David Bader Warns on Agentic AI Access as Real-Time Models Drive Data Center Demand
Updated
Updated · NJIT News | · May 29
David Bader Warns on Agentic AI Access as Real-Time Models Drive Data Center Demand
1 articles · Updated · NJIT News | · May 29
David Bader said agentic AI should get only narrow, reversible permissions, with humans approving high-stakes actions such as payments, posts or calendar changes.
Real-time AI is reshaping computing needs because users expect near-instant answers, forcing data centers to handle bursty, latency-sensitive workloads rather than predictable batch jobs.
Bader said his research targets irregular, data-intensive computations — hard-to-split problems increasingly underlying modern AI — to make advanced computing faster and more accessible.
On data centers, he argued smarter algorithms and scheduling can cut electricity, water use and noise by reducing unnecessary operations and shifting heavy workloads to cleaner, lower-demand hours.
Looking ahead, Bader said AI could let a single researcher tackle problems now requiring teams and specialized hardware, widening access to high-performance computing by hiding infrastructure complexity.
Can 'green' data centers truly offset the massive energy and water costs AI imposes on local communities?
As AI agents act independently, how can society build trust and ensure human control when things go wrong?