scripod.com

EP 498: Meta drops Llama 4, Microsoft Copilot levels up its AI game, GPT-5 roadmap hits snag and more AI News That Matters

Shownote

Welp, that escalated quickly. 🤯 After Google and OpenAI won the week last week, Microsoft and Meta decided to drop some pretty crazy AI features and models. How can anyone even keep up?! (With Everyday AI, and our weekly AI news that Matters segment, of...

Highlights

The world of AI continues to evolve rapidly, with major players like Microsoft, Meta, and OpenAI pushing boundaries. This episode delves into recent advancements in AI tools and models, including updates to Microsoft Copilot, the release of Meta's Llama 4, and Amazon's NovaAct AI agent. Additionally, it explores the implications of GPT-4.5 passing the Turing test and discusses the ongoing challenges and opportunities in the AI landscape.
05:49
Microsoft silently rolled out agent-based AI in browser with new actions features
06:16
Copilot Vision can generate podcasts with audio summaries
12:30
Midjourney V7 includes voice input for draft mode
15:34
Mini rag is now inside ChatGPT
15:50
ChatGPT Teams can retrieve and summarize Google Drive content accurately.
18:57
OpenAI delaying GPT-5 but releasing O4 Mini model soon
21:16
GPT-5 is delayed due to integration difficulties, despite exceeding expectations.
28:24
GPT-4.5 crushed the Turing test with a 73% win rate.
30:32
NovaAct is part of Amazon's AI initiative for diverse media input types
32:06
NovaAct positions Amazon among competitors in the autonomous AI systems race.
35:40
Llama 4 boasts a 10 million token context window
39:50
Llama 4 is not truly open source.
42:46
GPT-4.5 becomes the first model to officially pass the Turing test
47:30
Final two startups will be on a show and the prize will be announced live.

Chapters

Everyday AI News and Insights
00:00
Microsoft Expands Copilot Features
06:10
MidJourney V7: Voice and Draft Mode
09:39
Midjourney's Incremental Update Feedback
11:58
Google Gemini Accessibility Struggles
15:50
OpenAI Delays GPT-5, Announces Models
18:53
GPT-5: Delay Due to Integration Complexity
21:04
GPT-4.5 Passes Turing Test
24:31
Amazon Unveils NovaAct AI Agent
30:30
Understanding AI Agents vs. Models
32:06
Long Context Windows' Rising Impact
35:33
Chinese Models: Safety and Preference Insights
39:49
AI Updates: April 7 Recap
42:43
Subscribe to Everyday AI Newsletter
45:27

Transcript

Jordan Wilson: This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life. I hate saying this each Monday, but my gosh,...