I had a front-row seat to the social media revolution in global affairs roles at Twitter and Meta. The same mistakes are happening in AI

The rapid, unregulated development of AI, echoing the "move fast, break things" approach of early social media, presents significant systemic risks as it integrates into critical infrastructure like energy. Drawing parallels to social media's unchecked growth, the article warns that delaying governance until AI is ubiquitous will lead to uncontrollable problems, citing AI's massive energy consumption—projected to drive a 2-3x increase in data center electricity demand by 2030—and the potential for catastrophic failures if systems prioritize efficiency over safety. It advocates for proactive governance, robust testing, and collaborative public-private frameworks to ensure AI's safe and sustainable integration, emphasizing that the stakes are exponentially higher given AI's foundational role in modern society.

Analysis

The article conveys a "moderately negative" sentiment regarding the rapid, unregulated development of AI, drawing a direct parallel to the "move fast, break things" approach of early social media, which led to significant societal issues like misinformation and polarization. The author, with experience at Twitter and Meta (META), emphasizes that delaying governance until a technology is ubiquitous results in a loss of control, a lesson from the social media era. This time, the stakes are exponentially higher as AI is becoming foundational infrastructure, impacting sectors from energy to defense, and acting as a "public utility" that shapes resource allocation and decision-making.

Key risks include opaque models, algorithms prioritizing performance over safety, and a global governance void amidst rapid capital flows. A critical case study is energy, where AI racks consume 10-50 times more power than traditional systems, projected to drive a 2-3x increase in global data center electricity demand by 2030. This energy demand, coupled with potential AI-driven prioritization of industrial over residential customers during crises, underscores systemic vulnerabilities.

The article advocates for a new playbook, urging proactive governance, robust testing, and auditing from the outset, rather than retroactive fixes. It emphasizes treating AI infrastructure with long-term governance in mind, aligning public, private, and global actors to establish shared standards. This proactive approach is crucial to avoid repeating past mistakes and ensure AI's safe and transparent integration into critical systems.

The author stresses that the window for shaping these systems is closing, necessitating a shift from short-term optimization to long-term governance in the development of critical infrastructure like energy, compute, and data centers. Collaborative efforts, such as those seen at ADIPEC, are highlighted as essential for establishing shared standards and interoperable systems.

Key Decisions for Investors

• Investors should scrutinize AI companies' governance frameworks and long-term safety protocols, particularly those integrating AI into critical infrastructure sectors like energy.
• Monitor regulatory developments and international collaborations concerning AI standards, as proactive governance will significantly impact future operational costs, market access, and potential liabilities for AI-dependent firms.
• Evaluate the energy consumption profiles and sustainability strategies of AI-dependent companies, given the projected 2-3x increase in data center electricity demand by 2030 and its potential impact on operational expenses and environmental compliance.
• Assess the systemic risks associated with AI's integration into public utilities, considering potential for catastrophic failures, reputational damage, and the robustness of fail-safes in portfolio companies.