Jackson, Keoki

CTO of Lockheed Martin

Keoki Jackson: Lockheed Martin

Lex Fridman Podcast · Aug 19, 2019

• Keoki Jackson
• Lockheed Martin
• space exploration and Orion
• AI verification and validation
• autonomous weapons policy
• hypersonic weapons
• missile defense systems
• OODA loop
• nuclear deterrence
• military-industrial complex

Core ideas

Jackson traces his own path (via a love of space exploration and MIT) into Lockheed Martin, and frames the company's identity around "solving hard mission problems" rather than building specific products — citing the SR-71 (designed purely to outrun any air defense system of its era) as a template, and describing Lockheed's continuous involvement in essentially every U.S. planetary mission, including the Viking Mars lander and the OSIRIS-REx asteroid sample-return mission, whose "pogo stick" sample-collection method (a brief, seconds-long touch-and-go maneuver that blows regolith into a collection hopper) he cites as an example of engineer-driven improvisational ingenuity. On the long-term human spaceflight roadmap, he distinguishes "getting there" (already demonstrated in the 1960s) from "staying" — sustained, industrial-scale infrastructure (permanent habitation, vehicle reuse, fuel and consumables logistics) — as the real engineering and budgetary challenge behind current plans for a lunar base in the 2024-2025 timeframe as a stepping stone to Mars, with Lockheed's Orion spacecraft designed as the human-rated vehicle for that journey.

He describes Lockheed's "Mya" system — an AI assistant for astronauts in deep space, conceptually similar to Alexa — as emerging from the company's broader shift toward model-based systems engineering, where a full "digital twin" of a spacecraft (built from decades of accumulated materials, dynamics, and radiation-environment modeling) accompanies the physical vehicle and its test/flight history throughout its operational life, letting the AI flag anomalies before they escalate and surface additional exploration opportunities. He identifies verification and validation of non-deterministic, continuously-learning AI systems (as opposed to traditional deterministic aerospace systems) as a genuinely unresolved research problem — one Lockheed is actively funding at universities including MIT — and lays out the company's formal autonomy spectrum: no autonomy (human does the task), partial autonomy (human in the loop for every decision), supervisory autonomy (system does most work, human can intervene), and full autonomy (human out of the loop entirely), with different missions calling for different points on that spectrum.

Asked directly about the Boeing 737 MAX crashes, Jackson attributes aerospace's institutional safety culture — something he says isn't inherently taught in software engineering and may be irrelevant in other software domains like ad targeting — to decades of deliberate cultivation, and argues that as flight-critical software systems have grown enormously complex, exhaustively verifying every possible state and failure scenario has become computationally close to infeasible, requiring instead a culture of continuous skepticism, bounded testing, and rigorous root-cause analysis after failures (citing the Challenger accident and the commercial aviation industry's multi-year record of zero fatal accidents as evidence deliberate systemic improvement, not luck, drives safety outcomes).

Jackson highlights specific defensive systems, including Aegis (integrated fleet air and missile defense) and its land-based variant Aegis Ashore, and THAAD (theater-level ballistic missile defense), framing 1980s-era SDI-derived "hit-to-kill" technology (intercepting an incoming missile with a direct kinetic strike) as a genuine historical breakthrough, and describes hypersonic weapons (above roughly Mach 5) as an area of renewed U.S. investment given publicly claimed Russian and Chinese hypersonic capabilities that could potentially defeat existing missile defenses. On autonomy in aviation specifically, he describes "Have Raider" — a Skunk Works demonstration using an autonomous F-16 as a "loyal wingman" alongside a piloted aircraft — and Sikorsky's MATRIX "optimal piloting" technology (an FAA-certifiable, dial-adjustable autonomy system enabling zero-pilot, one-pilot, or two-pilot operation, including landing helicopters on oil rigs in poor visibility), alongside two concrete safety systems already deployed: Auto GCAS (automatic ground-collision avoidance for the F-16, credited with saving seven aircraft and eight pilots by preventing G-force-induced blackout crashes) and an integrated air-to-air collision avoidance system now also implemented on the F-35.

On AI-enabled autonomous weapons and the "AI arms race," Jackson states he doesn't see today's environment as a runaway race toward removing humans from lethal decision-making, citing DoD Directive 3000.09 (governing autonomous and semi-autonomous weapons systems), which mandates human agency and control, realistic testing and evaluation, operator training, transparent human-machine interfaces, and established doctrine for use of such systems — framing this as analogous to existing rules of engagement around chemical and biological weapons. He describes AI and autonomy's core near-term battlefield value through the OODA loop framework (observe-orient-decide-act): human reaction times operate on the order of tenths of a second while computers operate on nanosecond timescales (he draws the comparison that a nanosecond is to a second as a second is to 32 years), so AI-enabled systems that compress that decision loop confer a decisive tactical advantage — while also noting AI's complementary value at the opposite end of the timescale, in sustained tasks (surveillance, logistics, guarding) that exceed practical human endurance limits ("dull, dirty, and dangerous" work).

On nuclear deterrence, Jackson (whose company builds Trident submarine-launched ballistic missiles, among other strategic-deterrent systems) argues that despite the continued real possibility of human self-destruction, strategic deterrence has arguably prevented conflicts on the scale of the world wars of the early 20th century — while noting the deterrence landscape has grown far more complex since the Cold War, given that ballistic missile and space-based capabilities have proliferated from a handful of nations to over 70, and that new domains like cyber warfare (capable of targeting critical infrastructure like power grids and water systems) now carry comparably existential stakes to traditional weapons of mass destruction, motivating Lockheed's current strategic-systems recapitalization program. Addressing Eisenhower's warning about the military-industrial complex directly, Jackson responds by noting Lockheed's substantial non-military business (energy, commercial space, commercial aircraft), and argues that military spending as a fraction of GDP remains well below historical peak levels even amid what he characterizes as renewed "great power competition" and rising global threats — framing the close customer-contractor partnership typical of defense contracting as a necessary consequence of these systems' unique safety and mission requirements rather than a problematic lack of open market competition.

On the commercial space landscape, Jackson describes SpaceX and similar new entrants as welcome competitive forces (while emphasizing the importance of a level regulatory playing field), and cites falling costs of computing, communications, and additive manufacturing as driving an unprecedented wave of innovation across the industry, noting Lockheed's own early role (with Stanford) in launching one of the first CubeSats, and its current work on software-defined "smart satellites" that can be reconfigured for new missions on orbit. Looking a century ahead by analogy to the Wright Brothers, he declines to name a specific technological ceiling, predicting instead a future that is faster, more connected, more autonomous, and more complex — with continued investment in AI, autonomy, and quantum computing, and an eventual extension of both defensive and scientific presence from the deep oceans to the edge of the solar system.

Excerpts