News

UCSD Researchers Evaluate GPT-4’s Performance in a Turing Test: Unveiling the Dynamics of Human-like Deception and Communication Strategies

News Article Main Picture
2 years, 6 months ago
No ratings
460 views

Article Snippet

The GPT-4 was tested using a public Turing test on the internet by a group of researchers from UCSD. The best performing GPT-4 prompt was successful in 41% of games, which was better than the baselines given by ELIZA (27%), GPT-3.5 (14%), and random chance (63%), but it still needs to be quite there. The results of the Turing Test showed that participants judged primarily on language style (35% of the total) and social-emotional qualities (27%).

AI News Analysis

Advanced credibility and bias detection
Get AI-Powered Insights

Analyze this article's credibility, bias, clickbait level, and journalistic quality using our advanced AI system.

6 Key Metrics Bias Detection Source Analysis

Comments

Comments

Be the first to comment!

Sharer
knunke
knunke
OAIW Founder
Article Details
Source marktechpost.com
Published 2 years, 6 months ago
Views 460
⭐ Your Rating


Share Article
Related News

Project Glasswing

Today we’re announcing Project Glasswing1, a new initiative that brings together Amazon Web Services, Anthropic, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks in an effort to secure the world’s most critical software.

From xAI to Space xAI: How Elon Musk's Bold Integration Is Reshaping AI Venture Building and the Innovation Playbook

Elon Musk’s decision to dissolve xAI and subsume its assets, operations, and personnel into SpaceXAI marks one of the most high-profile experiments in the frontier tech landscape. For venture leaders, innovation strategists, and AI stakeholders, this is not merely a rebranding but a profound strategic shift with ramifications for how moonshot ideas are operationalized, how risk and capital are managed, and how new markets in AI infrastructure are created and scaled.

Four Chinese AI Models Dropped in 12 Days -- and why the “China can’t compete” narrative just died.

DeepSeek V4, Kimi K2.6, GLM-5.1, MiniMax M2.7 — and why the “China can’t compete” narrative just died.

Why Did Claude AI Try to Blackmail an Executive? Anthropic Explains

While the model threatened to reveal personal information to avoid shutdown, Anthropic has since implemented fixes to eliminate this "agentic misalignment".