OpenAI has announced an enhancement to its AI-powered deep research feature for ChatGPT, now enabling it to analyse codebases on GitHub. This new “connector,” slated for release in beta, allows developers using ChatGPT Plus, Pro, and Team subscriptions to ask detailed questions regarding codebases and related engineering documents. Enterprise and educational users will gain access soon.
The GitHub integration signifies a trend among AI companies to improve the functionality of their chatbots by linking them to external services. For instance, competitor Anthropic recently introduced Integrations for its AI chatbot, Claude, showcasing a similar approach. While OpenAI previously offered a plugin capability, it evolved towards custom chatbots dubbed GPTs.
Nate Gonzalez, OpenAI’s Head of Business Products, suggested that user demand for access to internal resources prompted the introduction of this connector. Through this integration, the ChatGPT tool can facilitate various programming tasks including breaking down product specifications, summarising code structures, and offering real code examples for new API implementations.
Despite the advantages, there are inherent risks such as the potential for AI to generate inaccurate information, commonly referred to as “hallucination.” OpenAI emphasizes that this feature is designed to save time for developers rather than replace their expertise. Additionally, the tool will respect organisational privacy settings, ensuring users only access codebases that are explicitly shared with ChatGPT.
OpenAI continues to develop its coding capabilities, having recently launched an open-source tool for terminal coding and enhancing its ChatGPT desktop app to read code across various developer-centric applications. Their focus on programming indicates its importance within their AI model offerings. Notably, OpenAI has also reportedly agreed to purchase Windsurf, an AI coding assistant, for $3 billion.
Furthermore, OpenAI has introduced fine-tuning capabilities for its latest models, allowing developers to customise the o4-mini “reasoning” model using a method called reinforcement fine-tuning, which enhances performance based on task-specific grading. However, fine-tuning for the o4-mini is restricted to verified organisations, while all paying developers can fine-tune the GPT-4.1 nano model.
This shift towards verification for accessing certain models and developer features started in April, with OpenAI asserting that it aims to mitigate abuse by requiring organisations to submit identity verification documents.
Overall, OpenAI’s latest features highlight a significant move towards integrating AI more deeply into the software development process, with an aim to empower developers while maintaining necessary safeguards.
Fanpage: TechArena.au
Watch more about AI – Artificial Intelligence
