AI Singularities: Are We Already There? - Mark Brady - The Agentic Insider - Episode #14

14 snips

Jul 10, 2025

Join Mark Brady, the Deputy Chief Data Officer at KBR Inc. and former Chief Data Officer of the US Space Force, as he unpacks the dual singularities of AI that could reshape humanity. He explores the critical need for safety objective functions in AI systems and the risks posed by autonomous AIs creating other AIs. The conversation shifts to the future of AI and human coexistence while diving into its applications in defense. Brady also shares insights from his own journey and career tips for aspiring data scientists.

Ask episode

AI Snips

Chapters

Books

Transcript

Episode notes

INSIGHT

Asimov's Laws as Safety Functions

Asimov's three laws of robotics function as safety objective functions to prevent robots from harming humans.
These laws provide motivation or objectives for AI behavior linked to safety instead of unrestricted goals.

ADVICE

Prioritize Safety Objective Functions

Embed safety objective functions that override other objectives in AI systems.
Avoid AI objectives that promote replication or resource maximization to prevent competition with humans.

ADVICE

Prepare for Existential AI Risks

Organizations should prepare for AI risks by embedding safety objectives in systems.
Focus less on bias concerns and more on existential risks like safety overrides in AI.

Get the Snipd Podcast app to discover more snips from this episode

Get the app

What if we've already lost control? Dr. Mark Brady, Deputy Chief Data Officer of TRMC at KBR Inc., unveils the two AI singularities that could redefine humanity's future. The episode dives into the significance of safe AI, exploring the six laws of safe AI and various facets of AI safety, such as the importance of safety objective functions and the challenges posed by AI evolution and machine learning. The conversation also touches on the applications of AgTech AI in defense and government, and the current capabilities and limitations of autonomous AI agents. To close, Mark shares personal hobbies and offers career advice for aspiring data scientists and future chief data officers.

Takeaways:

Understanding AI Singularities: Familiarize yourself with the concept of AI singularities. Recognize the symptoms and implications of these singularities on society and technology.
Safety Objective Functions: Implement safety objective functions in AI systems to ensure they prioritize human safety over other objectives.
Tracking Data Lineage: Recognize the challenges in data lineage and strive to improve transparency and understanding of how data flows through systems.
AI Creating AI: Ensure that any AI creating other AI systems includes its own safety objective functions in its creations. This prevents the propagation of unsafe AI systems.
Human and AI Coexistence: Consider the potential future coexistence of different intelligences, such as biological, machine, augmented biological, and biological augmented machine, and how they might interact and compete.
Large Language Models: Keep abreast of the advancements in large language models and their applications. Understand their current limitations and potential for future development.

Quote of the Show:

“If AI is smarter than we are, how are we gonna be able to compete with it? How are we going to be able to make sure something's not going awry? We need AI to help us.” - Mark Brady

Links:

LinkedIn: https://www.linkedin.com/in/mark-brady-18274321b/
Website: https://www.kbr.com/en
Book Link: https://a.co/d/gtODJjR

Ways to Tune In:

Transistor: https://agenticinsider.show/
Spotify: https://open.spotify.com/show/2fCCuVl6tL5p2kEAwFCQNJ
Apple Podcasts: https://podcasts.apple.com/us/podcast/the-agentic-insider/id1807346381
Amazon Music: https://music.amazon.com/podcasts/9fe0482d-d798-4420-a93a-fc6653bce437
Podchaser: https://www.podchaser.com/podcasts/the-agentic-insider-6057241
YouTube: https://www.youtube.com/@iridiusAI