OpenAI announced thatScale AI, a new American artificial intelligence startupThe two companies will work together to provide fine-tuning capabilities for their artificial intelligence models, allowing customers to fine-tune content according to application service requirements when applying artificial intelligence models such as GPT-3.5 or GPT-4, thereby ensuring that the application services will not affect privacy or interactive content.
Because many AI models are pre-trained, they may still produce inappropriate results for subsequent applications, such as providing erroneous information or leaking internal data. Unless the AI model is custom-designed and trained, with the ability to fine-tune specific details, most users who use open-source or publicly available AI models often face some inappropriate use cases.
OpenAI's collaboration with Scale AI is to use the latter's technology to fine-tune Open AI's previously released GPT-3.5 Turbo artificial intelligence model, allowing users to customize artificial intelligence models according to their needs and improve the accuracy of customized model operations through Scale AI's enterprise-level artificial intelligence technology and data engine.
On the other hand, the fine-tuning function of its artificial intelligence model also symbolizes that OpenAI hopes to create a more responsible artificial intelligence model application experience.
In addition to the current fine-tuning for GPT-3.5 Turbo, Scale AI's collaboration with Open AI also includes GPT-4 fine-tuning capabilities, which are expected to be officially launched this fall.
