Hardware manufacturers are doubling down on Edge Computing this year, with PC and laptop makers also launching various new products. Personally, I believe this is just the beginning; this industry trend will have various impacts on humanity at multiple levels. I am most optimistic about Apple, which has an advantage in hardware. As long as it integrates AI well in its software, there is still much it can achieve. As an iPhone and Mac user, I still look forward to Apple launching eye-catching products or services this year.
AI model 在進入2024年後也是各大巨頭各顯神通，Meta/Google/OpenAI/... 。對一個新 model 可能都還沒導入，且還在評估或測試階段，model 就有有新的進步，新版本又推出了。至於算力，那更是大家都拚了命在追加，但永遠不夠用，OpenAI 的 Sora (文字生成影片）一出，算力的需求又要再加大，之前在使用文字生成圖片時就感到算力吃緊。
現在在生成影片的需求下，看來永遠不夠用了。
As we entered 2024, major tech giants like Meta, Google, OpenAI, and others have been showcasing their prowess with AI models. When a new model is still not fully implemented, being in the evaluation or testing phase, it often happens that the model sees new improvements, and a new version is released. As for computing power, everyone is desperately trying to increase it, but it's never enough. With the launch of OpenAI's Sora (text-to-video), the demand for computing power has escalated even further. Previously, when using text-to-image generation, the strain on computing resources was already felt. Now, with the demand for video generation, it seems like it will never be sufficient.
AI model 在記憶體使用的優化及model 本身優化也是有些進展，這也是今年值得期待的事。
There have been some advancements in optimizing memory usage and improving the models themselves, which are also something worth looking forward to this year.
生成式AI 的提示工程是未來使用AI 很重要的技能之一，未來AI 普及後，世界上可能只有兩種人，一種是會使用AI，另一種是不會使用AI的人。
One of the crucial skills for future AI usage is prompt engineering for generative AI. With the proliferation of AI in the future, there may only be two types of people in the world: those who can effectively use AI and those who cannot.
