#AI

If you ask an #ai scientist if it would be possible to generate an AI that understands and extends research papers, he will appreciate the difficulty of the task since it is his domain. However, there are some who could claim end-to-end workflows are possible in some other areas where they do not have enough knowledge. We often oversimplify the domain with over complexified neural architectures. In most cases, we don't need full automation but better integration, faster and more transparent response, and more iteration in the real world to eliminate unnecessary tasks. It is not about replacing humans, it is allowing people to do better. Here is my joke to illustrate this.