MARK A. MACINNES, Ph.D.
Los Alamos
As a LANL retiree after 37 years of service, my career was focused on biological research and Nuclear Quality Assurance roles. I saw firsthand how competition between nations drove the world to make large investments in nuclear deterrence. Alarmingly, the arms race isn’t over, with more countries pursuing the ‘security’ of atomic weapons.
Today, with the advancement of large-scale Artificial Intelligence (AI) computing, we are experiencing a déjà vu moment. We need to find an alternative approach to the winner-takes-all ‘arms race’ mentality to develop AI superintelligence (ASI), which I define as machines operating far beyond human ability—unlike today’s AI tools, which we still understand and control reliably. The AI arms race is already spiraling out of control financially. There must be balanced and sensible limits, which we will discuss. Let’s compare the fever dream of an ASI arms race with the current practical applications of AI, both for societal good and for AI deterrence to prevent misuse.
AI for Good emphasizes the responsible use of AI in education and adult commitment to adapt to rapidly changing careers, supporting the region's economic sustainability. We can't expect children to use AI effectively unless their teachers, mentors, and parents are willing to engage. Let’s spark community interest and debate! Los Alamos has always attracted some of the most engaging speakers on a wide range of topics, including those leading the discussion on AI, both for and against. More specialized roundtables for educators will be crucial for professional growth. AI trade fairs showcasing the latest AI advances in the arts, history, culture, science, and philosophy would likely appeal to us. The Chamber of Commerce could promote speakers about AI best practices in business. Los Alamos could become, with support, a model for AI educational and business innovation hubs, as well as a leader in civic transparency.
AI for deterrence involves efforts to guard against genuine threats or misuse against malicious rogue actors, counterterrorism, cyberattacks, and disinformation. LANL will likely anticipate potential crisis scenarios and assist in developing effective AI security solutions. The United Nations can, with increased support, promote a global AI trust and verification framework. This must improve upon the legacy of Cold War nuclear treaties, which were hard-won but easily broken or ignored. Just as some nuclear threats have been mitigated through force, the threat of direct action or sanctions must remain part of our nation’s AI deterrence strategy.
Los Alamos plays a special role. Our crucial history in nuclear national security gives the Los Alamos Community strong moral authority and responsibilities. None of these goals, as outlined, will be easy, but we must start urgently.
Various AI tools contributed to the research and editing of this editorial. However, the thoughts, themes, facts and opinions expressed are my own.