Wu et al. 2026 argue that the barriers to agentic AI in biomedical laboratories are organizational rather than model-centric—technical fluency, standard operating procedures and data-governance rules that more capable models alone will not remove—and they built the open-source Murmurent layer beneath the agents to supply them. Nelson et al. 2026's interview study reaches the same conclusion from the innovation side: AI's benefits depend on integration within a robust ecosystem of conventional experimentation.
The Wu et al. portion is fully supported (barriers unlikely to be removed by more capable models alone, fluency/SOPs/governance rules, open-source Murmurent beneath the agents), but the cited Nelson et al. text never indicates the study is interview-based, and it states that sustaining innovation facilitated by AI—not AI's benefits generally—depends on integration within a robust ecosystem of conventional experimentation.
Written by Kimi K3 via Ollama Cloud · checked by GLM-5.3 via Ollama Cloud · 16 Oct, 04:35
Source chain
AEvery quote below was checked, without a model, to appear verbatim in its source.
- 01
“Significant barriers remain, however, which are unlikely to be removed solely by releasing more capable AI models.”
Abstract · no page number · evidence E23
- 02
“We present Murmurent, shared software that sits beneath the agentic AI and provides the skills, procedures and protections to address the concerns above”
Abstract · no page number · evidence E23
- 03
“even sustaining innovation facilitated by AI depends upon integration within a robust ecosystem of conventional experimentation, data collection, analysis, and theory development”
Full-text passage · p. 2 · evidence E56
The answer to both questions is a qualified “yes.” Our results deepen and refine prior suggestions that AI may help to speed up science, but may raise epistemic risks, by clarifying the role of AI -enabled science in processes of technological change; identifying mechanisms and boundary conditions underlyin g both affordances and downsides of AI; and drawing out implications of these mechanisms for research strategy and policy. Our key, novel findings are as follows: 1) AI’s contributions to technological progress come by extending prior computational utilit ies in reducing the “experiment intensity” of search through technological design spaces; 2) this utility is likely to facilitate sustaining innovation within existing “technological paradigms” (Dosi, 1982; Dosi, 2007; Nelson, 2008b) , but may undercut disruptive innovation; and 3) even sustaining innovation facilitated by AI depends upon integration within a robust ecosystem of conventional experimentation, data collection, analysis, and theory development. Consequently, at least thus far, AI/ML is more another tool in the research and development toolkit than a radical transformation of the innovation process.
Passage read from arxiv.org, which may be a preprint rather than the published version.