Because small, cheap, fast models are sufficient for much of the detection work, you don't need to judiciously deploy one expensive model and hope it looks in the right places. You can deploy cheap models broadly, scanning everything, and compensate for lower per-token intelligence with sheer coverage and lower cost-per-token.
这一观点提出了AI安全的经济新模式，通过广泛部署小型廉价模型来弥补单一大模型的不足。这种'广撒网'策略可能比依赖少数昂贵模型更有效，尤其在大规模代码库扫描场景中，为AI安全的经济可行性提供了新思路。
since reasoning models and agentic AI can rack up quite a bill
文章提醒了一个常被忽视的约束条件：AI的使用成本。在讨论AI替代人类时，人们往往默认AI是低成本方案，但推理模型和智能体的高昂算力成本意味着，仅凭能力覆盖并不等于经济上的可行替代，成本收益分析仍是决定性门槛。
合計数百回、時には数千回に及ぶLLM呼び出しの中で、有望な仮説をさらに深掘りするのか、まったく新しい角度に広げるかを、Sakana Marlinはその都度判断しながら探索します。
数百到数千次 LLM 调用完成一次研究任务——这个规模令人震惊。一个用户提交一个研究主题，背后触发的是数千次 AI 推理调用，形成一棵庞大的假设探索树。从成本角度看，如果每次 LLM 调用均价 0.1 美元，1000 次调用就是 100 美元的计算成本。「数周人力工作」的价值与「100 美元计算成本」之间的鸿沟，正是 AI 替代知识工作的核心经济逻辑所在。
The global annual market value of animalpollinated crops is estimated between USD 235–577 billion(OECD 2019)
for - stats - global annual economic cost of insect pollinators - 235 to 577 billion USD - OECD 2019
for - planetary emergency - economic cost of nature - from an insurance perspective - natural capital valuation - from insurance industry perspective - biodiversity - natural capital valuation - from insurance industry perspective - Swiss RE - Biodiversity and Ecosystem Services (BES) metric - from insurance industry perspective
James, N., & Menzies, M. (2020). Human and financial cost of COVID-19. ArXiv:2009.11660 [Physics, q-Fin]. http://arxiv.org/abs/2009.11660
Rodela, T. T., Tasnim, S., Mazumder, H., Faizah, F., Sultana, A., & Hossain, M. M. (2020). Economic Impacts of Coronavirus Disease (COVID-19) in Developing Countries [Preprint]. SocArXiv. https://doi.org/10.31235/osf.io/wygpk
Kozlowski, J., Veldkamp, L., & Venkateswaran, V. (2020). Scarring Body and Mind: The Long-Term Belief-Scarring Effects of COVID-19 (Working Paper No. 27439; Working Paper Series). National Bureau of Economic Research. https://doi.org/10.3386/w27439
Argente, D. O., Hsieh, C.-T., & Lee, M. (2020). The Cost of Privacy: Welfare Effects of the Disclosure of COVID-19 Cases (Working Paper No. 27220; Working Paper Series). National Bureau of Economic Research. https://doi.org/10.3386/w27220
The Cost of the COVID-19 Crisis: Lockdowns, Macroeconomic Expectations, and Consumer Spending. COVID-19 and the Labor Market. (n.d.). IZA – Institute of Labor Economics. Retrieved August 7, 2020, from https://covid-19.iza.org/publications/dp13224/
Modelling the Distributional Impact of the COVID-19 Crisis. COVID-19 and the Labor Market. (n.d.). IZA – Institute of Labor Economics. Retrieved August 5, 2020, from https://covid-19.iza.org/publications/dp13235/
Deryugina, T., Gruber, J., & Sabety, A. (2020). Natural Disasters and Elective Medical Services: How Big is the Bounce-Back? (Working Paper No. 27505; Working Paper Series). National Bureau of Economic Research. https://doi.org/10.3386/w27505
Rampini, A. A. (2020). Sequential Lifting of COVID-19 Interventions with Population Heterogeneity (Working Paper No. 27063; Working Paper Series). National Bureau of Economic Research. https://doi.org/10.3386/w27063
He, Z., Nagel, S., & Song, Z. (2020). Treasury Inconvenience Yields during the COVID-19 Crisis (Working Paper No. 27416; Working Paper Series). National Bureau of Economic Research. https://doi.org/10.3386/w27416
Slobin, S. (n.d.). How Remote Work Divides America. Reuters. Retrieved July 27, 2020, from https://graphics.reuters.com/HEALTH-CORONAVIRUS/USA-REMOTEWORK/xlbpgbrljvq/
Kundu, B., & Bhowmik, D. (2020). Societal impact of novel corona virus (COVID ̶ 19 pandemic) in India [Preprint]. SocArXiv. https://doi.org/10.31235/osf.io/vm5rz
halperin, daniel. (2020). The Covid-19 Lockdown “Natural Experiment” That Has Already Been Conducted [Preprint]. SocArXiv. https://doi.org/10.31235/osf.io/jzhe2
Hadavand, A., Hamermesh, D.S., & Wilson, W.W. (2020). Is scholarly refereeing productive (at the margin)? The National Bureau of Economic Research. https://www.nber.org/papers/w26614