Analysis: Saturating accuracy could limit the potential of NLP models
Explore the rise of AI language models and their market saturation. See why multimodal models may take the lead and what it means for chipmakers, cloud providers, and enterprises.
Popular AI applications, such as customer service, involve understanding the language and generating the most appropriate response to user questions. Such applications have gained popularity in recent years and are fueling the need for language models with the best possible accuracy. Our analysis of publicly available data from the HuggingFace (HF) website indicates that over two-thirds of the models on HF are related to natural language processing (NLP) and text. We feel that the language models have reached saturation in terms of accuracy. In time, different models, such as multimodal models, will become dominant. Chip manufacturers, cloud companies, and enterprises should pay close attention to this trend.
