The author suggests that while the current use of AI may seem mundane, it is a normal part of the adoption process for new technologies, drawing parallels with the early days of cinema, cars, and personal computers. The author believes that while we need new infrastructure, tools, and experience to fully realize the potential of LLMs, the current "boring" applications of Sober AI are valuable and worth our attention. Examples given include generating SQL statements from natural language input, which can turn enterprise data into a powerful tool.
Key takeaways:
- The field of Sober AI, which focuses on practical and realistic applications of AI, is gaining traction and becoming the default in the industry.
- Most of the work with Language Models (LLMs) is mundane and involves data acquisition, cleansing, preparation, synthesis, staging, evaluation of model results, and debugging.
- As AI continues to evolve, it is increasingly seen as just another function in a pipeline or program, rather than a magical solution.
- Despite the hype around AI, the relatively boring applications of Sober AI, such as generating SQL statements from natural language input, are valuable and worth our time and attention.