Recently, I focused on finishing the Beijing University of Chinese Medicine Management Institute Summer Class—Practical Application of Medical Artificial Intelligence Analysis. I also watched some tutorials on Bilibili. Regarding artificial intelligence, I learned a lot of foundational knowledge and got a general understanding of some high-level terms like large models, multimodal, attention mechanism, etc. It gave me a feeling that the future will be revolutionary (I won’t explain these terms here; please check out the AI tutorials I’ve been watching recently; those with a basic foundation should be able to understand).
Previously, my understanding of these advanced technologies was very superficial. I used GPT to help write code and Midjourney to create images, but the results were average, and I felt these tools weren’t that impressive. But when I really started to understand their underlying mechanisms, it’s amazing! Progress is rapid!
From attention to transformer, a new trend in AI is emerging: unification, even towards general artificial intelligence. Previously, tasks with time-series characteristics like audio, text, and video were completed using RNN and its variants, while tasks without time-series properties like image recognition and classification were done with CNN and its variants. Now, transformers and their variants are used to handle all these, elevating the level of information understanding to a new dimension. Moreover, I came up with a direction that I haven’t seen implemented yet: correspondence between speech and robot actions.
Future productivity will inevitably experience a new explosion because of this. New waves will inevitably bring new opportunities. Increase your abilities, broaden your horizons, and I wish everyone to ride the waves successfully!
Today’s post is brief because it contains many terms, and explaining each one in detail would require a lengthy article. Considering that the audience of this public account doesn’t have much interest in programming, I won’t go into details. As for those interested in artificial intelligence, start implementing the details early and make them work for you, instead of just watching media hype!