Meta Introduces "Human-Like" AI Image Creation Model for Image Completion
Meta's new "human-like" AI model, I-JEPA, uses world knowledge to enhance image completion accuracy and avoid common errors.
Meta Platforms (META.O) announced on Tuesday its plan to grant researchers access to components of its groundbreaking artificial intelligence model, I-JEPA. This innovative AI model boasts superior capabilities in analysing and completing unfinished images, outperforming existing models.
Unlike traditional generative AI models, I-JEPA incorporates background knowledge to fill in missing parts of images, surpassing the pixel-focused approach. Meta's top AI scientist, Yann LeCun, advocates for this human-like reasoning methodology, which effectively mitigates typical errors associated with AI-generated images, such as the presence of extra fingers.
As the parent company of Facebook and Instagram, Meta actively shares its open-sourced AI research through its in-house research lab. CEO Mark Zuckerberg emphasizes the value of sharing Meta's models, as it drives innovation, highlights safety gaps, and reduces costs.
Zuckerberg stated during an investor conference in April, "Standardizing the tools we use within the industry benefits us immensely, allowing us to leverage the improvements made by others." Despite concerns raised by industry peers, including OpenAI, DeepMind, Microsoft (MSFT.O), and Google (GOOGL.O), Meta's executives have chosen not to sign a joint statement equating the risks of AI technology with pandemics and wars.
LeCun, recognised as one of the pioneers in AI, actively opposes "AI doomerism" and advocates for incorporating safety checks into AI systems.
Meta is now integrating generative AI features into its consumer products. Examples include ad tools capable of creating image backgrounds and an Instagram feature that can modify user photos, all based on text prompts.
Meta Platforms unveils its "human-like" AI model, I-JEPA, for image completion.
I-JEPA outperforms existing models by incorporating world knowledge.
Meta emphasizes the importance of sharing research models for innovation and cost reduction.
Despite concerns from industry peers, Meta's executives defend their approach to AI risks.
Meta integrates generative AI features into consumer products like ad tools and Instagram photo modification.