Revolutionizing Human-Computer Interaction: New Computer's Approach to Natural, Contextual Communication

Join Sam and Jason, co-founders of New Computer, as they explore the future of human-computer interaction, where sensors, contextual modalities, and AI come together to create intuitive and natural interfaces.

  • 1. Sam and Jason, co-founders of New Computer, excited to start by showing their pores up close.
  • 2. They aim to change computing metaphors and abstractions, thinking from first principles about the future of intelligence.
  • 3. Define intelligence as the ability to process various types of information, reason, and find meaning.
  • 4. Humans perceive the world through senses, process information, form a theory of mind, and react accordingly.
  • 5. Communication involves more than just language; it includes non-verbal cues like eye contact, gestures, and tone.
  • 6. With the advent of chat GPT, computers are starting to approximate the human input-reasoning-output loop using language.
  • 7. New Computer aims to make this interaction feel even more natural for people.
  • 8. Demonstrating a live audio-visual demo using real-time pose measurements and an LLM (large language model) for context-aware interaction.
  • 9. The system decides whether to use keyboard input or text output based on the user's position and orientation.
  • 10. People are naturally sensitive to others, and computers should adapt to people rather than requiring users to adapt to them.
  • 11. Future of human-computer interaction may involve explicit social gestures and implicit gestures considering cultural norms.
  • 12. New physics could involve working with generative intelligence as a fluid probabilistic material, using flexible metaphors instead of rigid ones.
  • 13. Mixed reality experiences may allow users to point at objects and have devices understand their intentions through voice and gesture inputs.
  • 14. Designers must think on behalf of the user, considering context and appropriate output formats when working with multiple simultaneous inputs.
  • 15. New sensors and contextual modalities offer a wealth of opportunities for gathering context and extracting signals.
  • 16. LLMs can be used to decide how to respond in words and present information based on various inputs.
  • 17. Probabilistic interfaces are challenging due to their multiple possible outputs; familiar metaphors help ground these interfaces.
  • 18. Social norms should be considered when designing AI interactions, as certain conversations may not be appropriate.
  • 19. The future of truly intelligent interfaces remains to be determined and involves thinking beyond current isomorphism.

Source: AI Engineer via YouTube

❓ What do you think? What does it mean to "think on behalf" of a user, and how can designers and developers effectively integrate probabilistic interfaces into real-world applications? Feel free to share your thoughts in the comments!