AI Now Learns by Watching, Not Just Following Instructions
OpenAI and Anthropic announced features allowing AI systems to learn by demonstration.
1A New Era in AI
In the summer of 2023, OpenAI and Anthropic announced new features allowing AI systems to learn tasks by watching them performed once. This marks a significant departure from traditional instruction-based methods. OpenAI's Record Replay feature enables ChatGPT and Codex users to demonstrate workflows and turn them into reusable skills.
2Anthropic's Approach
Anthropic introduced Record a Skill within Claude Cowork. This feature allows users to record their screen while performing a task and narrate their reasoning. Claude then converts this into a skill that can be executed again. This approach closely mirrors OpenAI's solution.
3Competition and Innovation
Two competitors unveiled similar solutions within weeks of each other, signaling a new era in AI training. This is seen as an admission that instructions alone cannot take AI to the desired level. Learning by observation could enable AI to handle more complex tasks.
4Potential of Record Replay
OpenAI's Record Replay feature offers significant convenience, especially for developers. Users can demonstrate workflows and automate them, saving time. This could increase efficiency in software development processes and reduce error rates.
5Advantages of Claude Cowork
Anthropic's Record a Skill feature allows users to teach tasks more naturally. The combination of screen recording and voice narration helps AI better understand the logic behind tasks. This could be particularly useful for complex operations.
6The Future of AI Training
Learning by observation could become a new standard in AI training. This approach enables AI to learn in a more human-like manner, allowing it to take on more complex and customized tasks. This could revolutionize the industry.
7Company Strategies
The similar features introduced by OpenAI and Anthropic indicate a common need in AI training. Both companies are developing innovative solutions to make AI learning more effective and efficient. This competition could accelerate progress in the sector.
8The Future of AI
Learning by observation features represent a significant step forward for AI. This approach enables AI to learn in a more human-like manner, allowing it to take on more complex and customized tasks. This could revolutionize the industry.
Original article: Fast Company — Yapay Zeka Artık Sadece Talimatla Değil, Gözlemle Öğreniyor