Nov 14, 2024, 3:00 AM ETInterview

Ex Google CEO: AI Can Create Deadly Viruses! If We See This, We Must Turn Off AI! - Eric Schmidt

70-20-10 Investment Rule: Google adopted a strategy allocating 70% of resources to the core business, 20% to adjacent ventures, and 10% to experimental projects.
- This model generated an estimated $10–$40 billion in extra profits over a decade, primarily driven by the "Google Brain" machine learning team (initially 10–15 people).
Emphasis on "Divas" over "Knaves": Successful leadership requires aligning with brilliant, high-conviction individuals ("divas" like Steve Jobs) who pursue perfection, while avoiding those who act solely for self-interest ("knaves").
Failure Velocity: The mantra "fail fast" is critical; Eric Schmidt notes that the ability to quickly identify and cancel bad ideas prevents resource drain and allows teams to pivot to viable innovations.
Organizational Isolation for Disruption: Major innovations require separating the "hunters" (disruptors) from the "harvesters" (core business operators).
- Example: Steve Jobs isolated the Macintosh team in a separate building with a pirate flag to protect them from corporate inertia; this structure is necessary because incentives for disruption differ from operational stability.
Founder Mode Tenure: Average tenure of companies on the S&P 500 has declined from 33 years to 12 years, projected to reach 8 years by 2050 due to accelerated disruption cycles.

Existential Necessity: Schmidt asserts that failing to integrate AI into every aspect of a business is a fatal strategic error; "if you're not using AI... you're not going to make it."
Future Tech Stack: Successful future companies will operate with a specific architecture: mobile apps on Android/iOS connected to powerful backend AI computers that "discover answers" rather than following rigid code logic.
Productivity Multiplication: AI is expected to double human productivity, enabling individuals to produce twice the output (e.g., generating double the content volume) without a proportional increase in labor.
Language of Development: Python is identified as the primary programming language for AI; while AI can write code, the ability to use APIs to build new tools remains a critical skill.
Critical Thinking Imperative: The ability to distinguish between marketing narratives (lies) and verified arguments is essential, as AI can generate "perfect misinformation" and reinforce confirmation bias via algorithms.
Algorithmic Manipulation: Social media algorithms (e.g., TikTok) optimize for attention and outrage, often driving users into "rabbit holes" of negative content, contributing to recorded increases in teen self-harm and depression.

Primary Fear: Schmidt's specific fear is not AI becoming sentient, but rather humanity failing to adopt AI fast enough to solve critical global problems in education, healthcare, and infrastructure.
The "Plug" Condition: A specific safety threshold has been defined: if AI agents begin communicating in a language unintelligible to humans (recursive self-improvement beyond oversight), the system must be physically unplugged.
Existential Threats: Major risks include:
- Cybersecurity: Raw models potentially discovering "day-zero" vulnerabilities (previously unknown threats) faster than human defense systems.
- Biology: The risk of AI designing new pathogens or weapons.
- Warfare: A shift toward drone-centric conflict where low-cost drones ($5,000) can neutralize high-cost tanks ($5,000,000), changing the fundamental logic of warfare.
Regulatory Approach: Schmidt advocates for government guardrails and "trust and safety" teams to test raw models before release, likening the need for physical security around AI data centers to that of nuclear facilities.
Geopolitical Dynamics: He predicts China will likely develop a different AI model than the West due to a state bias against freedom of speech, potentially resulting in more controlled but aggressive AI weapons.

Hiring Philosophy:
- Startups: Should prioritize raw intelligence, speed, and risk-taking over experience.
- Scale: Large companies often fail to innovate because they are too conservative; they must identify specific "owners" to drive disruption.
Focus Strategy: Contrary to "focus on what you are good at," Schmidt argues companies must predict the world in five years and pursue high-impact areas even if they are currently unprofitable (e.g., Google's early free products).
Culture Setting: Company culture is set by founders and must be enforced.
- Google maintained a technical culture where product quality drove customer acquisition, rendering sales forces unnecessary for superior products.
- Google discontinued "TGIF" (all-hands meetings) when the company grew too large, leading to leaks and the inability to maintain intimacy.
Remote Work Data: While Schmidt personally favors office work for learning and community, he acknowledges data indicates productivity is slightly higher with flexible remote work options.

Job Market Outlook: Automation will displace dangerous, repetitive, or boring jobs (e.g., security guards, basic assembly), but will not reduce the total number of jobs due to demographic shortages and the need for higher human productivity.
Human-AI Collaboration: AI is expected to augment, not replace, human roles in fields requiring empathy, moral judgment, and charisma (e.g., sports, arts, leadership).
Universal Basic Income (UBI): Schmidt rejects UBI arguments, stating that human desire for status, purpose, and contribution will drive labor demand even in an age of abundance.
Information Overload: The shift from scarcity to abundance in content creation means celebrities and creators must use AI to amplify their reach rather than fearing substitution by synthetic media.

Google Founding: Larry Page and Sergey Brin founded Google at Stanford using a NSF grant; their first server was powered by a borrowed dorm room outlet and built with corkboard to manage heat.
First Investor: Andy Bechtolsheim (co-founder of Sun Microsystems) provided the initial $100,000 (some accounts suggest $1M) check, believing in the founders' intelligence.
DeepMind Acquisition: Google acquired DeepMind (based in London) to secure a lead in AI, recognizing early on that the technology would define the future of the company.
Book Collaboration: Schmidt's book Genesis was completed with the late Henry Kissinger, who viewed the advent of AI as a question of human survival; the book's final chapter was finished on Kissinger's deathbed.

Advice to Youth: Learn to write in Python and, more importantly, develop analytical and critical thinking skills to verify information rather than accepting social media narratives.
Advice to Entrepreneurs: Say "yes" to limited-time opportunities; success often depends on luck and timing, but the critical move is accepting the invitation to act.
Prediction for 5 Years: The emergence of a "polymath" AI assistant that guides users through information overload, likely running on specialized hardware with enhanced network speeds.