Howard Marks: A Layman Thinks About AI

Howard Marks approaches artificial intelligence as an investor without claiming technical expertise, arguing that its potential importance makes it impossible to ignore. Because investing requires judgments about the future, he emphasizes both the difficulty of forecasting AI and the risk of confusing confidence with knowledge. An investment thesis depends on getting many linked variables right: how the technology develops, demand for computing power, who benefits, how competition unfolds, whether financing remains available, and when changes occur. Their combined uncertainty makes confident predictions especially hazardous, though uncertainty cannot be a reason to avoid investing altogether.

Marks organizes his thinking around four questions. First, AI systems are built through training, but their internal decision-making remains only partly understood, raising questions about whether their explanations can be trusted. Second, their expanding abilities - from pattern recognition toward reasoning - leave open whether they can be reliable, ethical, creative, skeptical, and aligned with human goals. Third, productivity gains may benefit AI users rather than technology vendors if models become interchangeable and price competition erodes returns; alternatively, dominant providers could command high prices. He also raises the possibility of taxes on AI replacing workers. Finally, rapid progress makes it uncertain whether today’s AI leaders can defend their positions against large technology companies or future entrants. The memo’s central point is that AI may transform the economy, but its winners, profits, and investment returns remain exceptionally difficult to identify in advance.