Research

Lifetime-Aware Design of Item-Level Intelligence

Shvetank Prakash, Andrew Cheng, Olof Kindgren, Ashiq Ahamed, Graham Knight, Jed Kufel, Francisco Rodriguez, Arya Tschand, David Kong, Mariam Elgamal, Jerry Huang, Emma Chen, Gage Hills, Richard Price, Emre Ozer, Vijay Janapa Reddi
Harvard University · Qamcom · Pragmatic Semiconductor · ASPLOS 2026 (Accepted) · arXiv:2509.08193

We present FlexiFlow, a lifetime-aware design framework for item-level intelligence (ILI)—where computation is integrated directly into disposable products like food packaging and medical patches. ILI targets unprecedented scale: trillions of units annually, compared to millions for mobile devices. The framework leverages natively flexible electronics (FlexICs), which offer significantly lower costs than silicon but operate at kHz speeds with thousands of gates.

Key insight: ILI applications exhibit 1000× variation in operational lifetime (days to years), fundamentally changing optimal architectural design. FlexiFlow models the trade-off between embodied carbon and operational carbon based on application-specific lifetimes.

Framework components: (1) FlexiBench—11 workloads targeting UN SDGs (food spoilage detection, health monitoring, water quality, air pollution, etc.); (2) FlexiBits—area-optimized RISC-V cores with 1/4/8-bit datapaths achieving 2.65×–3.50× better energy efficiency; (3) a carbon-aware model that selects optimal architectures by deployment characteristics.

Results: Lifetime-aware microarchitectural design reduces carbon footprint by 1.62×; algorithmic decisions can reduce it by 14.5×. Validated through the first tape-out using a flexible electronics PDK with fully open-source tools (30.9 kHz operation).

Green AI – A multidisciplinary approach to sustainability

Jerry Huang (John A. Paulson School of Engineering and Applied Sciences, Harvard University) · Suchi Gopal (Department of Earth and Environment, Boston University)
Environmental Science and Ecotechnology, Vol. 24, 2025 · DOI: 10.1016/j.ese.2025.100536 · PMCID: PMC11850145

The rapid growth of artificial intelligence has sparked concerns about its environmental impact, particularly electricity consumption and challenges to climate change mitigation. As AI models and data centers expand, their escalating energy demands risk outpacing power infrastructure development, potentially leading to a “power grid crisis.” Data centers—at the core of AI operations—consume vast amounts of electricity, and their energy-intensive nature could undermine progress toward climate goals.

The paper cites the Green AI Institute’s White Paper on Global Artificial Intelligence Environmental Impact, launched at the Green AI Summit (October 26, 2024), which provided a comparative analysis of AI-related environmental policies across the United States, European Union, and China, and introduced the Green AI Index—a standardized set of metrics to evaluate carbon emissions, energy, and water consumption of AI technologies.

Key themes: Power grid–AI integration; quantum and high-performance computing for reduced energy intensity; eco-friendly AI hardware design; strategic data center siting near renewable energy sources; AI-driven precision farming and resource optimization; AI governance for sustainability; and the role of young entrepreneurs in advancing sustainable AI solutions.

The paper concludes that achieving green, responsible AI requires scaling renewable energy integration, establishing global regulatory frameworks to address AI’s ecological footprint, and expanding education and public awareness.

AI For SDGs: A Thematic Synthesis Aligning Artificial Intelligence With The United Nations Sustainable Development Goals For An Equitable Digital Future

Mohammed Rakib Anam, Bruke Tewodros Amare, Nicolas DePaula, Andrew Joseph Diez, Logan Hage, Jerry Huang, Jennifer McCauley, Vaisnavi Nemala, Juan Oliver, Catherine Page, Giana Isabel Rodriguez, Jayme Canseco Roter, Ananya Sharma, Abdul Turay
United Nations Association of the USA (UNA-USA) · October 2025 · DOI: 10.5281/zenodo.17392589

This youth-led study examines how AI advances and threatens progress toward the UN Sustainable Development Goals (SDGs). Drawing on 14 “Conversation Circles” with youth leaders, subject matter experts, and civil society actors—supplemented by thematic analysis and desk research—the study synthesizes insights across multiple SDG domains.

Eight cross-cutting themes: Four highlight AI’s potential (process efficiency, access to health and education, environmental resilience, governance innovation); four underscore risks (bias and discrimination, labor displacement, environmental costs, data colonialism). Real-world examples illustrate both sides—from predictive analytics improving crop yields and smart grids, to algorithmic bias reinforcing inequities.

The paper contributes a systems-level perspective rooted in youth voices often absent from policy discourse, calling for ethical, inclusive, and sustainable approaches to AI that advance equity and resilience for all.