What is Artificial Intelligence and How Does it Work? History | Types | Benefits | Risks
Is artificial intelligence the future? Find out all about the definition, types, uses, advantages and disadvantages of AI technology here!
When posed with the term ‘artificial intelligence’, we might be forgiven if our minds wander to autonomous robots bent on the destruction of humankind, like in The Terminator. To those of you who came bearing such expectations, it might be a tad disappointing that AI has not accurately made it that far. Nonetheless, the concept is one that is ingenious and worthy of a thorough discussion.
What is Artificial Intelligence?
Definitionally, artificial intelligence refers to the simulation of human intelligence within machines, providing them with the ability to replicate human thought patterns. This means that any machine that is capable of simulating problem-solving in a human-like fashion is considered to be artificially intelligent.
Whether it’s Siri trying to tell you what the weather forecast is, or Netflix trying to predict shows that you would like, AI technology can be found nearly everywhere.
How Does Artificial Intelligence Work?
The innumerable examples of AI around us beg the question: how does artificial intelligence work?
In a nutshell, artificial intelligence functions because humans incorporate sets of instructions (or algorithms) to machines which then ‘learn’ from data to solve problems. This is the basis of machine learning, which is the principle behind teaching machines how to ‘think’, and the foundation of AI. It refers to a series of IF-THEN statements (which loosely function according to the principle that if ‘x’ happens, then execute ‘y’) that the computing device adheres to as a set of strict rules.
Artificial intelligence in medicine, for example, involves the prescribing of medication to patients based on data regarding a wide variety of symptoms (and their variations) that the programme has been ‘trained’ to recognize. AI has gone so far as to be able to diagnose diseases even before symptoms have appeared.
Fundamentally simple, right? Well, AI technology is extremely sophisticated.
Deep learning, which is a growing branch of artificial intelligence, involves the use of ‘neural networks’ which are given extremely large volumes of data to make an accurate repository of data which can be used for a variety of different purposes. Deep learning is a subset of machine learning but is distinct in that it requires no human intervention in its learning process, and forms algorithms that can validate the truth of predictions on its own.
AI Through the Ages
Though artificial intelligence technology has only come to fruition in the last few years, the technology behind its modern form has been in the works for nearly a century.
Mid-20th Century:
Alan Turing and John Von Neumann both contributed to the technology that converted the decimal computing system (which took into account numbers between 0 to 9) to a more comprehensive binary computing system (which functioned based on ON/OFF signals represented by 1s and 0s). Binary number systems allowed for the use of conditional logic in computing, which serves as the building block for artificial intelligence.
Formal AI, however, was founded by John McCarthy in 1956 at Dartmouth College, where he coined the term “AI” as well as its formal definition.
In 1965, DENDRAL was created by MIT. Soon after, MYCIN was invented by Stanford in 1972. These two machines used simple IF-THEN statements, based on Turing and Neumann’s work. They functioned using inference engines, which became the basis for the creation of expert systems soon after.
Late 20th and 21st Century:
Fast-forward a few decades into the era of expert systems: a system that simulates a human expert in terms of decision-making and problem-solving.
1997 showed a breakthrough in AI development: an expert system created by IBM (which went by the name of “Deep Blue”) defeated former international chess champion, Garry Kasparov, proving the lengths that artificial intelligence research had reached by then. However, this machine was still very simple, as it did not think for itself, but instead was ‘brute-forced’ into winning by having all the possible outcomes of the chess game written into its programme.
The 21st century saw the uses of artificial intelligence expand to unimaginable heights, as an era that not only had extremely powerful computers but one that has nearly limitless access to massive volumes of data.
ASIMO, a humanoid robot that was made to interact with people and assimilate into society, was made in 2000 by Honda. Simultaneously, MIT also launched Kismet, which was a robot that was created to interact with people and experience some semblance of ‘emotions’.
The use of artificial intelligence in business was now being envisioned. Companies had realized the potential of artificial intelligence as a product and began investing extensively on AI projects. Siri, an artificially intelligent virtual help assistant, was launched by Apple on the iOS store in 2011. Siri used deep learning algorithms to function: using voice recognition technology to try to interpret the words that are spoken into it while attempting to ‘understand’ them by cross-referencing queries on its ever-expanding neural network.
