🤖 AI Systems - Human Exoskeletons | Drishtant Ghosh (Drix10)

🤖 AI Systems - Human Exoskeletons

AI systems should be built as exoskeletons for humans, not as total replacements. This approach focuses on effective ways for humans to do sense making with their swarm of agents, rather than relying solely on autonomy.

Key Points:

Human-AI Collaboration: Building AI systems as exoskeletons for humans enables effective collaboration and sense making, rather than relying on full autonomy.
Swarm of Agents: AI systems can be designed to work in conjunction with humans, leveraging the strengths of both to achieve better outcomes.
Trade-offs: This approach requires careful consideration of trade-offs between autonomy and human involvement, as well as the potential limitations of human-AI collaboration.

🔗 Resources:

📊 Startup Success Rates

Only 11% of Marc's 36 startups did well, while 50% never made money. This highlights the importance of perseverance and the need to take more attempts to achieve success.

Key Points:

Startup Success Rates: The success rate of startups is often lower than expected, with only 11% achieving significant success.
Perseverance: The key to success lies in perseverance and taking more attempts, rather than relying on a single idea or approach.
Trade-offs: This approach requires careful consideration of trade-offs between risk and reward, as well as the potential costs of failure.

🔗 Resources:

📸 Modeling and Consulting

The author had to choose between industry artist representation or modeling at Ford Models, and digital consulting for retail fashion brands, high-profile individuals, and artists.

Key Points:

Modeling and Consulting: The author had to make a choice between modeling and consulting, highlighting the diversity of career paths available.
Trade-offs: This approach requires careful consideration of trade-offs between creative pursuits and commercial opportunities.
Actionable Takeaway: Consider exploring multiple career paths to find the best fit for your skills and interests.

🔗 Resources:

🌳 California Live Oak

The California Live Oak is a type of tree native to the western United States, known for its distinctive shape and ability to thrive in a variety of environments.

Key Points:

California Live Oak: This tree is native to the western United States and is known for its distinctive shape and ability to thrive in a variety of environments.
Trade-offs: This tree requires careful consideration of trade-offs between its ability to adapt to different environments and its potential limitations.
Actionable Takeaway: Consider exploring the unique characteristics of this tree and how they can be applied to other areas of life.

🔗 Resources:

🤖 MLP-Splatting

MLP-Splatting is a technique for creating compact 3D scenes using hundreds of tiny MLPs, enabling natural decomposition and efficient rendering.

Key Points:

MLP-Splatting: This technique uses hundreds of tiny MLPs to create compact 3D scenes, enabling natural decomposition and efficient rendering.
Trade-offs: This approach requires careful consideration of trade-offs between the complexity of the scene and the efficiency of the rendering process.
Actionable Takeaway: Consider exploring the potential applications of MLP-Splatting in fields such as computer graphics and game development.

🔗 Resources:

🗣️ Expressive Multi-Indic Speech Model

The Rumik OSS-1 is an open-source expressive multi-indic speech model trained on less than 60k hours of data, highlighting the potential for data efficiency in speech modeling.

Key Points:

Rumik OSS-1: This speech model is trained on less than 60k hours of data, demonstrating the potential for data efficiency in speech modeling.
Trade-offs: This approach requires careful consideration of trade-offs between the quality of the model and the amount of training data required.
Actionable Takeaway: Consider exploring the potential applications of this model in fields such as language translation and speech recognition.

🔗 Resources:

🚫 See the Unseen Badge

The "see the unseen" badge is distributed based on the quality of the contributor's work, with low-quality trajectories failing QA and data not passing quality checks.

Key Points:

See the Unseen Badge: This badge is distributed based on the quality of the contributor's work, with low-quality trajectories failing QA and data not passing quality checks.
Trade-offs: This approach requires careful consideration of trade-offs between the quality of the contributor's work and the potential limitations of the badge system.
Actionable Takeaway: Consider exploring the potential applications of this badge system in fields such as data quality control and contributor evaluation.

🔗 Resources:

💸 Researcher Compensation

The average researcher makes $1.5k/day in cash compensation, with some researchers having access to 1000+ GPUs and earning up to $79k/day.

Key Points:

Researcher Compensation: The average researcher makes $1.5k/day in cash compensation, with some researchers having access to 1000+ GPUs and earning up to $79k/day.
Trade-offs: This approach requires careful consideration of trade-offs between the potential earnings of researchers and the potential costs of equipment and infrastructure.
Actionable Takeaway: Consider exploring the potential applications of this compensation model in fields such as research and development.

🔗 Resources:

🚫 OpenAI and Codex

OpenAI has been accused of training on in-progress work by Alpoge and Buckman, and offering to publish together after stealing their work.

Key Points:

OpenAI and Codex: OpenAI has been accused of training on in-progress work by Alpoge and Buckman, and offering to publish together after stealing their work.
Trade-offs: This approach requires careful consideration of trade-offs between the potential benefits of collaboration and the potential risks of intellectual property theft.
Actionable Takeaway: Consider exploring the potential applications of this case study in fields such as intellectual property law and research ethics.

🔗 Resources:

📚 Interactive Children's Book

Fable 5.1 is an interactive children's book that reads the story aloud, uses layered 3D illustrations, and animates every word, demonstrating the potential for AI-generated content in education.

Key Points:

Fable 5.1: This interactive children's book uses AI-generated content to create an immersive and engaging reading experience.
Trade-offs: This approach requires careful consideration of trade-offs between the potential benefits of AI-generated content and the potential limitations of the technology.
Actionable Takeaway: Consider exploring the potential applications of this technology in fields such as education and content creation.

🔗 Resources: