Are large language models actually smart like humans?
Large language models (LLMs) can give the appearance that they are very human like in our interactions with them.  This is because they can respond to our answers the way a human would respond.  Large language models under the hood are very good at pattern recognition.  They are training on vasts amounts of data from the internet and have billions of machine learning parameters that can be tuned.  These models can even be customized or trained on a set of data that is industry specific to be able to provide industry specific answers to industry specific questions.  They work by taking the user prompt and converting it into a pattern, and then it looks across all the different patterns the LLM has been trained on and when it finds a match it returns it to the end user.  It is important to note that large language models at this moment are not humans, do not exhibit human behavior, and are incapable of feeling human emotion and making decisions based on this.