Last Week in AI #217 - ChatGPT Agent, Kimi k2, Hiring Drama
507 snips
Jul 23, 2025 
In this discussion, John Krohn, CEO of Y-Caret and AI expert, shares insights on OpenAI's groundbreaking ChatGPT agent, capable of controlling computers and performing complex tasks. The conversation dives into fierce competition in the AI market with Google's acquisitions and Meta's strategic hires. Ethical concerns arise over AI-generated explicit content and facial recognition misuse. The episode wraps up with the recently resolved SAG-AFTRA strike, highlighting newfound rights and protections for video game voice actors amidst evolving tech challenges.
AI Snips
Chapters
Transcript
Episode notes
ChatGPT Agent's Powerful Leap
- OpenAI's new ChatGPT agent controls entire computers and excels in diverse tasks, combining browsing and research capabilities.
- It surpasses previous models in benchmarks, marking a major step in agentic AI performance.
User Experience with ChatGPT Agent
- Jon Krohn shared his experience using the ChatGPT agent, praising its efficiency and ability to perform complex computer tasks.
- Users can watch and even interrupt its actions, enhancing transparency and control.
Kimi K2’s Open Source Breakthrough
- Alibaba-backed Moonshot's Kimi K2 AI model challenges ChatGPT and Claude in coding yet costs less, showing rapid open-source progress.
- It uses a trillion parameters but activates only 32 billion, showcasing efficiency in large model design.