Artificial intelligence capabilities and inabilities in software solutions


more and more articles and advertisement for startups and applications stating that they “use AI”, have “AI capabilities” or similar “AI *” claims. In most of those cases that applications are using machine learning functionalities and models and do not have anything related to artificial intelligence in them.

Machine learning is not artificial intelligence, machine learning is one of building blocks on the path of developing artificial intelligence. Models and techniques that are based on machine learning and implemented in some software solution do not mean that that software has any relation with artificial intelligence.

I understand that it is easy to fool board member or investors on what is AI but it is really getting ridiculous on what is today labeled as “AI software solution”. Even Google search understand that this is misused so it packs in result machine learning together with artificial intelligence software solutions.


“Machine learning is the science (and art) of programming computers so they can learn from data,” writes Aurélien Géron in Hands-on Machine Learning with Scikit-Learn and TensorFlow.

Machine learning focuses on “teaching” computers how to create algorithms that learn from existing data and make predictions of future data. Therefore, machine learning requires careful preparation of lots of data. This is not technique where you will throw tons of data and it will miraculously find some sense in it and draw the conclusion.

Simple implementation of machine learning is your spam filter that you use on daily bases and one of more complex implementations is dealing with is ability to recognize objects in an image (computer vision). This task is something that all humans perform with ease. However, for computer to obtain that information there are several things that need to be taken into account as image resolution and how much of the object is visible on image. There is essential difference how images are processed by computer and by intelligence.

Partial images are something computer have problems dealing with and reason is simple, it lays in difference how computer and human algorithms work.

" algorithms adopt a "bottom-up" approach that moves from simple features to complex ones. Human brains, on the other hand, work in "bottom-up" and "top-down" modes simultaneously, by comparing the elements in an image to a sort of model stored in their memory banks." Weizmann Institute of Science. "Less than meets the eye: How do computers -- or our brains -- recognize images?

Weizmann Institute of Science. "Less than meets the eye: How do computers -- or our brains -- recognize images? How do computers -- or our brains -- recognize images?." ScienceDaily. ScienceDaily, 10 March 2016

Brown University. "Key weakness in modern computer vision systems identified." ScienceDaily. ScienceDaily, 30 July 2018. 


“Artificial Intelligence (AI) is the part of computer science concerned with designing intelligent computer systems, that is, systems that exhibit characteristics we associate with intelligence in human behaviour – understanding language, learning, reasoning, solving problems, and so on.” (Barr & Feigenbaum, 1981)

Reason for association of different techniques with AI is base on not understanding that AI research also overlaps with tasks such as robotics, control systems, scheduling, data mining, logistics, speech recognition, facial recognition and many others. It overlaps but all of those tasks are not therefore AI and all of applications are not AI enabled since they have such purpose.

Different types of AI:

  • Reactive AI - These systems can not go back to their past and learn from it. Systems based on this type of AI neither have a concept of the past, historic data and inferences from it nor the ability to understand a future.

  • Limited Memory AI - These systems can go back to their past for a limited period of time and learn from it. This concept is beautifully incorporated in self-driving cars, where the sensors detect instances of a pedestrian crossing, bad road conditions, weather, incoming vehicle, lane detection, traffic lights and more to make smarter driving suggestions (still not decisions).

  • Theory of Mind AI - This is more on the books and science fiction movies, where we talk about artificially intelligent and advanced systems that can perceive the concept of emotions in people and change their behavior accordingly

  • Self-Aware AI - This type of artificial intelligence is still a probability and has the potential for real application. In this type of artificial intelligence, machines or robots are aware of who they are, understand their internal traits, states and conditions and even perceive human emotions.

To achieve last one scientist are still trying to understand how human brain functions. In order to make efficient AI you must be able to mimic human brain. Brain synapse is one of the fundamental building blocks of our nervous system and creating artificial synapse capable of simulating a fundamental function of our nervous system. Therefore this is important task on our path of achieving AI.

American Chemical Society. "Hacking the human brain: Lab-made synapses for artificial intelligence." ScienceDaily. ScienceDaily, 28 June 2017.