Mathematician Alan Turing changed history once more with a straightforward query: “Can machines think ?” Less than ten years after assisting the Allies in winning World War II by cracking the Nazi encryption device Enigma.
Turing’s 1950 article “Computing Machinery and Intelligence” and the accompanying Turing Test established the fundamental goal and vision of AI.
Fundamentally, the field of computer science known as artificial intelligence (AI) aims to successfully address Turing’s challenge. This endeavour aims to mimic or duplicate human intelligence in machines. The broad goal of AI has generated a lot of debate and interest. In reality, there is no universally accepted definition of the field.
The biggest problem with merely “developing intelligent machines” as an AI goal is that it doesn’t define AI or describe what an intelligent machine is. The interdisciplinary science of artificial intelligence (AI) is approached from many different angles, but developments in machine learning and deep learning are driving a paradigm shift in practically every sector of the computer industry.
A 2019 research study titled “On the Measure of Intelligence” is one example of a new test that has been suggested recently and has received generally positive reviews. In the article, François Chollet, a seasoned expert in deep learning and a Google employee, makes the claim that intelligence is defined as the “pace at which a learner transforms their existing knowledge and experience into new skills at worthwhile activities that include uncertainty and adaptation.” In other words, the most intelligent algorithms are able to predict what will happen in a variety of situations with only a tiny quantity of experience.
In contrast, Stuart Russell and Peter Norvig address the idea of AI by organising their work around the theme of intelligent agents in machines in their book Artificial Intelligence: A Modern Approach. In this light, Artificial Intelligence (AI) is defined as “the study of agents that acquire perceptions from the environment and perform actions.”
FOUR TYPES OF APPROACHES ARE DEFINED AS ARTIFICIAL INTELLIGENCE
(i) Thinking like a human being means modelling thought after the human mind.
(ii) Rational thinking is the imitation of logical cognition.
(iii) Being humane means acting in a way that resembles human conduct.
(iv) Rational behaviour refers to behaviour that is intended to accomplish a specific objective.
The first two ideas deal with how people think and rationalise, whereas the remaining ideas are concerned with how people act.According to Norvig and Russell, “all the skills needed for the Turing Test also allow an agent to act rationally.” They place special emphasis on rational agents that act to achieve the greatest results.
“Algorithms enabled by restrictions, exposed by representations that support models focused at loops that tie thought, perception, and action together,” is how Patrick Winston, a former MIT professor of AI and computer science, characterised AI.
Although these concepts may seem esoteric to the average person, they assist to focus the discipline as a branch of computer science and offer a guide for incorporating ML and other branches of AI into programmes and machines.
The Four Categories of Machine Intelligence
Based on the kinds and levels of difficulty of the tasks a system is capable of performing, AI can be categorised into four categories. Automated spam filtering, for instance, belongs to the most fundamental category of artificial intelligence, while the distant possibility of creating robots that can understand human emotions and thoughts belongs to a completely separate subcategory of AI.
The most fundamental AI principles are followed by a reactive computer, which, as its name suggests, can only use its intellect to see and respond to the environment in front of it. A reactive machine cannot utilise past experiences to inform current decisions since it lacks memory. Because they can only experience the world right away, reactive machines can only carry out a limited number of highly specialised jobs.
However, intentionally limiting the scope of a reactive machine’s worldview means that this kind of AI will be more dependable and trustworthy – it will respond consistently to the same stimuli.
The chess-playing supercomputer Deep Blue, which was created by IBM in the 1990s and defeated Gary Kasparov in a game, is a well-known example of a reactive machine. Deep Blue was only able to recognise the chess pieces on a board, know how each moves according to the game’s rules, acknowledge each piece’s current position, and decide what would be the most logical move at that precise moment. The machine wasn’t striving to better place its own pieces or anticipate prospective movements from the other player. Every turn was perceived as existing independently of any earlier movements and as having its own reality.
Google’s AlphaGo is another illustration of a reactive machine that plays games. Due to its inability to predict moves in the future and reliance on its own neural network to analyse game developments in the present, AlphaGo has an advantage over Deep Blue in more difficult games. In 2016, champion Go player Lee Sedol was defeated by AlphaGo, which has already defeated other top-tier opponents in the game.
Reactive machine AI can achieve a level of complexity and offer dependability when developed to carry out recurring tasks, despite its constrained scope and difficulty in modification.
2. Limited Memory
When gathering information and assessing options, limited memory AI has the capacity to store earlier facts and forecasts, effectively looking back in time for hints on what might happen next. Reactive machines lack the complexity and potential that limited memory AI offers.
