2024:
LLM → LAM
words → action
At CES, the world's largest ICT convergence exhibition held last January, rabbit, an independent AI hardware developed by founder Jesse Lyu, garnered significant attention. In a LinkedIn post, Lyu hinted at a shift in how humans interact with AI, stating "2024: LLM → LAM, words → action." The device, with its vibrant orange square design, compact size, 360-degree rotating camera, and 2.88-inch touch screen, might initially be mistaken for a new smartphone. However, its voice command capabilities and portability allow for a more proactive transfer of human intent to computers in daily life. While the news of its initial orders selling out is noteworthy, a more significant aspect is the emergence of a device that replaces the smartphone, which we've come to believe must be constantly in hand. This implies that a change in the interaction between humans and computing, which has been built upon apps, is underway.
The interactions and task management portrayed in Iron Man's AI assistant, Jarvis, simultaneously instilled anticipation for future technological advancements and a realization of its current limitations. In the real world, it's fair to say that apps have shaped smartphones. The technological visions of IoT (Internet of Things) startups and large corporations, often filled with promises of a human-centric future, have consistently relied on smartphones as the connection point for diverse technological devices both inside and outside the home. However, this cold and transactional relationship based on boundaries has not delivered or fostered human-like communication or expectations. In this regard, the attention and anticipation surrounding rabbit suggest that corporate AI visions will become even more critical. Furthermore, as the presence of apps becomes less necessary, it hints at a shift toward a more conversational, human-like style of interaction between humans and computers.
Rachel Woods, a former Facebook data scientist and publisher of the AI-focused newsletter The AI Exchange, argues that companies need to understand how AI remembers user interactions to maximize its utility. She introduces a framework of short-term memory, long-term memory, and knowledge, suggesting that AI's memory systems will increasingly resemble human memory.
Just as we only need to remember a phone number long enough to dial it, AI's short-term memory refers to remembering ongoing chat conversations or information gleaned during a task. The industry aims to expand this capability for more information and context retrieval. Unlike short-term memory, which resets with each new conversation, AI's long-term memory provides a starting point within a predefined context, allowing for persistence over time. Similar to how we might recall memories by reviewing our personal journals or blog posts on a specific topic, OpenAI's custom GPT, instructions, and memory features all relate to this long-term memory. Finally, AI's knowledge function, which grants access to specific information or data that AI can use to complete tasks, is analogous to a well-organized Google Drive – a family photo album of memories or a collection of expert papers. This represents the core of maximizing specific knowledge, providing a structure for easy retrieval and utilization of the most relevant information.
unsplash: Priscilla Du Preez
Humans have lived in cities for over 6,000 years. Within these complex organizational structures, conversation has been an essential skill, maintaining a sense of the world and fostering empathy within relationships. Even when addressing our pets – a dog or cat waiting at home after work – conversation serves as a confirmation of our mutual existence, the beginning and process of emotional exchange. The day when interacting with AI becomes a more human-like experience is drawing closer. The evolution from smartphones to AI devices like rabbit signifies that AI is becoming increasingly intertwined with the fabric of human conversation. As technology progresses, the goal should not just be to mimic human conversation but to cultivate a form of communication that acknowledges the complexity of human emotions and the physicality of our existence. This journey towards a more human-like AI should focus on enhancing our ability to connect, empathize, and thrive in a world where the boundaries between the physical and digital realms continue to blur.
References
Comments0