Tag
OpenAI describes using GPT-4 for content moderation by enabling policy experts to develop and refine content policies in hours rather than months through an iterative process of comparing GPT-4 judgments against human labels. The approach reduces manual moderation burden while keeping humans in the loop for complex cases and bias monitoring.
Viable, an AI-powered platform founded in 2020, uses fine-tuned GPT-4 models to analyze large-scale unstructured qualitative data like customer feedback and support tickets with greater accuracy than traditional summarization tools. The platform helps businesses extract actionable insights for improving NPS scores, reducing support costs, and informing product decisions through seamless integrations with tools like Zendesk and Intercom.
OpenAI announces function calling capability for GPT-4 and GPT-3.5-turbo models, allowing developers to describe functions via JSON Schema and have models intelligently choose to output structured JSON for external tool integration. The update also extends support for older model versions until June 2024 and improves model evaluation methodology.
OpenAI proposes using language models (GPT-4) to automatically generate and score explanations for neurons in language models, open-sourcing datasets and tools covering all 307,200 neurons in GPT-2. The work demonstrates iterative and scalable approaches to mechanistic interpretability, though explanation quality still lags behind humans.
OpenAI outlines its comprehensive approach to AI safety, emphasizing rigorous testing, iterative deployment, real-world monitoring, and regulatory engagement to ensure powerful AI systems are built and used safely.
Duolingo launches two new GPT-4-powered features in a premium subscription tier called Duolingo Max: Role Play for AI conversation practice and Explain my Answer for contextual grammar feedback. This addresses key gaps in language learning by enabling free-flowing conversations and accurate error explanations beyond what GPT-3 could reliably deliver.
Iceland has partnered with OpenAI to use GPT-4 to preserve and promote the Icelandic language in the digital age, addressing concerns about linguistic extinction while creating resources that could benefit other low-resource languages globally.
Stripe deployed GPT-4 across multiple workflows including support customization, developer documentation assistance, and fraud/bad-actor detection in Discord communities, finding the model outperformed human reviewers in several tasks. The company ran an internal hackathon-style initiative with 100 employees to identify and prototype AI-powered features using OpenAI's latest models.
Khan Academy is partnering with OpenAI to integrate GPT-4 into its educational platform to enhance student learning through deeper conceptual understanding and personalized tutoring, while also helping teachers create customized instructional materials and track student progress.
OpenAI releases GPT-4, a large multimodal model that accepts image and text inputs and demonstrates human-level performance on professional and academic benchmarks, significantly outperforming GPT-3.5 across various evaluation metrics.
OpenAI has partnered with Be My Eyes to use GPT-4 for visual accessibility, enabling blind and low-vision users to navigate websites, e-commerce platforms, and physical spaces through intelligent summarization and real-time guidance. The system leverages GPT-4's vision capabilities to identify important content and provide contextual assistance that mimics how sighted users naturally scan information.