Siri was discovered to be a profitable venture for Apple, contributing to its revenue stream through affiliate links and leads. Similar technology, such as Alexa, soon followed suit.
What are the types of Artificial Intelligence?
As you’ve probably been able to gauge by now, there isn’t just one type of artificial intelligence. So, what do they look like? We will explore a few of them below. Some of them are currently in production, whereas the others are either work-in-progress or purely theoretical.
Reactive Machines:
Perhaps the simplest form of artificial intelligence, reactive machines function based on intuitive functionality. That is to say that they do not respond based on memory-based functionality, and hence, only operate within a select combination of inputs. These machines, therefore, do not possess the capacity to ‘learn’ the same way that other artificially intelligent machines do.
Prominent examples of this technology are Deep Blue and Kismet, which possess the limitation of not being able to improve with ‘practice’, as they are unable to store the experiences of functions that they have already performed.
Limited Memory:
Limited memory machines, in addition to possessing the functionality of reactive machines, can also store past experiences for a limited amount of time. These machines are ‘trained’ with large volumes of data and can refer to it when encountered with an unfamiliar situation.
Limited memory machines allow for a glimpse into the full extent to the advantages of artificial intelligence. A popular example of this technology is used in self-driving cars, where the cars store multiple pieces of information simultaneously for short periods, such as traffic signals, pedestrian movement, and the flow of traffic.
Tesla has taken advantage of this technology to become one of the most valuable car companies in the market.
Theory of Mind:
Though still in production, the theory of mind technology attempts to make massive feats for artificial intelligence in the future. Researchers aspire to create a computer that can simulate – to a reasonable accuracy – the human mind. This technology will also attempt to emulate emotional intelligence and how humans interact with entities and environments around them.
Self-awareness:
The ‘final form’ of AI is the creation of a self-aware machine. In this, an artificially intelligent entity is envisioned to recognize its individual needs, wants, and desires alongside understanding the inherent need to self-preserve. This technology currently exists only in theory, but companies are currently looking to work on production as soon as the resources are available.
What are the Benefits and Risks of AI?
Whether we’re talking about Baymax or the Agents in the Matrix: the advantages and disadvantages of artificial intelligence are aplenty. We’re going to dive into each of them and shed light on what they entail.
What are the advantages of artificial intelligence?
Little to No Mistakes:
Amongst the numerous benefits of artificial intelligence, arguably the most prominent rest with the fact that as AI develops, they will make no errors due to the near-perfect precision that these machines are capable of. “Human errors” that are caused by negligence, tiredness, or a lack of focus will cease to exist in a world where machines do the tasks for us instead. The decisions that are reached are never done based on limited information or pressure, but rather, on a set number of well-thought algorithms.
This would be particularly beneficial for the health sector, where artificial intelligence in healthcare will reduce the risks posed by dangerous surgeries, wrongful prescriptions of medicine, and slow-to-respond healthcare services such as ambulances.
Takes on Risky Jobs:
If there were artificially intelligent machines back then, did you know that the damage of the Chernobyl disaster could have been limited? This is because AI machines could have gone near the core of the power plant without suffering from the resulting radiation, unlike humans. Similarly, AI could take on several tasks that humans would find dangerous and are unwilling to do due to the risk of human life involved. This could involve tasks such as bomb-defusal or space exploration.
Works Longer and Faster:
AI machines will not have the backaches that the average 9 to 5 worker has by the end of the day, nor will they take lunch or bathroom breaks. They will be available to work non-stop without the concern of complacency or physical strain.
AI machines will also be able to compute decisions much faster than humans with advanced, completely undistracted processing. In the distant future, the need for human laborer's may cease to exist entirely. In their current state, they are able to streamline the work of people by assisting them do tasks, make computations, and solve problems several times faster than the average worker.
Hence, AI – despite being a machine that learns and improves nearly continuously – does not get tired, and therefore is the optimal worker.
What are the disadvantages of AI?
