top of page
Search
Writer's pictureSeema Chokshi

Will the real AI please stand up

Will the real AI please stand up


More recently the term is AI or Artificial intelligence has been very generously strewn around in almost every product description, expert or industry conference being held. It makes sense to stop for a moment and once again ask the foundational question “What is AI?”

With the recent popularity of ChatGPT almost all professionals are talking about AI in the context of their work and not only that, but they are also in-fact using the tool during the course of the day for personal and professional tasks of varying complexity.

While researching on the perceptions/ implications of AI adoption in organizational and managerial contexts, I encountered varying definitions and interpretations of AI by authors, researchers, bloggers, and industry specialists. One statement which is repeatedly used goes on to say that “AI systems are intelligent systems that learn on their own” or another one that quotes “machine learning algorithms learn on their own without being explicitly programmed.”  


In this article I demystify the vagueness of these statements. In my long association with the academia, I encountered numerous such situations where all resources keep repeating the same words without anything as a feeble attempt at unravelling the real meaning of the words. The real reason we will always needs good teachers even though books have been around all through this time ! Read on to know more about what is AI and how does it differ from other technologies.


What is AI and how is it different from other technologies?


One thing is clear that AI systems run on machine learning algorithms. Now, let’s talk about what is the real meaning of the statement “machine learning algorithms can learn on their own” or “perform tasks without explicit instructions.”


While pinning down a clear thought process to understand AI, I came across a very relevant example in the reference cited below*.  An early pioneer of machine learning, Samuel Arthur, used large number of checkers games to teach the computer to play checkers better than himself. It would have been impossible to write a code of something like this as the computer ended up playing the game better than Samuel and other checkers experts. In this example Samuel instructed the program to identify good moves against the bad ones based on a guide to play checkers but not the exact moves themselves. This is exactly the intuition behind the statement “Machine learning algorithms can learn and improve their performance based on data examples without being explicitly programmed”.   


While, I am not going to elaborate on the different types of machine learning algorithms in this write up, one thing is clear that we can classify the ML algorithms as non-deterministic systems, in other words the output of these systems cannot be predetermined exactly and with complete accuracy. So it’s clear based on the above perspective that an automation system that simply embodies a set of sequential steps to perform a task doesn’t classify as a type of AI system. However, an automation system could be an AI system if it automates the decisioning of a ML algorithm.  


This brings me to the different representations of AI. While all AI systems run on ML algorithms, they can have different embodiment types.  As a physical robot, a virtual agent or bot and as embedded intelligence which cannot be seen by users but is a part of another system.


Physical robots are programmed to perceive their environment and often have sensors to perform tasks. Examples include robots that work alongside humans in an assembly line. A virtual bot could be like a customer service chatbot or a virtual assistant like Alexa or Siri. Embedded systems are embedded algorithms that run behind websites, smartphones that help to enhance the functionality of these interfaces.

The ChatGpt which is being used by millions of non-tech users is based on deep learning models which are a type of machine learning models and is physically embodied as a virtual chatbot.


*Human Trust in AI: Review of empirical research by Glikson et. Al.

2 views0 comments

Commentaires


bottom of page