Eric Schmidt on Future War & AI Risks

Eric Schmidt describes the reality of modern war as more brutal than depictions in media, warning that cities face destruction and innocent lives are lost. He observed the widespread destruction of cities and the tragic loss of innocent lives firsthand, underscoring the grim reality of contemporary conflicts.

Schmidt stated that the only effective way to counter an aggressor is through direct resistance. He noted that ongoing innovation is driving advancements in both offensive and defensive military technologies.

The current conflict demonstrates a continuous cycle of technological development where new methods for attack are swiftly met with new forms of defense, illustrating an accelerating arms race.

Russia is now on an innovation path that mirrors Ukraine’s progress, Schmidt argues, leveraging a superior resource base to scale developments like jet-powered Shahed drones and independent glide bombs. This shift poses a critical challenge, as Ukraine continues to face a lack of effective defenses against ballistic missiles and shortages of Patriot systems.

Russia is rapidly scaling up new technologies, including jet-powered Shahed drones and independent glide bombs, which are proving difficult for Ukrainian forces to counter.

Beyond these immediate threats, Russia is developing a new satellite network, referred to as Rusevet, intended to enhance its communication and targeting capabilities across Ukraine. This technological advancement is compounded by its continued human mobilization efforts, increasing the strategic pressure on Ukraine.

The overall trajectory indicates that while Ukraine has shown remarkable ingenuity, Russia's ability to innovate and deploy technologies at scale, backed by superior resources, is creating a concerning imbalance on the battlefield.

Roughly a year away from becoming operational, the Russian-built 'Rusevet' satellite network is designed to enhance communication and targeting capabilities. Schmidt estimated this system is approximately a year away from becoming fully operational.

Once active, the network could render Russian operations virtually unblockable, allowing forces to 'hunt' anywhere within Ukraine. This capability would significantly escalate the strategic imbalance, giving Russia a considerable advantage in intelligence and precision strikes.

Ukrainian commanders are advocating for the use of Starlink within Russian territory to target critical infrastructure, such as oil facilities and ballistic missile launch sites. Zanny Minton Beddoes noted that commanders question why access to Starlink is restricted within Russia's borders, especially given the ongoing missile threats to Ukraine.

Schmidt stated that enabling Starlink access in Russia is a political decision, not a technical one. He also highlighted that the US government holds the authority to compel private companies like SpaceX to support national security objectives, suggesting a pathway for intervention.

Current technology on the front lines primarily uses machine vision for targeting, rather than true artificial intelligence. This technology allows weapons to lock onto specific targets with precision.

Schmidt clarified that machine vision is an engineering feat, but it does not involve the recursive self-improvement characteristic of true AI systems. True AI is defined by its ability to learn and improve autonomously over time.

AI's role in warfare is still in its nascent stages, with current applications largely focused on automating battle planning and synthesizing data. The more advanced capabilities of self-improving AI are just beginning to be explored in military contexts.

Future AI systems will leverage large language models and other tools to synthesize open-source, proprietary, and secret data, enabling them to reason through complex scenarios. This marks a significant shift from current AI applications.

Instead of merely assisting with targeting, AI will transition to providing high-level strategic advice to commanders. Schmidt envisions systems that can anticipate potential failures and recommend optimal solutions based on the positioning of assets and evolving battlefield conditions.

This advanced AI will act as a sophisticated advisor, offering proactive insights and solutions without replacing human decision-making. The goal is to enhance, not automate, strategic command.

Human intervention currently outperforms automated terminal guidance systems in contested environments, as jamming technologies frequently cause automated systems to lose their target locks. Schmidt noted the critical importance of the final mile of guidance for ensuring accurate military engagement.

For example, Switchblade 600 systems have demonstrated failures due to jamming during operations, underscoring the vulnerability of automated precision in the face of electronic countermeasures. Schmidt asserted that any situation where terminal guidance is not as effective as a human operator represents a critical flaw in the system.

Legal and moral accountability in warfare demands human oversight, as autonomous systems left without constraints lack the capacity to avoid targeting civilians or causing unnecessary conflict escalation. Schmidt emphasized that commands like 'attack this country' are dangerously broad and lack the necessary constraints.

Military targets must be defined by strict legal, time-bound, and spatial parameters, requiring human approval to maintain accountability and prevent indiscriminate actions. Without such oversight, AI systems, which lack moral judgment, cannot adhere to the laws of war.

The fundamental principle must be that humans retain ultimate control and responsibility, ensuring that ethical and legal considerations are paramount in the deployment of AI in warfare.

