The Prof G Pod with Scott Galloway · February 11, 2026 · 50m

Why CEOs Are Getting AI Wrong — with Ethan Mollick

Wharton's Ethan Mollick presents hard data on AI adoption: BCG randomized controlled trials show 40% quality improvement and 26% faster task completion with GPT-4, even without training. But workers hide AI use from employers — creating a massive gap between actual adoption and corporate visibility. Mollick argues most organizations lack the imagination to redesign work around AI. Covers the 'jagged frontier' of AI capabilities, open-weight models, and implications for education and medicine.

Highlights

•

AI produces 40% quality improvement in RCTs — workers hide it

BCG RCTs: 40% quality gain, 26% faster. Workers hide AI use from employers, creating a massive adoption visibility gap.

Claude ChatGPT Gemini

•

The jagged frontier of AI capabilities

AI is brilliant at some tasks and terrible at adjacent ones. Organizations can't predict which without testing.

Claude ChatGPT Gemini

References

■

Co-Intelligence — Ethan Mollick (2024) — Living and working with AI

Misc

✧BCG RCT: 40% quality improvement, 26% faster, even without training

✧Workers report 3x productivity on specific tasks but hide it from employers

✧The 'jagged frontier': AI is brilliant at some tasks and terrible at adjacent ones

✧'Most organizations lack the imagination to redesign themselves around new technology'