OpenAI executives hint that artificial general intelligence may emerge within the next decade, and it will be a highly complex intelligent system

OpenAI's CTO Mira Murati expressed the following main points in this conversation:

1. Regarding collaboration with Apple:

   • This is an important opportunity to bring ChatGPT to all Apple device users
   • OpenAI will not store Apple users' data or IP addresses
   • User data will not be used to train models unless explicitly permitted

2. Regarding data sources:

   • Mainly from publicly accessible data, partners, and paid labeling
   • Exploring different business models and compensation methods with publishers
   • Emphasizing information accuracy and news value

3. Regarding the relationship between data and models:

   • Models learn by inputting large amounts of data, rather than being pre-programmed
   • Increasing data volume and computing power will continuously improve model capabilities
   • Significant work is required to process and filter data

4. Regarding transparency:

   • Publicly sharing model specification documents, showcasing model behavior and decision-making processes
   • Hoping users feel confident and engaged when using the technology

5. Regarding AI risks:

   • More work on safe deployment is needed to gain trust
   • Encouraging people to deeply engage with AI interactions, understanding its potential and limitations
   • Forming a shared responsibility between development teams and users

Overall, Murati emphasized OpenAI's efforts in data usage, model development, and transparency, while calling for greater public participation in and understanding of AI technology.