What happens when guardrails aren't enough?

On this episode of Ground Control, host Perry sits down with Dr. Craig Kaplan, cognitive psychologist, AI researcher, and founder of SuperIntelligence.com, for a wide-ranging conversation on why safety has to be built into AI's architecture, not bolted on after the fact.

Craig's path into AI started as an undergraduate studying psychobiology, which led him to Carnegie Mellon, where he worked alongside Nobel laureate Herbert Simon, one of the field's founding figures. From there he founded IQ Company in 1993 and, later, Predict Wall Street, a crowdsourced hedge fund built on a simple but radical bet: that the aggregated opinions of millions of everyday retail investors could outperform the best-funded quant researchers on Wall Street. It worked. The fund ranked in the top 10 of its class before Craig sold the company in 2020, having proven what he calls collective intelligence in the most competitive arena imaginable.

That experience became the foundation for how Craig thinks about AI safety today. He argues that the dominant approach to building large language models, pouring in more data and compute to create ever-larger "black box" systems, is fundamentally the wrong design. Guardrails added after training are a losing game of whack-a-mole: easy to jailbreak, impossible to fully patch. Instead, Craig makes the case for something closer to a democracy than a dictator: architectures built from many smaller, checkable intelligences that create transparency, checks and balances, and distributed power, the same qualities that make human institutions resilient to bad actors.

Perry and Craig also dig into what this means practically for businesses adopting AI: why companies should never rely on a single model for mission-critical decisions, how to make sure a company's own values and ethics get explicitly built into the systems it deploys, and what obligations companies have to be transparent with employees and customers about where their AI comes from and how it's being used. The conversation closes with a rapid-fire round covering Craig's revised AGI timeline, what Herbert Simon might make of his work today, and why Craig believes the data shows humanity's track record is better than the doomer narrative suggests, if we can just get AI to match our values.

Guest: Dr. Craig Kaplan, LinkedIn | SuperIntelligence.com

0:00 Introduction: Meet Dr. Craig Kaplan
0:50 From Psychobiology to AI: Studying Under Nobel Laureate Herbert Simon
3:06 Founding IQ Company in 1993
4:06 The Three Eras of AI: Symbolic, Machine Learning, and Reasoning
7:58 Quantifying the Pace of AI Progress
10:15 Predict Wall Street: Proving Collective Intelligence in the Markets
16:22 Why AI Guardrails Fail: An Ounce of Prevention
23:07 Designing Safer AI: Black Boxes vs. Democratic Architecture
29:13 Sovereign Intelligence and Owning Your Data
34:36 Bounded Rationality and the Case for Mixture of Experts AI
40:53 What Do Companies Owe Users About Their AI?
49:27 Rapid Fire: Predictions, Regrets, and Final Thoughts

What does it mean to be "suffocated by success" and how do you break free? In this episode of Ground Control, Perry Robinson sits down with Jonathan Maddock, a systems engineer turned program manager turned attorney with nearly 30 years inside the Department of Defense. Jonathan brings a rare perspective forged at the intersection of military leadership, government bureaucracy, and commercial enterprise, where competing priorities are the norm and alignment is the mission.

Jonathan shares the thinking behind his book Suffocated by Success, exploring how founders and leaders unknowingly become the bottleneck in their own organizations. Drawing on his engineering and legal backgrounds, he breaks down why every business, no matter the size or industry, is a system of systems, and what happens when leaders fail to treat it that way.

Perry Robinson and Jonathan dig into founder ego, the temptation to jump in and solve every problem, and why the fastest path to sustainable growth often requires slowing down first. Jonathan uses a simple but powerful analogy: letting your kids struggle through the wrong answer on their homework so they actually learn how to find their way back. The same principle applies to building a resilient team and a scalable business.

The conversation also covers the role of AI in managing complex systems, how RocketDocs and businesses like it can benefit from systems thinking, how to distinguish between self-imposed urgency and real constraints, and why patience is not the opposite of speed but the foundation of it.

Jonathan closes with a message that goes beyond business strategy and operational efficiency: a reminder that each of us is our own system of mind, body, and soul, and that mental health awareness, including being kind to yourself, is part of operating at your best.

Chapters:

00:00 Introduction to Jonathan Maddock's Background
05:32 The Concept of Being 'Suffocated by Success'
11:12 Understanding Systems in Business
16:38 The Role of Founders and Ego in Business
22:17 Integrating Engineering and Legal Systems
23:35 Understanding Legal Systems Through Engineering Principles
26:22 Breaking Down Complex Problems into Manageable Components
28:55 The Role of AI in System Management
32:38 Navigating Fragility in Business Systems
35:13 The Importance of Patience in System Development
36:35 Learning Through Challenges: The Founder’s Journey
39:27 Balancing Speed and Systematic Growth
43:35 Mental Health Awareness in System Management

Connect with Jonathan on LinkedIn or visit fortivianstrata.com to learn more about his work and pick up a copy of the book.

Ground Control is sponsored by RocketDocs, the response management platform built for regulated industries and trusted since 1994. RocketDocs helps teams in financial services, healthcare, life sciences, and enterprise tech streamline RFP responses, DDQ automation, security questionnaires, and sales proposals using private AI. RocketDocs is designed to help teams win more business faster without sacrificing compliance or quality. Learn more at rocketdocs.com.

Breaking Into Government Contracts: What You Need to Know Before You Start

Selling to the government sounds like a dream. Steady clients, long-term relationships, and reliable revenue. But the path to becoming a government-approved vendor is anything but simple. In this episode, Bryan and Perry sit down with Lori Crooks, founder and CEO of Cadra, a compliance consulting firm helping small and mid-sized businesses navigate the complex world of government contracting.

Lori breaks down the alphabet soup of compliance frameworks (FedRAMP, GovRAMP, CMMC, NIST) and explains what they actually mean for your business. She shares why starting with compliance early is the single most important thing a company can do, how FedRAMP authorization can open doors far beyond federal agencies, and what continuous monitoring really looks like once you're approved.

Whether you're a small business curious about government contracts or already knee-deep in the process, this conversation is packed with honest, practical guidance on what it takes to get in and stay in the government contracting space.

In this episode:

• The difference between FedRAMP, GovRAMP, and CMMC
• Where to start if you're new to compliance frameworks
• Why policies and procedures matter more than most companies realize
• How AI is reshaping both cybersecurity threats and compliance tools
• The hidden business benefits of compliance beyond winning contracts
  

Connect with Lori: cadra.com | lori.crooks@cadra.com | LinkedIn: Lori Crooks

00:00 Introduction to Government Contracting
02:28 Navigating Compliance Frameworks
05:12 Understanding FedRAMP and GovRAMP
08:02 Starting Your Journey in Government Contracts
10:48 The Importance of Early Engagement
13:31 Maintaining Compliance and Continuous Monitoring
16:10 Automation in Compliance Processes
18:50 Organizing Information for Audits
21:39 Utilizing Tools for Compliance Management
22:55 Navigating Compliance Standards
24:06 Assessing Readiness for Compliance
25:32 The Importance of Comprehensive Policies
27:41 The Role of Documentation in Compliance
29:44 How to Reach Out for Compliance Help
32:06 Understanding the Rewards of Compliance
35:42 The Impact of AI on Compliance Standards
38:36 Future Changes in Compliance Landscape
40:58 Starting with Compliance Early