Artificial Intelligence definition
John McCarthy coined the first term for AI in the mid 20th century at Dartmouth conference.
“Every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it. An attempt will be made to find how to make machines use language, form abstractions, and concepts, solve kinds of problems now reserved for humans, and improve themselves.” – John McCarthy, pioneer researcher in AI
Download latest questions asked on Artificial Intelligence in top MNC's ?
An AI system is a machine that is designed to accomplish the task which humans perform with their natural intelligence. This comes with learning which involves garnering the rules and information for using the data. It also comes with self-correction and reasoning which involves following the rules to gain appropriate conclusions. You ought to know that even the most advanced systems of AI like neural networks were conceptualized way back in 1950-60s itself. But a major reason why it has got a major boost nowadays is due to the advent of big data.Neural networks and Machine Learning systems require huge processing and storage requirements which big data fulfilled effectively. Storage and computing costs have also reduced significantly. Development in AI owes its tribute to this factor also.
AI in the sense can be said as embodied and disembodied. An embodied AI is something that is connected to the world.From this kind of AI sense and react response can be derived. A robot typically matches with this kind of AI. Disembodied AI is where there is no sense and response operations on part of AI system. The input can be sourced from the web with the results being put out in computer screens.
AI can also be classified into strong AI and weak AI. Let’s see how they differ under various parameters.
|Strong AI||Weak AI|
|Definition||The system incepted to simulate human reasoning by AI systems is called as strong AI.||Weak AI is just where the systems are designed to accomplish a specific task.|
|Design||Building such a model is a Herculean task||Designing such a system is relatively easy|
|Functionality||Computer program stores the algorithm||Manual entry of tasks has to be done|
|Research undertaken||Intense research is underway to make this happen.||There are systems which accomplish this functionality but research is underway to create better systems.|
|Examples||There is no system created which can reach that level of sophistication or competence.||IBM’s Deep Blue, autonomous cars etc|
Watch this insightful video differentiating between strong AI and weak AI:
There is also another distinction of AI called narrow AI and general AI. Narrow AI is where the system is meant for specific tasks. General AI is where the systems are designed to provide apt reasoning capability.
The branches of AI: Machine Learning and Deep Learning
Machine Learning – The basic requirement of a Machine Learning system is to take a couple of input examples and then come up with outcomes based on statistical analysis. There are mainly two types of Machine Learning algorithms namely supervised and unsupervised algorithms. Supervised algorithms require humans to provide them sufficient input,desired output along with constant feedback so that expected output can be derived. Unsupervised algorithms don’t need to be trained with desired output and use Deep Learning to accomplish their tasks.
Watch this Matlab video which clearly differentiates between Machine Learning and Deep Learning:
Deep Learning – Deep Learning is a sub branch of Machine Learning where the learning methodology of humans is emulated to gain certain types of knowledge in order for the system to achieve its desired output. Compared to traditional Machine Learning algorithms which are linear, Deep Learning algorithms often work through multiple layers of complexity and abstraction. In Machine Learning the process of learning actually happens in a supervised way through a process known as feature extraction. The system’s ability to detect if a particular picture is a cat or not directly corresponds to how the programmer programs the system. Deep Learning is that training system where the system learns all this on its own without any supervision by the programmer. This system is not only accurate but is also faster.
“I believe that at the end of the century the use of words and general educated opinion will have altered so much that one will be able to speak of machines thinking without expecting to be contradicted.” – Alan Turing, Computing machinery and intelligence Specialist
Applications of Artificial Intelligence
Video games – AI has been used in games since the earliest of times when video technology gained ground. The complexity and uniqueness in behavior of AI characters has increased over the years. AI characters are adept at responding while learning the player’s behavior and respond to stimuli accordingly. Games like Far Cry use FSA (Finite State Automaton) where each AI character either attacks, runs away from the player characters based on the finite state model in which it is programmed. Monte Carlo Search Tree algorithms are also used in gaming. The AI would calculate all its possible moves. It would then calculate all the moves which the opponent would make in return and then it would place its move accordingly. Recently there has been a huge craze over Augmented Reality games like Pokémon Go.
