A quick revision of complex language models

To oversimplify, a complex language model is a type of language model designed to express and interpret information and knowledge on complex problems related to arithmetic and data that the human mind fails to grasp. This is inclusive of complicated mathematical structures or patterns in computer science or coding.

While complex language models are one of the most used and praised for solving unimaginable problems, they still lag behind in mimicking the human mind in every sense of it.

The basic problems that are not paid heed to

The artificial intelligence driven models and machines are made so prone to get acclimatized with complexities that basic problems are now unconsciously overlooked.

The AI and the language models can detect cancer but fail to differentiate between a cat and a dog. Artificial neural networks find their heavy usage in video gaming played by professionals but fail to deal with simple alphabets.

The problem lies in the algorithm. This is what experts have been highlighting. Not being able to deal with the basics is also counted as the unknown problems today which are very much in existence, leading to inequality in the field of artificial intelligence.

In a book named ' Algorithms are not enough' by Herbert Roiblat, highlights the same issues. He says how the complex language models fail to replicate the general intelligence and is still confined within its narrow domains.

It can hence be inferred that the AI algorithms continue to be evasive and the basic problems which are equally important like the complex ones, still remain in the dark chambers of our minds and AI.

Creating general artificial intelligence to bell the cat: Moravec's paradox

Hans Moravec, a computer scientist, states that computers outshine humans in problem solving. Even the most complicated ones can be interpreted and solved in seconds by computers but the machines start to stagger in emulating simple human skills and problems that a 5-year old can solve. This is named "Moravec's paradox".

This happens because the most complex of problems are structured and in most cases, patterned. But there are problems, which cannot be structured that falls under general intelligence. Complex language models fail to stray beyond a consistent set of rules that are programmed into them.

However, everything that we imagine is real. And so is general intelligence.

A point to note here is that the problem is not with the efforts of the scientists that make artificial intelligence and language structures to evolve. The problem here is that of unawareness.

First thing first. The present need of the hour is to identify the basic and unknown problems that are prevalent and the problems an ordinary person struggles with in everyday life. Consequently, the next step would be to design algorithms that can imitate the general human intelligence even if they are not structured or bound in a set of stringent rules.

The solution is flexibility. Language models should be customised to be substantial and impartial. Implementing so, it can blur the inequality and inequity the society battles with because artificial intelligence cannot be subjected to class constraints.

If language models are able to detect and interpret the movements of terrorists and rain clouds, it should also be able to help a housewife with her cooking and households chores.