Challenge Of Artificial Intelligence: Understanding & Using Natural Language

• The AI has made great strides since its first appearance on the scene of the investigation. Artificial intelligence, AI onwards, covers countless fields and has many applications. The most popular is machine learning and within it Deep learning. When someone mentions that it “interacts with artificial intelligence,” they are talking about communicating with a well-trained machine learning algorithm that, with more or less success, answers questions or solves problems.
• Actually, being more precise, one does not interact with an AI system, but with a set of specialized algorithms, each one in a single task. The combined effort of all the individual algorithms is what is perceived as “an AI” that interacts with us and offers us results.
• The road to get here has been long and, mainly, has been hampered by limitations in the processing capacity available to not only train, but also make systems that are common today viable from a “commercial” point of view. Thanks to the increased performance of processors, and particularly to cloud processing, systems that were not viable in the past are today.
• However, there is still a long way to go until AI systems master some “human” aspects that can be widely applied in various areas of interest. Specifically, we talk about processing and using natural language in the same way that a person would. The two branches of AI with the longest track record today are Natural Language Processing (NLP) and Natural Language Understanding (NLU) .
• The applications of the NLP are diverse, but we can list text classification, language modeling, voice recognition (rather, of expressions or ways of speaking), generation of subtitles or captions and captions, translation machines, preparing summaries of documents or answer questions.
• As for NLU, it is actually a subset of NLP that is responsible for machine reading comprehension. It is an IA-complete problem and therefore cannot be solved by simple specific algorithms.
• Specifically, mastering a natural language is extremely complex for an AI for multiple reasons including context. The context is the, according to the RAE, ” physical or situation, political, historical, cultural or any other environment, in which it is considered a fact .” There is another definition of context that is supported by the rules and linguistic environment, but that is easier to master. The daily language that we use, our expressions and conversations have, to a large extent, a context that our interlocutor should know to understand the conversation. Or, you should know how to inquire about the context in which we are situated.
• In addition to the context, we must take into account a multitude of details such as visual and body language, tone of voice, the immediate environment … Very complex tasks that machines will take years to master. The great challenge that Artificial intelligence AI faces is to understand language beyond knowing how to answer simple, concrete and direct questions. To achieve this we would be able to develop real personal assistants who, for example, give us an efficient summary of our daily mail or who are able to give us advice on where an email conversation is going. Real applications capable of understanding, in a given context, what is being talked about and how to respond to the interlocutors.

Also Read: WiFi: How To Improve Your Home WiFi Network