Lex Fridman PodcastKeoki Jackson: Lockheed Martin | Lex Fridman Podcast #33
CHAPTERS
- 0:00 – 4:44
Engineering awe: Skunk Works aircraft and early space milestones
Lex opens by framing Lockheed Martin’s legacy of extreme engineering, then asks Keoki what projects have inspired him most. Keoki highlights iconic aircraft (P‑80, F‑104, SR‑71) and pivots to his personal passion for space missions.
- •Skunk Works as a symbol of ambitious, fast-turn engineering
- •F‑104 Starfighter as an early Mach 2 milestone; SR‑71 as ISR speed/altitude icon
- •Keoki’s path to Lockheed via love of space and MIT
- •Lockheed’s deep involvement in planetary exploration, including Mars missions
- 4:44 – 6:00
Humanity to Mars: Orion, deep-space travel, and “Mars Base Camp”
Lex asks about the dream of humans stepping onto Mars and beyond. Keoki describes Lockheed’s role through Orion and a staged approach: Moon first, then Mars with architectures that support human missions and eventual longer-term presence.
- •Personal and organizational motivation for human deep-space exploration
- •Orion as a human-rated spacecraft for deep-space journeys
- •Mars “base camp” concept and mission flow (orbit to surface and back)
- •Colonization as further out, but enabled by early sustained missions
- 6:00 – 9:40
Sustained presence and a space economy: Moon base, cislunar infrastructure, and launch costs
The conversation turns to long-term strategy: not just visiting space, but staying and building an economy. Keoki outlines near-term cislunar infrastructure, the importance of lowering launch costs, and the engineering realities of maintaining a permanent presence.
- •Goal: sustained, sustainable presence beyond LEO with a space economy
- •Cislunar base enabling regular transfer to lunar surface and back
- •Launch access improvements: cost, flexibility, speed, reliability
- •Key challenge is ‘go vs stay’: habitation, reuse, supply chains, infrastructure
- 9:40 – 13:13
Robotic exploration at scale: OSIRIS‑REx and sampling an asteroid
Keoki emphasizes how robotics and autonomy have transformed exploration, enabling ‘virtual presence’ across the solar system. He uses the OSIRIS‑REx mission to asteroid Bennu to illustrate autonomous mapping and an ingenious sample-collection mechanism.
- •Rovers/satellites enable continuous Mars operations despite time lag
- •AI and processing expand what autonomous spacecraft can do
- •OSIRIS‑REx year-long mapping of Bennu and autonomous navigation
- •‘Pogo stick’ short-touch sampling: gas-driven collection into a hopper
- 13:13 – 15:08
Why humans still matter in space: adaptation, anomalies, and real-time decision making
Lex presses on the role of humans versus robots. Keoki argues humans remain uniquely strong at adapting to new information and handling anomalies, citing Apollo 13 as an example of ingenuity under constraints.
- •Human advantage: rapid adaptation and improvisation under uncertainty
- •Communication time-lag makes local decisions valuable on Mars missions
- •Apollo 13 as a canonical ‘adaptation under failure’ case
- •Best near-term approach: humans + autonomy rather than either alone
- 15:08 – 16:54
Human–AI teaming in deep space: MAIA (‘Alexa in space’) and the digital thread
Keoki introduces MAIA as an onboard AI assistant concept tied to model-based systems engineering and a ‘digital library’ of spacecraft knowledge. The vision is proactive anomaly detection, decision support, and enabling more exploration by monitoring massive telemetry streams.
- •MAIA as an AI assistant concept for astronauts
- •Model-based systems engineering and the ‘digital tapestry’ across lifecycle
- •Onboard access to test/flight history to spot anomalies early
- •AI monitors high-dimensional data humans can’t track; humans provide judgment
- 16:54 – 21:58
Testing the untestable: digital twins, non-determinism, and V&V for learning systems
Lex asks how you validate AI in space when you can’t easily test in the real environment. Keoki explains Lockheed’s reliance on sophisticated simulation/digital twins and highlights the new challenge: verifying non-deterministic learning systems that may operate outside designed envelopes.
- •Digital twins and decades of modeling space environments
- •Simulation of dynamics, radiation, controls, and mission scenarios
- •AI introduces non-determinism and online learning behaviors
- •Active research: trust, verification, validation, and safety cases for autonomy
- 21:58 – 27:20
Safety culture for complex software: lessons from aviation failures
The discussion shifts to software safety and culture, using the Boeing 737 MAX as a widely known case study. Keoki emphasizes a ‘mission success’ culture: skepticism, bounding behavior, deep testing, and continuous improvement grounded in root-cause analysis.
- •Safety culture differs across industries; aerospace emphasizes mission success
- •System complexity explodes state space; exhaustive verification is hard
- •Need for continuous inquiry: ‘did we consider the right possibilities?’
