrations from the targeted task, can be surprisingly suboptimal. Using the same empirical framework, we analyze optimal prompts on frontier models, revealing patterns similar to the binary examples and previous findings. Taken together, this work takes an initial step towards understanding optimal prompts, from a statistical and empirical perspective that complements research on frontier models. } }
Endnote
%0 Conference Paper %T Why is prompting hard? Understanding prompts on binary sequence predictors %A Li Kevin Wenliang %A Anian Ruoss %A Jordi Grau-Moya %A Marcus Hutter %A Tim Genewein %B Proceedings of The 29th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2026 %E Emtiyaz Khan %E Yingzhen Li %E Arno Solin %E Aaditya Ramdas %F pmlr-v300-wenliang26a %I PMLR %P 2134--2142 %U https://proceedings.mlr.press/v300/wenliang26a.html %V 300 %X Frontier models can be prompted or conditioned to do many tasks, but finding good prompts is not always easy, nor is understanding some performant prompts. We view prompting as finding the best conditioning sequence on a near-optimal sequence predictor. On numerous well-controlled experiments, we show that unintuitive optimal conditioning sequences can be better understood given the pretraining distribution, which is not usually available. Even using exhaustive search, reliably identifying optimal prompts for practical neural predictors can be surprisingly difficult. Popular prompting methods, such as using demonstrations from the targeted task, can be surprisingly suboptimal. Using the same empirical framework, we analyze optimal prompts on frontier models, revealing patterns similar to the binary examples and previous findings. Taken together, this work takes an initial step towards understanding optimal prompts, from a statistical and empirical perspective that complements research on frontier models.
APA
Wenliang, L.K., Ruoss, A., Grau-Moya, J., Hutter, M. & Genewein, T.. (2026). Why is prompting hard? Understanding prompts on binary sequence predictors . Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 300:2134-2142 Available from https://proceedings.mlr.press/v300/wenliang26a.html.

Related Material