Creativity, Automation, and the Power of Phi-4 | Ep. 02 with Jennifer Marsman
AI as a productivity booster: From automation to creativity, Phi-4’s power, and why humans still lead in innovation.
Our speakers
Jim Bennett is head of developer advocacy at Pieces, focusing on enabling developers to be more productive by leveraging contextual awareness of not only the code they write but the content they read and the conversations they have.
Jennifer Marsman is the Principal Engineer of Microsoft’s “AI for Earth” group, where she uses data science, machine learning, and artificial intelligence to aid with clean water, agriculture, biodiversity, and climate change. Jennifer is a frequent speaker at software development conferences around the world. Since 2016, Jennifer was recognized as one of the “top 100 most influential individuals in artificial intelligence and machine learning” by Onalytica, reaching the #2 slot in 2018.
🎧 Listen on Spotify.
Insights
🟢 AI is a powerful assistant for productivity, handling busy work so humans can focus on creativity and decision-making.
🟢 AI can be used for fun and practical tasks, from writing bedtime stories to managing pet medication schedules.
🟢 AI is best as a copilot, not a pilot, meaning it should assist humans rather than replace them entirely.
🟢 Humans still outperform AI in creativity, while AI works best as a tool for automation and augmentation.
🟢 Not all problems require AI—sometimes traditional coding is the better solution for efficiency and control.
🟢 Phi-4 is a compact AI model, trained on high-quality textbook data, capable of running on-device for enhanced privacy and speed.
🟢 Phi-4 has practical applications, such as trash sorting and other real-world problem-solving tasks.
🟢 Engineers need to develop skills in reviewing AI-generated code, ensuring quality, accuracy, and reliability in AI-assisted development. 🚀
Timeline
00:00 - Intro
01:16 - The ways Jennifer uses AI to be more productive - writing poems and bedtime stories
02:14 - Using AI to convert time-based rules into a schedule
03:44 - AI is great for busy work, so we can do the fun stuff!
04:37 - Jennifer and Jim agree that AI is best as a copilot, not a pilot
05:00 - Humans are better than AI at creativity, AI is just a tool
07:40 - Not all problems need AI, sometimes code is better
11:15 - Phi-4, trained on textbook quality data and runs on device
13:30 - Phi-4 for trash sorting
16:00 - Phi-4 is available in Pieces! 🎉
16:22 - How Phi-4 was trained on synthetic data
18:50 - How do engineers gain skills to review AI generated code?
