OpenAI Plans to Integrate Speech, Images, and Videos into GPT: Sam Altman

In a recent podcast featuring Microsoft co-founder Bill Gates and OpenAI CEO Sam Altman, the two tech leaders explored the intricacies of artificial intelligence (AI) development, focusing on the rapid growth of the technology and the potential advancements in OpenAI's Generative Pretrained Transformer (GPT) models.

Altman highlighted the current limitations of GPT-4, acknowledging that while it can handle a multitude of questions, there is room for improvement in terms of reliability. He emphasized the importance of enhancing reliability to ensure that the model consistently provides the best response out of a pool of potential answers.

Looking ahead, Altman discussed the anticipated incorporation of speech, images, and videos into future iterations of GPT, foreseeing these enhancements as crucial in the next two years. Although OpenAI's current models already support image and audio capabilities, Altman expressed the need for further improvements to create a more comprehensive understanding of the real world beyond textual data.

GPT-4.5 or GPT-5, the next significant update in the Generative Pretrained Transformer series, is expected to bring substantial changes in terms of personalization and customizability. Altman highlighted areas such as understanding user preferences, integrating with calendars and emails, and connecting with external data sources as key aspects that will undergo significant improvement.

Altman also addressed GPT-4's current shortcomings, specifically its limited reasoning abilities. Improving inferencing and reasoning capabilities is a priority for OpenAI, aiming to enhance the model's ability to synthesize data, analyze prompts, and draw insightful conclusions.

These ambitious changes are part of OpenAI's strategic priorities for the next two years, showcasing a commitment to refining and advancing their AI models.

In the conversation, Bill Gates expressed amazement at the evolution of ChatGPT, admitting initial scepticism he shared, “I was privileged to see your work as it evolved, and I was very sceptical. I didn't expect ChatGPT to get so good. It blows my mind.”