The Power of Language

Natural language processing (NLP), the branch of artificial intelligence or AI is what gives computers the ability to understand text and spoken words in much the same way human beings can. That is, NLP drives computer programs that translate text from one language to another, respond to spoken commands, summarise large volumes of text rapidly, as exemplified by spam detection, Google Translate and Chatbots. NPL also has the ability to convert text to an image using text-to-image generating AI models such as ChapGPT and Stable Diffusion. The power of NLP, however, should come as no surprise - after all, language models have been around for decades. Indeed, natural language is the main means of communication, between humans, between humans and computers, and even between computers. Moreover, the human brain is good at pattern recognition or making connections between seemingly unrelated things and this ability is boosted by AI, and just by using words (text data or text prompts). AI, then, is a transformative tool, indeed, an ideation tool. An open question remains though: are complex AI models (machine learning) truly doing something new or just getting really good at statistics? Well, since machine learning uses “statistical techniques” it can easily be construed as rebranded statistics. But the way statistics is used by statisticians is different than the way it is used by the machine learning community. That is, and according to US statistician Leo Breiman, statisticians use data modelling whereas machine learning practitioners use algorithmic modelling*. Both models can be used to understand data and make predictions. But machine learning lets nature, data and trial-and-error speak about the function that drives inputs to outputs in a complex system: "Let the data do the talking". In contrast, statisticians believe they can guess about this mechanism using best practices making upfront assumptions about the process that generated the data.*  https://projecteuclid.org/journals/statistical-science/volume-16/issue-3/Statistical-Modeling--The-Two-Cultures-with-comments-and-a/10.1214/ss/1009213726.full