Overview
Wiley worked with the Sustainable AI Group (Boris Gamazaychikov and Sasha Luccioni) to estimate the energy used by Landscape View, a playbook in Wiley Applied Research Intelligence, and published the results on 15 October 2026. The estimate compared Landscape View with an agent built for the comparison that used a large language model alone, using token counts and model names supplied by Wiley.
Benefits and outcomes
- Wiley reports, from the Sustainable AI Group’s estimate for one Alzheimer’s disease Landscape View, that the LLM-only approach used about 119 Wh and 81 g CO₂ and processed about 27 million tokens.
- Wiley reports that ARI used about 15 Wh and 10 g CO₂, and about 6,800 tokens a run, after a one-time extraction into its knowledge graph.
- Wiley reports that, with the one-time extraction spread across 10 runs, ARI uses an estimated 88% less generative AI energy than the LLM-only approach, and that the extraction pays for itself after about 1.2 runs.
Notes and quotes
- Wiley says the estimate covers only the generative AI parts of one playbook, based on one run of each approach on one topic, and uses industry proxies for closed models. Wiley says it is not a certification, audit or verification.
Sources
- How Landscape View in Wiley Applied Research Intelligence uses an estimated 88% less generative AI energy than an LLM-only approach, Apoorva Shah, Wiley, 15 October 2026
- Apoorva Shah on the partnership with the Sustainable AI Group, LinkedIn, October 2026
This entry was written with the help of AI, from the sources listed. Please check the sources before relying on it, and let us know if anything needs correcting.
