This Time It's Different: AI Startups Across Three Generations

AI's Rollercoaster History

• 💡 The history of AI has been a rollercoaster of booms and busts, with periods of hype often followed by "AI winters" where the term became less favored.
• 📌 Early AI, such as expert systems in the 1980s, faced significant limitations due to inadequate technology and incorrect assumptions about necessary hardware like Lisp machines.
• 🧠 Many advances in AI occurred under different names, including voice recognition and statistical machine learning, before the term "artificial intelligence" fully re-emerged in recent years.

Evolution of AI Applications

• 🚀 Siri was initially conceived as a "do engine" focused on performing actions and integrating with an open ecosystem of partners, rather than just providing search results.
• 🎯 Steve Jobs was instrumental in boldly proclaiming Siri an AI company, which helped to reinvigorate the term "artificial intelligence" before its widespread adoption.
• 🛠️ The original Siri aimed to excel at natural language processing, be extensible across various domains, support conversational dialogue, and execute actions, laying groundwork for what is now known as agentic AI.

Liquid Neural Networks

• 🔬 Liquid AI is developing liquid neural networks, a novel approach inspired by biological systems like C. elegans, specifically for safety-critical applications in the physical world.
• ⚡ These networks are remarkably smaller and more energy-efficient (e.g., 19 neurons for autonomous driving compared to 100,000 in traditional models) and offer causal reasoning capabilities.
• 💡 Liquid neural networks demonstrate better generalization, can understand physics, and are designed to run on-device (edge AI), thereby reducing cloud dependency and democratizing AI access.

AI's Societal Impact

• 📈 AI represents an advance in automation that fundamentally changes the nature of work, creating new job categories while requiring individuals to develop adaptability and interpersonal skills.
• 🌍 AI possesses the potential to solve complex global problems, such as protein folding and climate change, positioning it as a powerful tool for humanity.
• ⚠️ Concerns include the possibility of an economic bubble and the high energy consumption associated with large language models, highlighting the need for more efficient AI solutions.

Entrepreneurship and Future Outlook

• 🔑 Successful entrepreneurship is primarily driven by passion, curiosity, and a clear vision for addressing real-world challenges, rather than solely by the desire to start a company or accumulate wealth.
• 📚 Developing AI literacy is crucial for the public, encompassing an understanding of AI's statistical nature and its implications across diverse professions.
• 🔮 Large language models (LLMs) are evolving search engines, but the future of AI will require balancing powerful capabilities with energy efficiency and on-device processing for broader accessibility and impact.