Google DeepMind's AGI framework

[Skip to main content](https://deepmind.google/blog/google-deepmind-at-icml-2024/#page-content)

July 19, 2024
Research

# Google DeepMind at ICML 2024

Share

![](https://lh3.googleusercontent.com/v9vMUwpupK73jF5l-Blwsxd7CCKQF7yUeoU7N7GSoTHoMxW6G_LTB5opS7qOEn8n38NqLM3V7yD7EdovPh4f6uCht2EeBoPDPkUgo1pH9bRCkmav5VY=w1440-h810-n-nu)

Exploring AGI, the challenges of scaling and the future of multimodal generative AI

Next week the artificial intelligence (AI) community will come together for the 2024 [International Conference on Machine Learning](https://icml.cc/) (ICML). Running from July 21-27 in Vienna, Austria, the conference is an international platform for showcasing the latest advances, exchanging ideas and shaping the future of AI research.

This year, teams from across Google DeepMind will present more than 80 research papers. At our booth, we’ll also showcase our multimodal on-device model, [Gemini Nano](https://deepmind.google/technologies/gemini/nano/), our new family of AI models for education called [LearnLM](https://blog.google/outreach-initiatives/education/google-learnlm-gemini-generative-ai/) and we’ll demo [TacticAI](https://deepmind.google/discover/blog/tacticai-ai-assistant-for-football-tactics/), an AI assistant that can help with football tactics.

Here we introduce some of our oral, spotlight and poster presentations:

[Google Research at ICML 2024](https://research.google/conferences-and-events/google-at-icml-2024/)

## Defining the path to AGI

What is artificial general intelligence (AGI)? The phrase describes an AI system that’s at least as capable as a human at most tasks. As AI models continue to advance, defining what AGI could look like in practice will become increasingly important.

We’ll present a framework for [classifying the capabilities and behaviors of AGI models](https://icml.cc/virtual/2024/poster/35180). Depending on their performance, generality and autonomy, our paper categorizes systems ranging from non-AI calculators to emerging AI models and other novel technologies.

We’ll also show that [open-endedness is critical to building generalized AI](https://arxiv.org/html/2406.04268v1) that goes beyond human capabilities. While many recent AI advances were driven by existing Internet-scale data, open-ended systems can generate new discoveries that extend human knowledge.

![](https://deepmind.google/static/images/fallback/video_poster_fallback_light.57b61d57842a.jpg)![](https://deepmind.google/static/images/fallback/video_poster_fallback_dark.4d55bc99efa6.jpg)



Your browser does not support the video tag.


At ICML, we’ll be demoing Genie, a model that can generate a range of playable environments based on text prompts, images, photos, or sketches.

## Scaling AI systems efficiently and responsibly

Developing larger, more capable AI models requires more efficient training methods, closer alignment with human preferences and better privacy safeguards.

We’ll show how using [classification instead of regress

... (truncated, 6 KB total)