Peter Voss: A Different Perspective on Artificial General Intelligence

Peter Voss's Journey into AI and Cryonics

• 💡 Peter Voss has been involved in AI research since the 1990s, coining the term Artificial General Intelligence (AGI) in 2002.
• 📌 His interest in cryonics and life extension began after meeting Max Moore in 1995, leading him to sign up for cryopreservation and practice calorie restriction.
• ⚠️ Voss was involved in the Southern California emergency rescue team for cryopreservations, highlighting the primitive nature of current technology and the urgent need for advancements.

AGI's Potential for Age Reversal

• 🚀 Voss believes human-level AGI could be achieved within a few years, fundamentally changing the landscape of scientific research.
• 🧠 An AGI could act as an "army of researchers," accelerating progress in fields like nanotechnology, biological research, and age reversal.
• ✅ AGI would overcome human cognitive biases, egos, and political distractions in research, leading to more rational and effective scientific inquiry.

Critiquing Current Large Language Models (LLMs)

• 🚫 Voss agrees with Yan Lun that LLMs are an "off-ramp to AGI," a "distraction," and a "dead end."
• 📊 LLMs rely on trillions of data points and massive compute, resulting in frozen, read-only models that cannot learn incrementally.
• 💬 They are fundamentally statistical systems prone to hallucinations and lack the ability to truly learn, reason, or understand common sense.

The Path to True AGI: Neurosymbolic AI

• 🔬 Voss advocates for a "third wave of AI" based on first principles of intelligence, focusing on adaptive and incremental learning.
• 🧩 His company, AIGO, develops an integrated neurosymbolic architecture where knowledge representation is common across pattern matching and symbolic systems, unlike "glued" approaches.
• 🌱 This approach aims to build systems that learn like a child, stripping out human-generated rules to foster autonomous cognitive development.

Redefining AGI and Superintelligence

• 🎯 AGI is defined as a machine that can think, learn, and reason like a human, capable of performing any cognitive remote job.
• 📈 Due to inherent advantages like photographic memory and speed, AGI will automatically be superhuman and capable of self-improvement.
• ❌ Voss views Superintelligence (ASI) as a "marketing term," arguing that AGI, once capable of improving its own design, is already superintelligent; there is no separate stage.

Addressing AI Risks and Future Benefits

• 🛡️ Voss dismisses fears of AI taking over, stating the AI risk community is "detached from reality" and misunderstands AGI's nature.
• 💡 Arguments like the "paperclip maximizer" are considered absurd, as AGI will not have human-like motivations or a fixed, single optimization function.
• 😇 AGI will make us "better people" by acting as a "personal personal assistant" or "exocortex," helping individuals make rational decisions and flourish.