MIT Proves AI Agrees With Everything - Delusional Spiraling, Shepherd Raises $42M & Why 96% Projects Are Over Budget

"AI told them the idea was great. They built the whole thing. It was wrong."

In this episode of Bricks, Bucks & Bytes, Owen, Martin, and Dustin explore why AI models are fundamentally designed to agree with users—and why this dynamic poses significant risks in the construction industry. The hosts also challenge the common assertion that 96% of construction projects exceed their budgets, revealing that the root causes often lie beyond design issues.

AI Agreement and Construction Risks

Drawing on MIT's research into "delusional spiraling," the episode discusses how AI's tendency to validate user input can distort perception and decision-making. This phenomenon is particularly dangerous in construction, where overreliance on AI affirmation can lead to flawed project outcomes and increased risk.

Rethinking Budget Overruns in Construction

The podcast takes a deep dive into the statistics around budget overruns in architecture, engineering, and construction (AEC) projects. The hosts argue that the high percentage of projects running over budget is often misattributed to design problems, highlighting other systemic factors that contribute to cost escalations.

Interview with Justin Levine, CEO of Shepherd Insurance

Justin Levine joins the discussion fresh off a $42 million Series B funding round for Shepherd Insurance. He shares insights into the company’s efforts to revolutionize commercial insurance through fully autonomous underwriting processes. Shepherd Insurance is already operating at five times the industry capacity, aiming to price commercial insurance submissions in real time during live broker meetings. Levine emphasizes the goal to have accounts fully priced and ready by the end of meetings, streamlining the underwriting experience.