Key insights

• ⚙️ The internet connected existing computers, while AI is creating new computer capabilities.
• 🧠 AI involves creating new capabilities and functions within computers.
• 🤖 AI is different from traditional Von Neumann machines and has limited ability to interact with people and understand the world.
• 🔤 New computers have different nature and capabilities like understanding language and images.
• 💼 The limited number of entities that were perceived to need vast computing capability, such as governments and big insurance companies.
• 📱 Evolution of computers from expensive mainframes to affordable devices like smartphones and embedded systems.
• 🌐 Future computing will feature AI models of diverse shapes, sizes, and capabilities.
• 🔒 AI's user-friendliness challenges traditional lock-in concepts.

Q&A

• What does the future of computing entail?

  The future of computing will involve a variety of AI models of different shapes, sizes, and capabilities, trained on various data and subject to different policies. AI's user-friendliness may challenge traditional concepts of lock-in, creating a different landscape from prior computing generations.

• How has the evolution of computers progressed to the present day?

  The evolution of computers has transitioned from expensive mainframes to small, affordable devices like smartphones and embedded systems. Today, almost everything contains a computer chip, draws electricity, and is connected to the internet.

• What was Thomas Watson Sr.'s view on the need for computers, and how is it reflected in old science fiction?

  Thomas Watson Sr. famously stated that the world would not need more than five computers due to their size, cost, and maintenance requirements. This concept is reflected in old science fiction, where big supercomputers frequently played a significant role in the plot.

• What differentiates AI from traditional computers?

  AI involves creating new capabilities and functions within computers, which have different nature, including composability and understanding of language and images. AI is more likely to draw lessons from the early days of the computer industry rather than the early days of the internet.

• How do the internet boom and the AI boom differ?

  The internet boom was driven by network effects and positive feedback loops, whereas the AI boom is more akin to the creation of a new kind of probabilistic computer with different industry, competitive, and startup dynamics.

• Why doesn't the analogy between the current state of AI and web 1.0 work?

  The analogy between the internet and AI doesn't work because the internet was a network that connected existing computers, while AI is focused on creating new computer capabilities and functions.

• 00:03 The analogy between the current state of AI and web 1.0 doesn't work because the internet was a network while AI is a computer. The internet connected existing computers, while AI is creating new computer capabilities.
• 01:45 The internet generation of startups was driven by network effects, while AI is more like a computer or a new kind of probabilistic computer.
• 03:13 The nature of new computers is different, allowing for composability and capabilities like understanding language and images. Lessons more likely to be drawn from early days of computer industry than early days of the internet. Networks are fundamentally different in evolution and adoption curve. Industry could be dominated by a few God models or a large number of different models and sizes.
• 05:03 Thomas Watson Sr., the creator of IBM, famously stated that he didn't think the world would ever need more than five computers because of their size, cost, and maintenance requirements. This concept is reflected in old science fiction where big supercomputers play a significant role in the plot.
• 06:28 The evolution of computers has gone from expensive mainframes to small, affordable devices like smartphones and embedded systems. Today, almost everything contains a computer chip, draws electricity, and is connected to the internet.
• 08:00 The future of computing will involve a variety of AI models of different shapes, sizes, and capabilities, trained on different data, running at various scales, and subject to different policies. AI's user-friendliness may challenge traditional concepts of lock-in, creating a different landscape from prior computing generations.