OpenAI Forms Safety and Security Committee, Begins Training Next-Generation AI Model With Aim to ...

Artificial intelligence research organization OpenAI unveiled the establishment of a Safety and Security Committee by its Board of directors. This Committee will oversee and guide the direction of AI development concerning safety and security measures.

The Safety and Security Committee’s initial responsibility will involve assessing and refining OpenAI’s procedures and protective measures over the next 90 days. Upon the completion of this period, the Committee will present its recommendations to the entire Board. Following the Board’s evaluation, OpenAI will release a public update regarding the endorsed recommendations, maintaining a focus on safety and security protocols.

In the announcement, OpenAI also outlined that the company has initiated training for its upcoming frontier model, signaling progress towards achieving “next-level capabilities” en route to artificial general intelligence (AGI), suggesting the potential launch of “GPT-5.”

However, the forthcoming iteration of large-scale models might not be accessible to the public until the following year. Typically, training AI models is a time-consuming process, often spanning several months or even years, and necessitates extensive fine-tuning by the development team before public release.

Additionally, as OpenAI advances in its development of AI systems approaching human-level intelligence, the organization has adopted a more cautious approach in its discussions regarding AGI. As Anna Makanju, the Vice President of Global Affairs at OpenAI, highlighted in a recent interview, OpenAI’s objective is to develop AGI capable of performing tasks at the current human cognitive level. She also stressed that OpenAI’s aim is focused on developing AGI rather than superintelligence, which would significantly surpass human intelligence.

OpenAI Introduces Enhanced GPT-4o, Elevating Real-Time Interaction Capabilities

OpenAI introduced GPT-4 in March 2023, followed by a gradual rollout of various text, image, audio, and video generation applications based on this model.

Recently, the company unveiled the latest iteration, GPT-4o, boasting improved interaction capabilities that mimic real-person conversations in real time. However, despite enhancements, GPT-4o remains within the scope of GPT-4’s capabilities, albeit slightly stronger across all aspects. 

The post OpenAI Forms Safety And Security Committee, Begins Training Next-Generation AI Model With Aim To Achieve ‘Next-Level Capabilities’ Of AGI appeared first on Metaverse Post.