Demo the cBook NOW!

De-cryptifying Artificial Intelligence for eLearning

May 25, 2020

We hear Artificial Intelligence a lot in connection with modern everyday technology. For most of us of a certain age, we think about it in terms of the sci-fi genre (Skynet!), or in terms of video game enemy intelligence (or the utter lack of it before Half-Life came around). In the first part of my deep dive into understanding the importance of AI in eLearning and corporate training, we look at some basic terms that make up artificial intelligence. 

To read about how AI is transforming the Learning and Development landscape, catch Part II of this article here. You can also download a handy guide covering these aspects in a gist from here.


What is Artificial Intelligence?

If a machine can analyse information, factor in real-world variables, apply reasoning, and then draw a conclusion to a certain query, it can be categorised as AI.

In simplified terms, the ability of any system to use human-like logic to think and perform tasks can be categorised as AI. It can be further broken down into Strong AI & Weak AI.

Strong AI

Strong AI can solve general, undefined problems as well or even better than what a human being can. Most people would associate their understanding of AI with this, and is the primary contributor to the misunderstanding of scope and potential, suspicions and general apathy towards the use of AI.

Skynet and Replicants fall under Strong AI, and they are exactly that, science fiction. In reality, Strong AI will only be possible after at least 20-30 years’ worth of development.

Weak AI

Currently, Weak AI is where we are at, enabling applications such as Siri, image recognition, recommendation systems, chatbots, humanoid robotics, gaming and others. It refers to systems that solve specifically defined problems by using fixed methods. They gain experience and become more skilled by way of introducing new data.

Actually, AI, or Weak AI, can be broken down into subsets like Machine Learning and Deep Learning.

What is Machine Learning?

While Artificial Intelligence is used as a blanket term for what the technology aspires to be, Machine Learning is grounded in reality. It can be defined as a subset of AI which allows a machine to automatically learn from available data. The idea of ‘learning’ in Machine Learning is that the machine will continually look for patterns in vast amounts of data, and then deduce accurate results based on what the defined tasks are. If one introduces more data to the same system, it will analyse, or ‘learn’, from the new data and update its conclusions to reflect the most accurate result.

Example

Lets say we have an algorithm based on a sales funnel. We feed a vast database of contacts to this algorithm, with detailed information in addition to the regular contact properties, like industries, career position, income and so on. We then feed in the contacts that actually moved to being Sales Qualified into the algorithm. We can then let the algorithm figure out on its own the success rate based on the contact profiles that turned to leads, and voila, we have ourselves a better strategy to sell our product in the future.

It utilises pre-defined algorithms, analyses vast amounts of data and works within the scope that it has been designed for, to provide the most accurate solutions.

For public consumption though, marketers use the terms ‘Machine Learning’ & ‘Artificial Intelligence’ interchangeably.

What is Deep Learning?

Deep Learning, in turn, is a subset of Machine Learning. Whenever we discuss autonomous cars, the kind of AI that defeats Starcraft professionals, or speech recognition, we are talking about Deep learning.

 In simplified terms, Deep learning is the addition of neural networks, layering and interconnectivity (just like the make-up of the human brain) to a set of algorithms to process more complex, human-like behaviour.

Examples

When we are presented a collection of image of a creature, say a cat dressed in drag, dressed as an alien and in one dressed in fur. Its easy for us to spot the imposter as a cat, but how does the machine come to the same conclusion?

As an example in language processing, widely in use in the form of voice assistants and chatbots, say we have an incomplete sentence like thus:

I like Indian cuisine. Lately I have experimented with Japanese and Italian too, though I keep going back to ________.

For us, it’s a safe bet that the answer is probably Indian, or some dish related to Indian cuisine. How does the machine know though?

To answer that, one must consider the seemingly instantaneous process humans use to arrive at that conclusion – The context of the image or the sentence, the fact that it is an animal or gastronomy that is being discussed; the features of a cat’s face, the relevance of the words preceding the blank space are some of them.

The aforementioned ‘neural networks’ which are integral to deep learning, help the machine understand this process.


Anuj Vyas is a member of Learn Tomorrow. cBook.AI aims to create a new learning experience providing a personal learning feed, which smartly selects learning content on the basis of Key Performance Indicators (KPIs) and artificial intelligence.

Want more information ? 

Join our Newsletter.
Learn Tomorrow GmbH
Taubstummengasse 7/3
Vienna: A-1040 

call us:  +43 1 312 555  
write us:  office@cbook360.com

ATU 75141437 |  HG Wien - FN 525562 k

© Copyright 2020 - Learn Tomorrow - All Rights Reserved
linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram
Share via
Copy link
Powered by Social Snap