anything that cannot be patternized or modeled using math, cannot be implemented with AI. examples like: emotions, care, love, creativity, strategies out of the box, new theories?
This is surely an interesting topic which blends ideas from philosophy.
While emotions/care/love/creativity seem like abstract concepts, you might be surprised at how creative software developers and data scientists can be when it comes to modelling abstract concepts. All behaviors of AI programs occur within some programmed environment. If the environment causes the AI to act in a way that benefits others, is that care? AI can certainly come up with out of the box strategies. See this video of AIs playing in an environment and their creative strategies within the defined environment.
That means, AI regression has a limit along its own curve, and dataset knowledge only encompasses the past, and cannot invent the future.
Most AI models learn from past data and tries to predict the future, that's true. In many cases, it does succeed if the underlying truths don't change much. Many AI models also are constantly trained on newly acquired data to keep up to date with the present. For example, Google is constantly looking at what people are searching for, and they will train their AI model to predict what you want to search for, even on a daily basis.
You can learn more about AI yourself through AI4E and then we can discuss more concrete examples too.