2024 Volume 8 Issue 1 Pages 33-46
Since 2020, AI has gained renewed attention due to the development of generative AI, particularly large-scale language models (LLMs). Initially designed to improve efficiency, AI is now expected to enhance creativity by combining existing knowledge to create new value. While AI cannot generate entirely new ideas from scratch, it may eventually evolve to make Nobel Prize-level discoveries. However, caution is necessary to prevent AI from becoming uncontrollable. The shift from tool-based AI to autonomous general-purpose AI is underway, where AI can understand situations and make independent decisions without human instructions.
At the same time, the Internet is flooded with vast amounts of information, including false data generated by AI, making it difficult for people to access reliable information. Solutions like originator profiles have been proposed to clarify information sources, but human cognitive limitations make it hard to process non-physical information, unlike physical objects with tangible attributes. Before the Internet, people could understand information based on context, but today, they are overwhelmed, leading to misunderstandings and misinformation.
Autonomous AI is expected to address this by collecting and selecting relevant information to foster innovation while maintaining diversity. Trust in AI's decisions is crucial, and explainability—how AI arrives at decisions—is key. The development of causal reasoning techniques will further enhance trust. Though it may take time for people to fully accept AI decisions, trust will grow through successful experiences. Eventually, delegating complex problems to AI will become commonplace. This delegation is based on trust, not domination, and in a future where humans and AI coexist, AI will become a valuable partner in enhancing human creativity.
