Artificial Intelligence seems more like marketing jargon these days. Every company, startup, or business you know now promotes its products and services with the term ‘AI-powered’ as its USP.
True to this, artificial intelligence sure seems to be inevitable nowadays. If you notice, almost everything you have around you is powered by AI. From the recommendation engines on Netflix and algorithms in dating apps to some of the most complex entities in the healthcare sector that help in oncology, artificial intelligence is at the fulcrum of everything today.
Read more: AI vs. Machine Learning vs. Data Science
But how well do we know about AI? We’ve been using the term in our conversations but is our understanding of artificial intelligence just basic?
Turns out that it mostly is. That’s why we decided to explore a little further and shed some light on less touched upon topics. In this post, we will understand the and shatter our conventional beliefs and understanding of AI being the same across all use cases and real-world examples.
At the outset, it might appear that AI is just a single entity and universal across all of its applications. However, only when you deep-dive into the topic that you will understand there are specialized wings in AI that are responsible for the execution of specific tasks, concepts, and delivery of corresponding results. Let’s start with machine learning.
Probably the immediate thought that strikes our mind when we think of AI is machine learning. For the uninitiated, is a subset of AI that is responsible for making machines learn concepts, algorithms, and conditions autonomously and perform specific actions. When there are vast volumes of datasets to be processed and analyzed, machine learning modules enter the picture and take charge of detecting patterns from datasets, preparing inferences and visualizations, and more.
More complex than machine learning, deep learning is a subset of machine learning. To give you a better idea, imagine a Russian doll. The outermost doll that you can visualize is AI and the subsequent dolls nested within each other are your machine learning and deep learning concepts respectively.
The reason we quoted deep learning to be more complex than machine learning is because it works on a purpose that is far more intricate. It intends to mimic the behavior, emotions, and responses of humans and sort of instilling them in machines. It aims at replicating the working and functioning of the human brain – through artificial neural networks – and simulates how the brain would attempt to solve presented problems.
To do this, deep learning deploys concepts like Natural Language Processing (NLP), text, audio files, and more to learn human actions (and reactions).
If you’ve interacted with a chatbot, you would know that the efficiency at which you get responses from machines comes from the concept called natural language processing. Even consider your virtual assistant – Siri or Alexa – for instance. They are quirky, super-fast, precise and of late getting quite sarcastic as well.
All these advancements are because of the evolution in NLP and this is a kind of AI that teaches machines to talk and respond like humans. It could be as simple as speech-to-text or as complex as chatbots you interact with on fintech and healthcare portals. For accurate responses, massive volumes of data are fed into NLP modules. Both deep learning and machine learning concepts and algorithms are deployed for this as well.
Computer vision is the visual counterpart of NLP. While the latter is all about text and notes, the former is all about videos, images, and all forms of visual references including computer imaging.
Computer vision is the branch of AI that lets machines identify, understand and process visuals and classify them for diverse purposes. For example, if you use Google Lens and point it at a product or a label, it would accurately pull out information on what it is.
Even a reverse image search on search engines is the result of computer vision. By blending concepts like deep learning and machine learning, computer vision is unlocking new levels in the way machines ‘look’ at the world and comprehend information.
Explainable AI is a fairly new concept that is gaining prominence because of several concerns associated with the decisions machines make, the rationality behind it, and more. To give you a simple idea of what explainable AI is, visualize that you’re driving in an avenue, heading to your workplace like usual. Suddenly, your GPS or navigation system asks you to take a detour and drive a mile extra and reach your destination.
Now, what’s the reason behind AI changing the route suddenly? Is it because it wanted you to save 10 minutes despite driving a mile extra or is it because of factors unknown? Explainable AI tries to answer such instances but on a more complex and intricate level.
|AI Type||Use Cases|
|Machine Learning||Ideal for risk and fraud analysis, navigational systems, segmented advertising campaigns, predictive analytics, and more.|
|Deep Learning||Recommendation engines, virtual assistants, preventive maintenance modules, processing data from multiple sensors and computer vision modules in autonomous vehicles, workflow optimization, and more.|
|NLP||Voice search, speech-to-text, dictation, chatbots or conversational bots, interfaceless devices, extracting data from unstructured data sources such as emails, social media feeds and more.|
|Computer Vision||In healthcare to detect minute tumours, oncology, in self-driving cars for their autopilot modules, applications like Google Lens, smart home automation systems, facial recognition systems and more.|
|Explainable AI||Predominantly deployed in DARPA, manufacturing, fintech and automotive devices.|
We believe this extensive post on the different types of AI was quite eye-opening. As the month’s pass, we’re sure there will be newer branches responsible for super-niche purposes and tasks.
is a serial entrepreneur with more than 20 years of experience in software and services. He is the CEO and co-founder of Shaip, which enables the on-demand scaling of our platform, processes, and people for companies with the most demanding machine learning and artificial intelligence initiatives.