# They Edited Pigs to Grow Human‑Ready Kidneys!

Source: https://www.youtube.com/watch?v=yMfktG1U-ak
Recap page: https://rapidrecap.app/video/yMfktG1U-ak
Generated: 2025-09-12T15:03:41.519+00:00

---
## Quick Overview

AI models can now interpret human social scenes and emotions with human-like accuracy, according to a new study on GPT-4V, which researchers found to be more effective than previous AI models in understanding complex social cues.

**Key Points:**
- GPT-4V demonstrates human-like ability to interpret social scenes and emotions, according to a new study.
- Researchers tested GPT-4V's understanding of social cues like laughter, touch, and dominance/submission.
- The AI model was able to correctly interpret 85% of visual features and 95% of video features.
- GPT-4V's ability to interpret social contexts is attributed to its advanced reasoning and multi-modal understanding capabilities.
- The study suggests that AI models are becoming increasingly sophisticated in their ability to understand and interact with human social dynamics.
- This advancement could have significant implications for human-AI interaction, virtual companionship, and the development of more empathetic AI systems.

![Screenshot at 21:43: A split screen showing a human subject on the left and an AI model's interpretation of the same scene on the right, highlighting the AI's ability to discern social cues.](https://ss.rapidrecap.app/screens/yMfktG1U-ak/00-21-43.png)

**Context:** A recent study published on PsyPost explores the capabilities of GPT-4V, a new AI model developed by OpenAI. The research investigates whether AI can truly understand complex human social interactions and emotions. Researchers tested GPT-4V's ability to interpret a range of social cues, including laughter, touch, dominance, submission, and even subtle emotional states, using a dataset of images and videos.

## Detailed Analysis

A recent study has found that GPT-4V, a new AI model, can interpret human social scenes and emotions with remarkable, human-like accuracy. Researchers from the University of Turkey tested GPT-4V's ability to understand various social cues, such as laughter, touch, dominance, and submission, across static images and videos. They found that the AI model performed exceptionally well, correctly interpreting 85% of visual features and 95% of video features. The study attributes GPT-4V's advanced capabilities to its sophisticated reasoning and multi-modal understanding, allowing it to process and connect different types of information—visual, textual, and contextual. This breakthrough suggests that AI is becoming increasingly adept at navigating complex human social dynamics, potentially leading to more sophisticated human-AI interactions and applications in areas like virtual companionship and emotional support.

### AI Capabilities

- GPT-4V interprets social scenes and emotions with human-like accuracy
- Deciphers cues like laughter, touch, dominance, submission, and emotional states
- Achieves high accuracy rates: 85% for visual features, 95% for video features

### Underlying Technology

- Advanced reasoning and multi-modal understanding enable GPT-4V's performance
- Integrates visual, textual, and contextual information for comprehensive scene analysis

### Implications

- Advancements in AI's social understanding could revolutionize human-AI interaction
- Potential applications in virtual companionship and AI-driven emotional support systems
- Raises questions about the future of AI in social contexts and its impact on human relationships

![Screenshot at 19:32: Screenshot of the article title 'AI vision: GPT-4V shows human-like ability to interpret social scenes, study finds', highlighting the core subject of the study.](https://ss.rapidrecap.app/screens/yMfktG1U-ak/00-19-32.png)
![Screenshot at 20:01: An image of a computer chip with 'AI' and a brain icon, representing the AI model being studied.](https://ss.rapidrecap.app/screens/yMfktG1U-ak/00-20-01.png)
![Screenshot at 21:07: A graphic showing a comparison of LLMs and human performance on spatial understanding benchmarks, illustrating the AI's capabilities.](https://ss.rapidrecap.app/screens/yMfktG1U-ak/00-21-07.png)
![Screenshot at 22:27: A screenshot of YouTube comments discussing AI's impact on the internet, showing public perception and concerns about AI content.](https://ss.rapidrecap.app/screens/yMfktG1U-ak/00-22-27.png)
![Screenshot at 23:37: A close-up of a t-shirt with an Abraham Lincoln graphic and the quote "Not everything you read online is true," symbolizing the need for critical evaluation of information, including AI-generated content.](https://ss.rapidrecap.app/screens/yMfktG1U-ak/00-23-37.png)
![Screenshot at 24:43: A graphic illustrating the concept of 'AI reasoning' and its potential to mimic human understanding of social cues.](https://ss.rapidrecap.app/screens/yMfktG1U-ak/00-24-43.png)
![Screenshot at 25:56: A demonstration of how AI tools like chatbots can provide conversational interaction and potentially emotional support.](https://ss.rapidrecap.app/screens/yMfktG1U-ak/00-25-56.png)