Gen AI Hype - the Never Ending Excitement

The author has been following the Generative AI agenda closely and finds it exciting, but also senses frustration with the technology not always delivering. Despite significant investments, the gap between expectations and real utility of Generative AI tech has not closed much in recent years. Influencers and media outlets often share sensational achievements and breakthroughs, but these claims may not translate to real-life scenarios. The author cites examples of inflated expectations, such as the reported performance of the Qwen2.5-Coder 7B model, which they were unable to replicate. They also note that influencers are not journalists and are not expected to verify claims, leading to a lack of skepticism and critical thinking. The author suggests that the constant stream of sensational achievements can be misleading and that it's essential to be cautious when evaluating claims about Generative AI. They also express disappointment with Small Language Models, which they find to be inferior to larger models from OpenAI and Anthropic. The author concludes that LLMs are task-specific and that trial and error is the best way to find the right model for a particular use case.