Three Takes on AI

In episode 3, hosts Brian, Michael and Campbell dig into the uncomfortable gap between AI’s glossy headline numbers and the risk‑adjusted reality leaders actually live with. Using fresh data on global AI investment, enterprise spend and reported 3.7x returns, they unpack what happens when 40%+ of projects deliver zero ROI, most initiatives die in pilot purgatory, and “3–6 month wins” quietly stretch to 12–18 months. The trio walks through true costs versus vendor quotes, why 40–50% productivity claims often net out closer to 5%, and how to manage AI like a portfolio where one winner has to pay for several failures. If you are being asked to “do something with AI” and justify the business case, this conversation will help you reset expectations, risk‑adjust your forecasts, and design AI bets that actually have a chance of paying off.

Three Takes on AI returns with a fiery Episode 2, where three operators with very different convictions tackle one big question: are we actually in an AI bubble, or are we just getting started?

In this episode, Michael Muhlfelder argues the market is deep in bubble territory, fueled by speculative valuations, financial engineering around GPUs, and an explosion of “AI companies” with thin evidence of real productivity gains.

Brian Silverman takes the other side, making the case that we are still in the “browser moment” of AI—early in the adoption curve, with massive infrastructure still being built and the true AI economy yet to arrive.

Campbell Robertson lands in the messy middle, suggesting this boom may not end with a dramatic pop at all, but instead fizzle, morph, or shift because of unprecedented capital flowing into real-world infrastructure, power, and data centers.

For business and technology leaders, the debate quickly gets practical:

• What signals actually distinguish a durable AI boom from a classic tech bubble?​

• How should you think about all the GPU spending, cloud build-out, and vendor noise when you are making your own AI investment decisions?

• And if this is a bubble, what is your downside protection and exit strategy—while still capturing upside if the AI economy keeps compounding?

If you care about AI strategy, capital planning, and avoiding expensive hype cycles, you will want to hear where each of the three lands—and where they reluctantly agree. Listen to Episode 2 of Three Takes on AI: “The AI Bubble — Burst, Fizzle, or Early Days?” at ThreeTakesOnAI.com.

The 95% Failure Rate Nobody Understands: MIT’s AI Study & The Shadow AI Revolution

Everyone’s freaking out about the MIT study claiming 95% of AI projects fail—but did anyone actually read it? We did. And the real story is way more interesting than the clickbait headlines.

In this debut episode, we dissect what the study actually says (spoiler: the 5% that succeed are crushing it), why traditional ROI metrics are broken for AI, and the fascinating phenomenon of “Shadow AI”—employees getting real value from ChatGPT and other tools while official projects struggle.

What You’ll Learn:

• Why most AI projects are being judged too early

• The governance nightmare of attorneys putting client data in public AI models

• How back-office AI succeeds while front-office implementations struggle

• Real examples of AI productivity gains that traditional metrics can’t measure

• Why change management matters more than the technology itself

Perfect for: Business leaders questioning their AI strategy, tech implementers looking for honest insights, and anyone tired of AI hype.

Resources & links: Visit threetakesonai.com for the full MIT study, our analysis, and to share your own AI experiences.