NEAR AI Office Hours with Atlas Computing

Опубликовано: 14 Август 2024
на канале: NEAR Protocol

217

Illia & Alex discuss the latest AI research topics and unpack some NEAR.AI focus areas.

NEAR AI: https://near.ai/
Illia Polosukhin https://x.com/ilblackdragon
Alex Skidanov https://x.com/AlexSkidanov

Atlas Computing
https://x.com/safewithatlas
https://atlascomputing.org/

In this episode, the team from Atlas Computing joins the discussion on the critical importance of AI safety and trust. They explore the challenges of ensuring AI behaves as intended, the role of formal verification, and how to design systems that prioritize human-centered outcomes.

High-Level Takeaways:
• AI Safety Requires More Than Testing: Traditional methods like testing and evaluations aren't enough for ensuring AI safety, especially in critical systems. A more rigorous approach, including formal verification, is necessary.
• Human-Centered AI Design: It's crucial to design AI systems that prioritize the needs and safety of users, rather than just the interests of the companies developing them.
• The Role of Formal Verification: Formal verification is becoming essential in building trust in AI systems, especially in high-stakes environments like blockchain and critical infrastructure.
• System-Level Safety Concerns: AI safety isn't just about the AI itself but the entire system in which it operates. Understanding and verifying system behavior is critical to preventing unintended consequences.
• Global Collaboration and Governance: Ensuring AI safety and trustworthiness requires collaboration across industries, governments, and academic institutions to develop standards and best practices.

Timestamps:
00:00:00 - Introduction
Introduction to AI Safety and Atlas Computing
00:03:00 - AI Safety Challenges
Discussing the Current State of AI Safety and Critical Risks
00:07:00 - Formal Verification in AI
The Role of Formal Verification in Ensuring Safe AI Systems
00:14:00 - Human-Centered AI and User Safety
Focusing on User-Centered Design for AI Safety
00:19:00 - The Future of AI Safety Tools
Emerging Tools and Technologies to Enhance AI Safety
00:27:00 - Challenges of Specifying Safety Requirements
The Difficulty of Defining and Ensuring AI Safety Standards
00:34:00 - Global Consensus and Interoperability
The Need for Global Collaboration in AI Safety Standards
00:43:00 - Rewriting the Internet and AI Safety
How Blockchain and AI Could Lead to a Safer Digital Infrastructure
00:52:00 - Moving Beyond Fear: Designing Safe AI Systems
Focusing on System-Level Safety Over Fear-Mongering
00:56:00 - Closing Thoughts
Final Reflections on the Future of AI Safety and Trust

Join NEAR's community:
Website: https://near.org/
NEAR AI: https://near.ai/
Twitter: / nearprotocol
Blog: https://near.org/blog/
Medium: / nearprotocol
GitHub: https://github.com/near
Dev Docs: https://docs.near.org/
#NEAR #nearprotocol #NEARAI