Healthcare – Companies are using Machine Learning to do better diagnosis than humans. IBM Watson is an AI technology which can answer questions when asked while understanding natural language. Chatbots in AI is an online computer program that is used to answer questions to inquisitive customers. This is useful to patients who want to set up appointments or want help in billing process. Virtual health assistants give essential medical feedback.
“A year spent on Artificial Intelligence is enough to make one believe in God”– Alan Perlis, Computer Scientist and professor at Purdue university
Education – Teachers can have more time as automatic grading by AI is there to help. Grading essay is another use case of AI. Each essay graded can be added to the database which can be used as a reference for grading future essays. AI can help teachers be more effective in teaching. It can assess the tests given to a class and identify those parts of the curriculum where the majority of the students are giving wrong answers. Teachers can concentrate on that part more and dispel the doubts of the students. If the learning is given as a game then students engage it with more vigor and hence this increases the retention rates of the students. More importantly the learning experience is made more fun. AI can tweak the educational experience for the differently-abled so that they can better learn the course content as per their capabilities. AI after carefully taking into consideration of the student habits can design study schedule for them.
Natural Language Processing – It is the processing of human language by the computer. Spam detection is one example where it is used. It involves looking to the subject line and text of an email address and determining whether if it’s a fraud or not. Human language is complex considering the ambiguity and the complexity in the linguistic structure like slang, context, and regional dialects and so on. Sentiment analysis is another major use case of NLP and AI. Plethora of comments on social media are analyzed by a company to determine as how exactly their brand is performing.
The Google AI neural network translation tool is one of the pioneering applications of AI. As a part of its work it was designed to translate English to Korean and from Korean to English. It was also trained to translate English to Japanese and from Japanese to English. The AI system learnt all this on its own and designed and was able to design a translation system through which it could easily translate Japanese to Korean and from Korean to Japanese. It is to be noted that the designers of the system had only trained it with English. The fact that the AI translation system was on its own able to translate without English left them baffled.
You will be astounded by the potential and current applications of AI through this video:
Fraud detection – You might have atleast once in your life received a message from your bank asking to report whether you had made a particular transaction or not on the credit or debit card. The bank systems deluge the AI system with both fraudulent and non-fraudulent transactions. The AI system would then learn which are fraudulent and which are not based on the huge training set it is given.
Autonomous cars – When the blind Steve Mahan traveled miles together on Google’s autonomous car the entire AI community hailed this feat. Even Tesla autopilot feature is very impressive in itself. Do you know if the human dozes into sleep the Tesla car would automatically slow the car down if it’s in autopilot mode? Washington Post reported that Google had developed an algorithm which lets the self-driving cars learn to drive on their own through experience just as humans would do. Much research is underway in this field as such cars are not yet completely autonomous as in Tesla case. Even though Google drove miles together with this technology it was near California where every traffic density and all such relevant information had been provided to the computer of the car.
Music and movie recommendations – Did you know that Mark Zuckerberg had created Synapse, a music player which suggested songs which the user was likely to listen? Netflix, Spotify and Pandora also recommends music and movies based on your past interests and past purchases. They accomplish this by garnering the choices you had made and inputting them into a learning algorithm.
AI autopilot in commercial flights – The pilot only needs to put the system in autopilot mode and then majority of the flight will be taken care of by the AI itself. It is reported by New York Times that only seven minutes of human intervention is required for the average flight of a Boeing plane. That pertains only for takeoff and landing.
There is a growing fear that widespread implementation of AI will erode human jobs. Not just commoners but entrepreneurs like Elon Musk are voicing alert at the growing pace of AI research. They view that AI systems may give place to large scale violence in the world. But that is a very myopic way of looking at things. In recent decades technology has grown rapidly and massively. For every job lost to technology there are many other vacancies. If it had been the case where a new technology will take all our jobs then majority of the world would have been jobless. Similar to Elon Musk, even Internet during its inception garnered much negative reviews. It is now obvious that internet just can’t be replaced. You wouldn’t be able to read this blog if that had been the case. Similarly AI will rise through its potential and goodwill and benefit the mankind in general.
Dive deep into the world of Deep Learning AI through Intellipaat’s course on Deep Learning!
- Find out how an application developer masters big data analysis
- Frequently Asked Question from clients planning to adopt Hadoop
- From a Confused Beginner to a Gleeful Hadoop Ninja: Ramya’s Journey