Historical examples, such as the Nuremberg trials, confirm that participants in war face personal liability for their actions, challenging the notion that total war nullifies legal constraints. Schmidt highlighted that awareness of personal accountability serves as a critical deterrent.

Even in intense, existential conflicts, the belief that justice will eventually follow persists. This historical precedent ensures that legal and ethical considerations remain relevant, influencing behavior even under extreme pressure.

Greater AI autonomy drives higher combat effectiveness, an edge that complicates Ukraine's defense strategy because tactical solutions are processed at speeds unattainable by humans, according to the logic of escalation. Schmidt warned about the inherent logic of escalation in this context.

If Russia is able to escalate its use of AI faster and with fewer limitations, Ukraine would face a considerable disadvantage, irrespective of international law. This dynamic underscores the urgency of addressing AI's role in military strategy.

The immediate priority must be to support Ukraine's defense capabilities. This approach is essential to avoid yielding to potentially illegal escalation tactics that could arise from an unchecked AI arms race.

US leadership recognizes the evolving nature of warfare, yet the military system is inherently resistant to change. Eric Schmidt pointed out that the military's structure is heavily influenced by Congressional actions, lobbying efforts, and special interests, rather than solely by the President.

Military conservatism is largely driven by an extreme aversion to loss, making leaders hesitant to replace proven, albeit expensive, systems with newer, cheaper, and unproven technologies. This resistance creates significant inertia within the defense establishment.

Schmidt proposed maintaining current budgets for existing infrastructure while allocating a separate, smaller budget specifically for emerging, effective, and cost-efficient technologies. This dual-budget approach could facilitate the adoption of innovation without immediately destabilizing the current defense framework.

Ukraine's drone brigades achieve remarkable success by being led not by traditional Soviet-era generals, but by former tech and business leaders. Schmidt highlighted that this civilian leadership has been crucial to Ukraine’s innovative achievements in warfare.

The strategic replacement of traditional military hierarchies with agile, tech-savvy civilian leadership has enabled Ukraine to rapidly adapt and implement new technologies, demonstrating a successful model for modern military innovation.

Reassigning existing military personnel in European countries such as Germany and the UK to AI-targeted systems would lead to full employment. Eric Schmidt asserted that every role would evolve, but positions would remain secure, addressing concerns about job displacement.

European labor laws provide inherent protections against mass layoffs, ensuring workforce stability even during significant technological transitions. This legal framework would support a smooth shift to AI-driven defense architectures without widespread job losses.

Eric Schmidt suggests the US holds an advantage over China in military innovation because China has not engaged in a war in 50 years. He believes China primarily copies current US methods, lacking the nuanced tactical doctrine developed through direct combat.

Ukrainian commanders have demonstrated world-class, subtle reasoning and tactical proficiency, which were honed through their extensive battlefield experience. This practical, real-world application of strategy is something China's military lacks.

The absence of recent combat experience means China's understanding of intricate tactical situations and rapid adaptation under pressure is likely undeveloped, giving the US a critical edge in the ongoing tech race.

An AI pause is unlikely, as the American incentive system fundamentally favors continuous progress rather than stagnation. Eric Schmidt explained that pausing would only create opportunities for local or global competitors to gain a strategic advantage.

The proposed solution involves simultaneously advancing AI technology while developing safeguards for the inherent risks it creates. This dual approach acknowledges the rapid pace of innovation and the competitive environment.

He emphasized that preventing unintended escalation is the primary diplomatic goal in the age of AI.

Schmidt noted that cybersecurity remains a critical, unresolved area, and without an agreement on limiting offensive and defensive cyberattacks, broader AI treaties are unlikely to succeed. This foundational issue must be addressed first.

Current oversight mechanisms, such as UN supervision, are impractical due to a lack of technical capability to verify code within secret military labs. Effective treaties require verifiable measures that can genuinely limit AI risks.

The mechanism involves a system where countries could remotely shut down each other's data center power if one side develops a dangerously large lead in AI innovation. This action would force a restart, leveling the playing field.

Schmidt described this as a serious, albeit unconventional, policy concept designed to act as a safeguard. The goal is to ensure no single entity pulls too far ahead in the AI race, thereby maintaining a global balance.

The interviewer expressed growing concern over the lack of an implementable AI pause, noting increasing alarm among Silicon Valley researchers. Eric Schmidt acknowledged that his own optimism has been tempered by the recent escalation in lethal innovation witnessed in Ukraine, which presents new challenges to AI safety.

Despite these concerns, Schmidt maintains that AI remains manageable as long as it stays under human control and alignment. He conceded, however, the critical need for effective diplomacy and international agreements to navigate the rapid advancements in AI technology.