Kathleen Kenealy on Building Google's Open Gemma Models and Her Journey in AI

Kathleen's Path to AI and STEM

• 💡 Kathleen's journey into AI began with a love for math and problem-solving, initially double-majoring in math and theater at Stanford.
• 🚀 A pivotal internship led her to discover TensorFlow, sparking a realization that technology could be used to understand human behavior and communities.
• 🎭 Her background in theater fostered a deep appreciation for language and empathy, which she later found directly applicable to her work in natural language processing.

The Power of Open Technology

• 🔑 Kathleen highlights the profound impact of Google's open-source ecosystem, including TensorFlow and the Transformer architecture, on her own research and the broader AI community.
• 🤝 She emphasizes that open models allow for rigorous scientific evaluation, foster collaboration, and enable a wider audience to benefit from and contribute to AI advancements.
• 🚀 The development of Gemma, Google's family of open models, was driven by the belief that state-of-the-art technology should be accessible.

Building and Releasing Gemma

• 🧠 Kathleen was instrumental in initiating the project that led to Gemma, driven by the need to keep the open-source community from falling behind internal development.
• 🌟 Gemma, built from the same research as Google's Gemini models, offers both lightweight, on-device models and larger cloud-based versions, available across various platforms.
• 🛠️ The release of Gemma aims to empower developers and researchers to explore new frontiers in AI and large language models.

Navigating AI's Future and Risks

• ⚠️ Kathleen acknowledges the immense impact of AI as a double-edged sword, stressing the need to be aware of risks and actively avoid negative consequences for individuals and society.
• 🌱 She believes in the potential benefits of AI, particularly in healthcare, while advocating for a balanced approach that considers both risks and opportunities.
• 🚗 Drawing parallels to the development of the automobile, Kathleen trusts that society will develop necessary frameworks and safety measures, like traffic laws and seatbelts, to manage powerful new technologies.
• 🔬 Google employs rigorous processes, including red teaming, to ensure the safety and responsibility of its AI models before public release.