- •Learning from failures: root cause, corrective action, and improved standards
- 27:20 – 32:45
What Lockheed Martin builds today: aircraft, space systems, and integrated defense
Lex asks for a broad overview of Lockheed’s categories of work. Keoki frames the company as ‘solving hard mission problems’ across tactical aircraft, satellites (GPS, comms, missile warning), civil exploration, and integrated defensive systems like Aegis and THAAD.
- •Mission-first framing: systems are outputs of mission needs
- •Tactical aircraft at scale (e.g., advanced fighters) and future successors
- •Space portfolio: GPS, communications, missile warning, civil planetary missions
- •Defense integration: sensors, command-and-control, and human–machine interfaces
- 32:45 – 38:21
Skunk Works innovation playbook: small teams, stealth, and hypersonics
Keoki explains Skunk Works as both an organization and an idea: small, empowered teams operating close to the customer. He touches on stealth fundamentals (geometry/materials, continuous cat-and-mouse) and the renewed urgency around hypersonics driven by global competition.
- •Kelly Johnson’s principles: small expert teams, reduced bureaucracy, tight customer loop
- •Stealth basics: shaping/geometry and materials; ongoing radar vs stealth evolution
- •Hypersonics history (X‑15, Shuttle) and modern urgency due to geopolitical threats
- •Importance of sustained R&D investment and government–industry collaboration
- 38:21 – 42:18
Secrecy, inspiration, and mission impact: why engineers still join
Lex laments that much work is classified, potentially limiting inspiration for young engineers. Keoki argues there’s still plenty visible (e.g., F‑35 as an ‘information system’ aircraft) and shares an anecdote about GPS’s real-world impact on soldiers’ safety.
- •Balancing secrecy with public inspiration via visible platforms and capabilities
- •F‑35 described in unclassified terms: stealth, sensor fusion, mission management
- •Mission motivation: technology that protects lives and enables outcomes
- •GPS story as an example of engineering impact felt by billions and by troops
- 42:18 – 50:06
Autonomy in the cockpit: loyal wingmen, collision avoidance, and ‘optimal piloting’
Lex asks whether fighter jets will become fully autonomous. Keoki describes today’s reality: remote piloting, high autonomy in platforms like the F‑35, autonomous wingman demos (HAVRater), and autonomy as a safety backstop (Auto G‑CAS) plus Sikorsky’s Matrix for scalable piloting levels.
- •F‑35 shifts pilots toward mission management; high stability/automation aids carrier landings
- •HAVRater demo: autonomous F‑16 as loyal/trusted wingman in mission scenarios
- •Auto G‑CAS saves pilots from G‑LOC ground collisions; integrated collision avoidance expands to air-to-air
- •Sikorsky Matrix ‘optimal piloting’: dialable autonomy from two pilots to none; commercial and military use cases
- 50:06 – 58:48
AI arms race fears and governance: OODA loops, DoD policy, and ethics principles
Lex raises concerns about an AI arms race and losing human control. Keoki argues the near-term focus is augmenting humans (e.g., compressing OODA loops and handling dull/dirty/dangerous tasks) within policy constraints, pointing to DoD Directive 3000.09 and Lockheed’s internal ethics commitments.
- •AI advantage in time compression: computers vs human reaction times; OODA loop acceleration
- •Autonomy for sustained operations: logistics, surveillance, guarding beyond human endurance
- •DoD Directive 3000.09: human agency, realistic testing, training, transparent interfaces, doctrine
- •Ethical use inside the company: privacy, bias reduction, human accountability, strong ethics program
- 58:48 – 1:04:25
Deterrence and modern strategic threats: nuclear recapitalization and cyber as WMD-like risk
Lex asks why civilization hasn’t ended in nuclear war and what deterrence looks like now. Keoki discusses strategic deterrence logic (credible retaliation), modernization needs, and how cyber threats to critical infrastructure increasingly resemble strategic, national-survival risks.
- •Deterrence as prevention via credible capability and unacceptable cost to adversary
- •Recapitalization: ensuring systems work reliably and remain adaptable for decades
- •World is more complex/multipolar with proliferated missile and space capabilities
- •Cyber as a strategic threat to grids, water, and infrastructure—requiring new defenses
- 1:04:25 – 1:13:05
Military-industrial complex, competition in space, and a 100-year technology horizon
Lex asks about Eisenhower’s warning, then pivots to SpaceX and the future. Keoki emphasizes changing threat realities, the non-free-market nature of defense procurement, the value of competition and innovation in space, and a future that is faster, more connected, more autonomous—driven by investments like AI and quantum.
- •Defense investment framed by evolving global threats and mission needs
- •Close customer partnership and high-assurance engineering as a core competence
- •Space competition (e.g., SpaceX) as a catalyst; Lockheed innovations like CubeSats and software-defined satellites
- •Looking ahead: autonomy, connectivity, complexity, quantum computing, and expanded solar-system presence
