LLM Novice Uplift on Dual-Use, In Silico Biology Tasks
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866914392354848768 |
|---|---|
| author | Zhang, Chen Bo Calvin Knight, Christina Q. Kruus, Nicholas Hausenloy, Jason Medeiros, Pedro Li, Nathaniel Kim, Aiden Orlovskiy, Yury Breen, Coleman Cai, Bryce Götting, Jasper Liu, Andrew Bo Nedungadi, Samira Rodriguez, Paula He, Yannis Yiming Shaaban, Mohamed Wang, Zifan Donoughe, Seth Michael, Julian |
| author_facet | Zhang, Chen Bo Calvin Knight, Christina Q. Kruus, Nicholas Hausenloy, Jason Medeiros, Pedro Li, Nathaniel Kim, Aiden Orlovskiy, Yury Breen, Coleman Cai, Bryce Götting, Jasper Liu, Andrew Bo Nedungadi, Samira Rodriguez, Paula He, Yannis Yiming Shaaban, Mohamed Wang, Zifan Donoughe, Seth Michael, Julian |
| contents | Large language models (LLMs) perform increasingly well on biology benchmarks, but it remains unclear whether they uplift novice users -- i.e., enable humans to perform better than with internet-only resources. This uncertainty is central to understanding both scientific acceleration and dual-use risk. We conducted a multi-model, multi-benchmark human uplift study comparing novices with LLM access versus internet-only access across eight biosecurity-relevant task sets. Participants worked on complex problems with ample time (up to 13 hours for the most involved tasks). We found that LLM access provided substantial uplift: novices with LLMs were 4.16 times more accurate than controls (95% CI [2.63, 6.87]). On four benchmarks with available expert baselines (internet-only), novices with LLMs outperformed experts on three of them. Perhaps surprisingly, standalone LLMs often exceeded LLM-assisted novices, indicating that users were not eliciting the strongest available contributions from the LLMs. Most participants (89.6%) reported little difficulty obtaining dual-use-relevant information despite safeguards. Overall, LLMs substantially uplift novices on biological tasks previously reserved for trained practitioners, underscoring the need for sustained, interactive uplift evaluations alongside traditional benchmarks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_23329 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | LLM Novice Uplift on Dual-Use, In Silico Biology Tasks Zhang, Chen Bo Calvin Knight, Christina Q. Kruus, Nicholas Hausenloy, Jason Medeiros, Pedro Li, Nathaniel Kim, Aiden Orlovskiy, Yury Breen, Coleman Cai, Bryce Götting, Jasper Liu, Andrew Bo Nedungadi, Samira Rodriguez, Paula He, Yannis Yiming Shaaban, Mohamed Wang, Zifan Donoughe, Seth Michael, Julian Artificial Intelligence Computation and Language Cryptography and Security Computers and Society Human-Computer Interaction Large language models (LLMs) perform increasingly well on biology benchmarks, but it remains unclear whether they uplift novice users -- i.e., enable humans to perform better than with internet-only resources. This uncertainty is central to understanding both scientific acceleration and dual-use risk. We conducted a multi-model, multi-benchmark human uplift study comparing novices with LLM access versus internet-only access across eight biosecurity-relevant task sets. Participants worked on complex problems with ample time (up to 13 hours for the most involved tasks). We found that LLM access provided substantial uplift: novices with LLMs were 4.16 times more accurate than controls (95% CI [2.63, 6.87]). On four benchmarks with available expert baselines (internet-only), novices with LLMs outperformed experts on three of them. Perhaps surprisingly, standalone LLMs often exceeded LLM-assisted novices, indicating that users were not eliciting the strongest available contributions from the LLMs. Most participants (89.6%) reported little difficulty obtaining dual-use-relevant information despite safeguards. Overall, LLMs substantially uplift novices on biological tasks previously reserved for trained practitioners, underscoring the need for sustained, interactive uplift evaluations alongside traditional benchmarks. |
| title | LLM Novice Uplift on Dual-Use, In Silico Biology Tasks |
| topic | Artificial Intelligence Computation and Language Cryptography and Security Computers and Society Human-Computer Interaction |
| url | https://arxiv.org/abs/2602.23329 |