An AI environment is developed so that models can be automatically taught and refreshed, or AI is created when a team continuously teaches a model in how to understand and use new data.
The following six actions must be taken when using ML with restricted memory AI: 1.Training data must be created 2.The ML model must be developed,3. be able to generate predictions, 4.be able to accept feedback from humans or the environment, 5.be able to store that feedback as data, and 6.all of these stages must be repeated in a cycle.
3.Theory of Mind
It is only speculative to have a theory of mind. The technological and scientific advancements required to reach this advanced level of AI have not yet been attained.
The idea is founded on the psychological knowledge that one’s own behaviour is influenced by the thoughts and feelings of other living creatures. This would suggest that AI computers would be able to reflect on and decide for themselves how people, animals, and other machines feel and make decisions. Robots ultimately need to be able to understand and interpret the concept of “mind,” the fluctuations of emotions in decision-making, and a litany of other psychological concepts in real time in order to establish two-way communication between people and AI.
4. Self Awareness
The final stage of AI development will be for it to become self-aware after theory of mind has been created, which will likely take a very long time. As conscious as a person, this kind of AI is aware of both its own presence and the presence and emotional states of others in addition to its own. It would be able to comprehend what other people could need based on both what they say to them and how they say it.
AI self-awareness depends on human researchers being able to comprehend the basis of consciousness and then figure out how to reproduce it in machines.
How is AI used ?
DataRobot CEO Jeremy Achin gave the following definition of how AI is used now in his lecture to a crowd at the Japan AI Experience in 2017.
“AI is the ability of a computer system to carry out operations that often require human intelligence… These artificial intelligence systems are frequently powered by machine learning, occasionally by deep learning, and occasionally by really dull stuff like rules.
Based on its capabilities, artificial intelligence can be categorised in three different ways. These are stages through which artificial intelligence (AI) can develop rather than different varieties, and only one of them is currently feasible.
1.Narrow AI, sometimes known as “weak AI,” is a replica of human intellect that only operates in specific contexts. Even while these machines may appear clever, they are functioning under many more restrictions and limits than even the most primitive human intelligence. Narrow AI is frequently focused on executing a single task exceptionally well.
2. Artificial General intelligence (AGI) AGI, often known as “strong AI,” is the type of artificial intelligence (AI) that we see in movies, such as the machines in Westworld or Data in Star Trek: The Next Generation. A machine with general intelligence, or AGI, can use its intelligence to solve any problem, much like a human being.
3.Superintelligence : This will probably mark the apex of AI development. Superintelligent AI will be able to not only mimic but also outperform human intelligence and complex emotion. This could entail forming its own opinions and conclusions, as well as its own ideologies.
Advantages and Disadvantages of Artificial Intelligence
Although AI is undoubtedly seen as a valuable and rapidly developing asset, this young area is not without its drawbacks.
In 2021, the Pew Research Center polled 10,260 Americans about their views on AI. According to the findings, 37% of respondents are more concerned than excited, while 45% of respondents are both excited and concerned. Furthermore, more than 40% of respondents said they believed driverless automobiles will be detrimental to society. Even still, more respondents to the survey (almost 40%) thought it was a good idea to use AI to track the spread of incorrect information on social media.
AI is a blessing for increasing efficiency and productivity while also lowering the possibility of human error. However, there are some drawbacks as well, such as the expense of development and the potential for robots to take over human occupations. It’s important to remember, though, that the artificial intelligence sector has the potential to provide a variety of occupations, some of which haven’t even been imagined yet.
Importance of Artificial Intelligence
AI has a variety of applications, including speeding up vaccine research and automating fraud detection.
According to CB Insights, 2021 witnessed a record-breaking year for AI private market activity, with global funding rising 108% from the previous year. Due to its quick acceptance, artificial intelligence (AI) is creating a stir in a number of businesses.
Business Insider Intelligence found that more than half of financial services companies now use AI technologies for risk management and revenue generation in its 2022 research on AI in banking. The application of AI in banking could result in savings of up to $400 billion.
According to a 2021 World Health Organization study on medicine, despite challenges, integrating AI in the healthcare sector “has tremendous potential” since it might lead to benefits like better health policy and more accurate patient diagnosis.
AI has also impacted the entertainment industry. According to Grand View Research, the global market for AI in media and entertainment would increase from a value of $10.87 billion in 2021 to $99.48 billion by 2030. In that extension, AI applications like detecting plagiarism and creating high-definition visuals are included.
Also Read ,