High Initial Costs:
Despite the cost-cutting nature of artificial intelligence in the long-run, high-quality AI comes with high upfront costs. These costs include installation and maintenance costs.
If the AI is to work with other employees, then there also need to be training costs for the remainder of the staff. Therefore, not only does artificial intelligence pose the threat of high initial costs, but also of consistent costs throughout its usage, which means constant money down-the-drain for businesses.
This becomes particularly difficult for SMEs to achieve competitiveness in their industries because they simply do not possess the same level of capital that large corporate giants do.
Unemployment:
Though a workplace where salaries do not need to be paid is a dream for a cost-cutting CEO, this will come at the expense of several labourers who will become liabilities in the case that an AI machine ends up replacing them. This will particularly harm minimum wage workers, where tasks are repetitive and easy to perform by non-human entities.
Lack of “Human Thinking”:
Despite the efficiency with which Artificially Intelligent machines operate, they still cannot replicate human rationale or emotions – at least, not within the next century. Given present technology, AI still might fail when faced with a situation that is too foreign for them to comprehend, or even making decisions that could potentially put people’s lives at risk – and the development in technology that will resolve this is still at least decades away. A common thought experiment cited for the latter is the self-driving car trolly dilemma.
These issues had massive consequences, such as when Facebook had to shut down its artificial intelligence chatbot programme in 2017 because it ‘invented its language’ and started communicating in gibberish.
[cta_blog_inner]
Real-World Examples of Artificial Intelligence
When all is said and done, the question remains: how can artificial intelligence help businesses grow? There are several prominent examples of the use of artificial intelligence in businesses.
Artificial Intelligence in Banking
Banks have grown to recognize the importance and benefit of employing AI for their needs. Currently, the banking sector employs AI in preventing fraud and analyzing contracts. This is an area that otherwise requires humans a considerable amount of time to detect, cross-reference, and report.
JP Morgan and Chase are currently using Contract Intelligence (COiN), which is an AI system that reads through and interprets essential contracts, which has saved nearly 360,000 hours for the company annually.
AI services like Teradata can detect suspicious transactions in real-time and can check them against legitimate transactions, and the banker’s past transactions, which several companies have employed so far.
Artificial Intelligence in Education
In the education sector, AI is primarily used to streamline administrative tasks and gauge student performance. It is used to grade tests, mark attendance, check homework and perform other menial tasks that would take away from meaningful teaching time.
Currently, AI is being researched to assist students outside the classroom with assistive tutoring and specialized learning that identifies and assesses individual students’ needs and weaknesses.
Carnegie Learning has developed MathiaU, which is an artificially intelligent system that assists college-level students with coaching in Mathematics.
Artificial Intelligence in Finance
Besides banks, AI can assist other financial institutions such as insurance firms and investment banks to vast degrees. Using natural language processing, investment research could be done through the conducting of sentiment analysis on stocks.
Predictive programmes may also be used to gauge the performance of the stock market in the future, where the data about the health of other stocks, the economy in general, and past data is inputted to predict accurate future results.
Kavout’s “K-Score” is an example of the use of an AI programme to derive “equity rating scores” to ease the process of buying and selling in the stock market.
What is the future of Artificial Intelligence?
The question on many people’s minds is: how far can artificial intelligence take us – is it humanity’s ultimate friend or our eventual foe?
AI is currently attempting to achieve a perfect text-to-speech synthesis, unlimited access to data, and commercial-grade fully autonomous self-driving cars – with progress accelerating at unprecedented rates.
According to Pew Research, the coming 12 years will enable massive developments in “saving lives”: in both healthcare and policing. In a world where Airbnb is developing technology that will almost instantly create products out of their designs, humanity is looking to make massive strides in AI production.
Though we are still a long way from actual artificial consciousness, the fact that it is a very real possibility is a very intriguing prospect. Will AI make humans entirely obsolete? After all, they’re predicted to do our jobs better than we possibly can. Or, will they become our most trusted companions, making our lives easier with each passing development?
Whichever school of thought you might ascribe to; one thing is for certain: humankind will continue to develop artificial intelligence until it has either achieved perfection, or Skynet finishes it for us.