Google Building AI World Model to Replicate Human Cognition

eWEEK’s content and product recommendations are editorially independent. We may earn money when you click our affiliate links. Find out more.

Google DeepMind is taking a decisive step toward the future of artificial intelligence by assembling a dedicated team of researchers to develop models of the world, an approach that could bring artificial intelligence closer to human cognition. Unlike today’s artificial intelligence simulations, which rely heavily on vast data sets and pattern recognition, world models aim to model the physical world and make predictions at large scales. This initiative will allow Google to achieve its main goal of achieving artificial general intelligence (AGI).

Bye Development of artificial intelligence Having made significant progress in recent years, world models represent an entirely new frontier that can take AI beyond reactive reactions into the realm of true understanding and foresight.

Who else creates models of the world?

Google DeepMind isn’t the only company pushing boundaries AI simulation with world models. NVIDIA recently introduced Cosmosplatform built on proprietary global models designed to revolutionize autonomous vehicles (AV) and robotics. Founder and CEO Jensen Huang believes these models will be a game changer, predicting a breakthrough moment similar to ChatGPT’s impact on large language models.

“This is the ChatGPT moment for robotics,” Huang said, emphasizing that global models are essential for the development of AVs and intelligent machines.

Meanwhile, AI Visionary Fei Fei Li’s World Labs is a pioneer in building large models of the world, with a focus on spatial intelligence, which allows AI to process and reason in our 3D world. These developments herald a new era in which AI doesn’t just recognize patterns, but understands and interacts with the physical world.

How world models can change AI

If Google and other artificial intelligence companies succeed in the world’s models, AI modeling can take unprecedented leaps forward. Instead of simply reacting to data, AI could anticipate, plan, and learn from experience, just like humans. This will make artificial intelligence systems significantly more efficient, adaptable, and capable of solving complex real-world problems with unprecedented autonomy.

World models may also be the missing piece in Google’s quest for AGI, which is Google’s ultimate goal. AGI refers to AI that can think, learn, and solve problems across a wide range of tasks, rather than performing specific functions such as the AI ​​models available today. AGI will also have the ability to reason, adapt, and make autonomous decisions. OpenAI CEO Sam Altman recently predicted that AI will be able to think like humans already this year.

However, determining whether AGI is achievable or still out of reach may take several years. Meanwhile, if the world’s models can deliver on their promises, they could mark a turning point in the evolution of AI when it truly understands the world in which it operates.