Technology
The Realm of Artificial Intelligence
The Entirety of its Resources at your Disposal.
Artificial Intelligence has become a part of our everyday vocabulary rather than a technological advancement of late. We are all familiar with smartphones, smart sensors in cars, smart locking systems, etc. AI-assisted home appliances have expanded to washing machines, refrigerators, health trackers, fans, and lights! What does AI actually deal with, and how does it work?
Introduction to AI:
AI refers to human-like intelligence shown by a computer, robot, or any other machine or the machines’ ability to mimic human activities like learning from examples, recognizing objects, understanding language, solving problems, and more. The workflow involves acquiring large chunks of data and processing it with fast, iterative algorithms allowing the software to learn from patterns in the data. Research in AI has focused chiefly on intelligence components: learning, reasoning, problem-solving, perception, and language. AI is further divided into
- Narrow AI —designed to perform specific tasks within a domain like language translations, face recognition, touch sensor
- General AI — a hypothetical concept which deals with cognitive abilities, emotions, and thought processes in machines
Narrow AI:
This division of AI is an application of artificial intelligence systems programmed to handle a specific task. It operates within a predefined range of functions that it is designed to perform. It is also termed weak AI because it lacks genuine intelligence like humans and consciousness. Various modes of machine intelligence that we are familiar with using narrow AI include:
- Image processing and facial recognition systems:
This deals with a high-level understanding of digital images or facial features, which work using a pattern or numerical representation of the dimensions of an element (image) stored in a database and compared for its match.
We encounter image processing in google image search or online shopping sites. It has advanced features to recognize any picture taken with the camera and browse the database for related products. Facial recognition systems are used in workplaces, security features in devices for locking and unlocking, and law enforcement agencies to crack down on criminals.
- Conversation assistants and chatbots:
These assistants come across when we open any organizational websites, banking, and transactions and are used to respond to queries, assist customers, and quickly come up with results as in search engines by checking through a vast database.
- Speech recognition and translation services:
The primary examples of voice and speech recognition include voice assistants like Google Voice Assistant, Alexa by Amazon, Siri by Apple, Cortana by Microsoft, and controlling smart home appliances.
Initially, these assistants communicate with users and convert the phrases or words to text or machine language and perform the specified task. In the case of translation services, the software uses machine translation, and each word is translated to the specified language.
- Recommendation engines:
This feature is often encountered in Google Search for similar results, movie suggestions in OTT platforms, and recommended products on online shopping sites. It works with algorithms that filter objects based on users’ preferences and compares them with other customers’ choices.
- Autonomous vehicles:
Self-driving cars and autonomous vehicles are all set to top the trend in several countries soon. The striking features include radar sensing, navigation systems, cruise control, etc. Companies like Mercedes-Benz, Nissan, and Tesla have started their venture in this field, and Google launched a project, Waymo, that facilitates research to achieve full automation.
General AI:
General AI (or) strong AI is a vast domain currently under research and mainly deals with the machine possessing cognitive abilities and intelligence to solve problems with greater efficacy than humans. And the foremost task we have in hand to establish strong AI is that we should be capable of modelling and replicating the functions and reasoning ability of the human brain. Some notable achievements that reach par with general AI include supercomputer K and Fugaku by RIKEN, a scientific research institute in Japan, summit supercomputer by IBM, and Param Siddhi-AI, owned by the government of India, to name a few.
How does AI work?
The algorithms and models used in narrow AI are programmed using machine learning. It allows the computer instructions to learn from the data instead of step-to-step instructions given by the programmer. In simple words, it is to train the data to learn algorithms. The algorithms then generate a new set of rules based on the data. This generates another algorithm, formally referred to as the machine learning model. We can use the learning algorithm for different models.
For example, a computer program code for chess problems includes randomly generated moves until a mate is found. The program then stores the solution with the positions obtained so that the next time the computer encounters the same position, it would recall the previous solution. This simple memorization of individual items and procedures, known as rote learning, is relatively easy to implement on a computer. But implementing the domain of generalization is much more difficult as it involves applying experience to analogous new situations.
