Multiple 2026 perspectives are converging on staged autonomy taxonomies as the way to measure agentic-science maturity: Zhao et al. 2026 introduce a 6-level "Spectrum of Agentic Autonomy" for drug discovery, while Hebenstreit et al. 2026 synthesize four prior frameworks into a biomedical framework distinguishing cognitive from physical capabilities.
E23 explicitly introduces a 6-level 'Spectrum of Agentic Autonomy' to evaluate system maturity in drug discovery, and E59 describes synthesizing four prior (level-based) frameworks into a domain-specific biomedical framework distinguishing cognitive from physical capabilities, supporting the convergence framing.
Written by Kimi K3 via Ollama Cloud · checked by GLM-5.3 via Ollama Cloud · 15 Oct, 04:34
Source chain
AEvery quote below was checked, without a model, to appear verbatim in its source.
- 01
“To evaluate system maturity, we introduce a 6-level "Spectrum of Agentic Autonomy."”
Abstract · no page number · evidence E23
- 02
“We synthesized concepts from these frameworks (i.e., general reasoning from AGI models and autonomy levels from engineering standards) to create a domain-specific framework for biomedical research that distinguishes between cognitive and physical capabilities”
Full-text passage · no page number · evidence E59
We identified and reviewed four relevant frameworks: DeepMind’s “Levels of AGI,” which differentiates between narrow and broad AI systems35; Society of Automotive Engineers’ “Driving automation systems,” which describes levels of interaction between humans and increasingly autonomous vehicles36; “AI agents for biomedical discovery,” which details increasing levels of AI agency in biomedical research37; and “Self-Driving Laboratories,” which maps the integration of software and hardware processes for increasing laboratory autonomy38. We synthesized concepts from these frameworks (i.e., general reasoning from AGI models and autonomy levels from engineering standards) to create a domain-specific framework for biomedical research that distinguishes between cognitive and physical capabilities (further described in the Results section). This integration was necessary because biomedical research uniquely requires high proficiency in both abstract reasoning and physical manipulation, which was not fully captured by any single prior framework. Research process mapping: We systematically analyzed the biomedical research lifecycle to identify major research tasks and their subcomponents.
Passage read from www.ebi.ac.uk, which may be a preprint rather than the published version.