ROUNDTABLE DISCUSSION: MIT REPORT: 95% OF GENERATIVE AI PILOTS AT COMPANIES ARE FAILING

A recent MIT report reveals that 95% of Generative AI pilots fail to deliver measurable ROI, often because they are treated as isolated experiments rather than integrated into business workflows. The study identifies a "GenAI Divide" where success is found in specific back-office tasks, while flashy, customer-facing projects fail.

Topics of discussion will include, yet will not be limited to:

• The "Boring" ROI Paradox -  ~50% of budgets go to low-ROI, flashy pilots, while the 5% success stories are in "boring" back-office automation
• The "Learning Gap" - Internal builds fail twice as often as vendor-led solutions, despite vendors having less business knowledge
• The "Data Swamp" - Enterprise data is often too messy or siloed for effective use, leading to inaccurate AI outputs
• The "Compliance Cliff" - Pilots fail in production because organizations apply legacy governance instead of specialized AI Trust, Risk, and Security Management strategies
• Change Management - Without proper change management, employees abandon tools due to a "Trust Gap" or fear of replacement

Marty Samadder, Head of Enterprise Architecture, NOVELIS

Adam Smykowski, Director of Agentic AI & Automations Solutions, VALLEY BANK

Karthik Josyula, Head of Data & AI Platforms - Data & Analytics, KOHLER

Poornakala Sethuraman, Ph.D., Vice President, Data Science & Product Strategy, BANK OF AMERICA