MIRAGE: The Phantom Web and Its Real Footprint

AI is adding computing, storage, and internet traffic, and a growing share of it comes from bots: accounts that post on Instagram, Snapchat, LinkedIn, and Facebook, and AI bots that drive other bots. Professor Swarup Bhunia, in UF’s Department of Electrical and Computer Engineering, asked how large that footprint is in energy, cooling water, and carbon, and how it compares with AI-powered search and the AI services people use directly.

Write a perspective paper under his supervision, with Habibur Rahaman as my mentor, in two parts: U.S. AI data-center growth and its resource demand, then bot activity and its footprint, compared with AI search and human-initiated use of services like ChatGPT, Claude, and Gemini. One question to test, if the evidence holds: can bots outweigh both?

• Review the literature and collect public data on data-center energy, cooling water, and emissions, and on how bot activity and AI search are changing.
• Keep measured figures, estimates, scenarios, forecasts, and company-reported numbers apart.
• Document each comparison’s system boundary, units, geography, workload assumptions, and uncertainty.
• Count each task once: a search request, its agent tool calls, retrieval, and model inference are parts of one task, not separate workloads.
• Tell apart traditional automation, content-generating AI bots, social-media account bots, crawlers, retrieval agents, and multi-agent workflows.
• Write the paper in LaTeX on Overleaf, with plots from the data.

• Pending: no results or data yet.