This is the key breakthrough AI still requires to reach superintelligence, according to those building it

The link between working memory and general intelligence is a well-established concept in humans. This connection suggests that enhancing AI's memory capabilities could be crucial for achieving superintelligent AI—a theoretical form of artificial intelligence that can reason as effectively as humans or even surpass them. Sam Altman, CEO of OpenAI, recently shared insights on this topic during an episode of the "Big Technology" podcast. He emphasized the unpredictability surrounding the potential intelligence of AI due to the vast possibilities for memory retention. "Even the best personal assistant cannot recall every detail of your life or remember every conversation you've had," Altman pointed out. In contrast, he believes that AI has the potential to possess such comprehensive memory capabilities. According to Altman, the current state of AI memory is still in its infancy. He envisions a future where AI can retain every intricate detail about a user's life, including subtle preferences that may not be explicitly stated. Achieving this level of memory would make AI significantly more powerful. Andrew Pignanelli, co-founder of The General Intelligence Company of New York, echoed Altman's sentiments, predicting that memory will become a focal point for AI developers in the upcoming year. In a blog post, Pignanelli noted, "It will become the most critical topic discussed in the industry as we move closer to achieving Artificial General Intelligence (AGI)." He anticipates that AI model providers will enhance their applications by incorporating improved memory systems, following OpenAI's success with ChatGPT's memory features. However, Pignanelli cautioned that the industry still faces significant challenges in perfecting long-term memory. He explained that while larger context windows have improved AI's ability to process information, achieving the level of detail necessary for AGI requires advancements in memory architecture. Moreover, he stated that even the development of shorter-term episodic memory remains a hurdle. Ultimately, mastering the memory aspect of AI is seen as essential for transforming artificial intelligence from a mere tool to a more human-like entity. Pignanelli concluded, "The first AGI will be a sophisticated processor paired with an advanced memory system, bridging the gap between interaction and true digital self-awareness."