Machines learn by
- Supervised learning — Giving labelled data to get desired output (Eg. Identifying rules to classify pictures)
- Unsupervised learning — Unlabelled data in the input used to identify patterns (Eg. E-commerce websites discovering similar items bought together)
- Reinforcement learning — Interacting with a dynamic environment with feedback and penalties (Eg. Self-driving cars abiding by traffic rules)
Other than machine learning, other methods include
- Cognitive computing — simulate human-like interaction by machines
- Natural language processing — ability to analyze, understand and generate speech
- Computer vision — pattern recognition and understanding images
- Deep learning — using massive interconnected processing units to learn patterns
- Internet of Things (IoT) — generate a large amount of data from connected units
- Application programming interfaces (APIs) — add AI functionality to software packages
and more fields that are untapped or under research, that are being added daily.
What can AI do?
We’ve been associated with AI for quite a few years now through our regular activities but haven’t properly acknowledged our dependence on it. Various fields like
- interpreting video feeds from drones,
- ensuring the quality of structures like oil pipelines by visual inspection,
- organizing personal and business calendars,
- responding to simple customer-service queries,
- plagiarism checkers,
- combining data from other intelligent systems and enabling related search results like hotel booking at a suitable time and location,
- diagnosing devices and X-rays,
- flagging inappropriate content online,
- monitoring data gathered by IoT devices related to wear and tear in elevators
- e-commerce and internet shopping,
- 3D mapping through satellite imagery,
and much more path-breaking inventions and technology, which we use daily, functions with the principles of AI.
Artificial perception and its response to stimuli have advanced to realistic activities like identifying individuals through optical sensors, automatic or autonomous vehicles on the road, and robots performing cleaning and maintenance tasks in offices, etc. One of the earliest systems which experimented on perception and action was FREDDY, a stationary robot with a moving television eye and a pincer hand, constructed under Donald Michie, at the University of Edinburgh, Scotland, during the period 1969–71. FREDDY recognised various objects and assembled simple artefacts, such as a toy car, from a random heap of components.
Myths revolving around AI:
We are quite familiar with the term AI and robots owing to the exposure from movies and showbiz. Moviemakers and writers have experimented in this field over the years through various perspectives; the advancement of AI in the field of space travel in Interstellar (2014), highly advanced AI machines designed for destruction in Terminator (1984), robots and personal assistants developing emotions and conscience in Ex Machina (2014), Her (2013), Bicentennial man (1999), I, Robot (2004), the most loved childrens’ movie Wall-E (2008) depicting an environment-friendly robot, our very own attempt in Endhiran (2010), which taps into the field of robotics and AI-powered devices and several movies that worked out brilliantly in theatres as well as taking a realistic approach (and some others that didn’t pose their best shots).
In most movies, there is the deceptive notion of all AI-powered machines developing feelings and reasoning, taking decisions on their own, and turning against humankind. The fundamental flaw with this notion is that automated machines and other devices are programmed by engineers and researchers using AI with a specific set of instructions to solve or decipher any task or problem posed and obtain solutions. They cannot act on their own unless programmed so. Moreover, the functioning of the human brain and intellect is an intriguing domain that needs extensive research and experiments to be consolidated as an algorithm. Hence, programming machines to mimic human nature is bound to be a strenuous task.
Effects of AI on society:
We experience several benefits of using AI as it motivates research in areas like economics, law, advanced technology, security, and control. Chatbots in websites and services are available 24/7 for queries and responses, provided that they have reliable accuracy through data analysis.
But AI and increased dependence on it by humans has a negative impact on society. It helps you carry out exactly what you asked for and is unaware of what you actually wanted. It could be turned into an indestructible weapon for warfare if misused by anti-social elements and can cause loss of life and property to a disturbing extent.
It doesn’t make sense to go behind controversial myths supporting or criticising our dependence on AI, but research on safety and privacy certainly plays a vital part. It must not be left out of its leash while we explore its valuable resources, and it should be made a point that we have control over them and use them only for the betterment of the living standard of people.
