Jun 24, 2025

Google Gemma and Open Source AI Models - Clement Farabet

Clement Farabet argues true autonomous AI isn't about human-mimicking bots, but highly constrained, domain-specific tools. Discover how combining open-weight Gemma with cloud-based Gemini unlocks this new architectural paradigm.

Pause
Mute Enter Fullscreen
#1 about 5 min

Early developments in neural networks and AI training

How academic research in neural networks laid the groundwork for modern AI infrastructure.

#2 about 2 min

The rapid progression of AI computation and hardware

The transition from awaiting faster hardware to accelerating capabilities with generative AI and transformers.

#3 about 3 min

Understanding the differences between Gemma and Gemini models

How Google separates local devices from powerful cloud-based API solutions to balance workloads.

#4 about 3 min

Distilling cloud AI capabilities into local device models

The process of compressing large language configurations to run efficiently on edge processors.

#5 about 2 min

Customizing foundation models for specific software domains

Enabling developers to fine-tune base open implementations into specialized variants for targeted tasks.

#6 about 2 min

Growing an active open-source developer AI community

Improving external accessibility and progressive feature updates to engage developers using local models.

#7 about 3 min

Licensing models and the monetization of bespoke applications

Enabling developers to build and commercialize localized pipelines through permissive open-source model weights.

#8 about 6 min

Building web apps and local agents with AI tools

Structuring entire applications and multimodal local assistants directly within AI Studio integrations.

#9 about 2 min

Addressing iterative code generation and developer challenges

Highlighting current difficulties around modifying targeted codebase paths versus regenerating full code iterations.

#10 about 4 min

Shifting application design toward contextual operating systems

Replacing rigid file schemas and applications with natural language processors and dynamic rendering.

#11 about 3 min

Balancing execution speed and mobile battery consumption constraints

Reducing real-time continuous computing loads on edge devices via structured cloud server delegation.

#12 about 6 min

Distinguishing true autonomous agents from basic automation scripts

Structuring AI behavior tightly around bounded objective goals to minimize unsafe systemic operations.

#13 about 5 min

Optimizing algorithmic forms for physical and digital execution

Creating efficient models over imitating humans while matching input designs to practical reality.

#14 about 2 min

Bootstrapping application AI projects without a startup budget

Leveraging Python platforms, flexible infrastructure, and free open model tiers to rapidly test capabilities.

#15 about 1 min

Establishing permissive but safe AI usage licenses

Providing flexible community agreements that explicitly forbid malicious behaviors like automated media manipulation.

#16 about 6 min

Contributing meaningful layers to the AI software ecosystem

Constructing accessible integrations and responsive interfaces instead of attempting to build new foundation algorithms.

Matching moments

4:04 min

Building practical AI agents using Google Gemini

Philipp Schmid Philipp Schmid · WWC 2025

2:54 min

Leveraging Chrome AI and nano models for web applications

Raymond Camden · Perf + AI

3:02 min

Understanding Google Gemini history and available context models

3:55 min

Technical catalysts driving real-world artificial intelligence

Clemens Wasner Clemens Wasner +3 · WWC Europe 2026

3:13 min

Embedding generative AI in enterprise software platforms

Mike Butcher Mike Butcher +3 · WWC 2024

5:05 min

Exploring popular generative AI models and applications

Mary Grygleski Mary Grygleski · LIVE

Upcoming sessions on this topic

Open session

World Congress 2026 North America

Building Stuff with GenAI - The Open Minded Workshop beyond OpenAI

Andreas Erben

CTO for Applied AI and Metaverse at daenet

Andreas Erben
Open session

World Congress 2026 North America

No Single Model to Rule Them All: Building Resilient AI Agents Across Open & Closed LLMs

Emmanuel Acheampong

Senior Manager Developer Relations at Crusoe AI

Emmanuel Acheampong
Open session

World Congress 2026 North America

Beyond Vibe Coding: Using CLI Tools as Your AI Counterpart in Pair Programming

Ron Veen

Java enthusiast and Special agent for Team Rockstars IT

Ron Veen
Open session

World Congress 2026 North America

From Software Agents to Physical Devices: Inside the Agentic Hardware Stack

Michael Yuan, Vivian Hu

Michael Yuan
Vivian Hu
Open session

World Congress 2026 North America

What's new, what's next: the latest models and developer tools from Google DeepMind

Paige Bailey

Member of the Technical Staff at Google DeepMind

Paige Bailey
Open session

World Congress 2026 North America

The Broken Rung: How AI is Rebuilding Software Development from the Ground Up

Tomislav Tipurić

Chief Technology Officer, Nephos

Tomislav Tipurić