Imagine teaching a super-smart AI something new without ever directly changing its code. That's the tantalizing possibility explored by researchers who've developed a novel approach to continual learning for large language models (LLMs). Traditional continual learning methods require tweaking the model's internal parameters, which can lead to "catastrophic forgetting" where the AI loses previously acquired knowledge. This new research introduces CLOB (Continual Learning Over Black-box LLMs), a paradigm that leverages the power of prompting alone. The key innovation is CIS (in-context CL via Incremental Summarization). As the LLM encounters new information, CIS creates concise summaries representing different concepts. These summaries act like memory snapshots, allowing the LLM to retain knowledge even as it learns new tasks. Whenever the AI sees more data related to previously learned concepts, CIS cleverly updates the summaries. This approach tackles the limited input size of LLMs by distilling information into manageable summaries. Experiments demonstrated impressive results, with this prompt-based learning method significantly outperforming traditional continual learning techniques. This innovative approach opens doors to more dynamic and adaptable LLMs, capable of continually expanding their knowledge base without losing what they've already learned. While this research primarily focuses on text classification, its implications are far-reaching. Imagine LLMs seamlessly integrating new information, constantly evolving and becoming even more powerful tools for understanding and interacting with the world. However, there are limitations. This method might struggle with extremely long documents that exceed the LLM's input capacity. Additionally, applying this technique to other fields like computer vision, where summaries might not be as easily represented, presents a challenge. Despite these limitations, this research presents a significant step forward in continual learning for LLMs, promising a future where AI can learn and adapt continuously, just like humans.