What new capabilities will GPT-4 offer with image inputs?

Question

Accepted Answer

GPT-4 Will Process Image and Text Inputs for Enhanced Functionality

In the future, GPT-4 is slated to offer enhanced capabilities by processing both image and text inputs. This dual input functionality is expected to represent a significant advancement in AI technology, allowing for more comprehensive understanding and interaction with user queries. While this feature is not yet released, its potential applications, such as interpreting visual information and generating relevant text-based outputs, are a key development for the technology.
Identification and Recipe Suggestion from Images

A notable future capability of GPT-4 will be its ability to interpret visual data and provide actionable information. For instance, if presented with an image of the contents within a refrigerator, the AI is designed to not only accurately identify each item but also to generate recipe suggestions based on the available ingredients. This demonstrates a move towards more practical and user-centric applications of advanced AI.
Collaboration with Be My Eyes for Advancements

OpenAI is actively preparing for the release of these new multimodal features through collaboration. The company has been working with Be My Eyes, an organisation focused on assisting visually impaired individuals, to help develop and refine these upcoming advancements. This partnership suggests a focus on leveraging AI's visual processing capabilities for accessibility and broader utility.
Improved Accuracy and Reduced "Hallucinations"

Beyond its multimodal potential, GPT-4 is also engineered for greater accuracy and reliability compared to its predecessors. OpenAI has stated that the new model is significantly less likely to generate disallowed content, with an 82 per cent reduction in such responses, and is 40 per cent more prone to delivering factual answers. This focus on safety and alignment was a six-month effort by the company.

Reading history

Hide →

This history is cleared when the page is closed.

What new capabilities will GPT-4 offer with image inputs?

Loading

Was this article helpful?

Helpful

Not helpful

Report an issue

1

0

0

Share with friends