This article takes inspiration from the Harvard Graduate School of Arts and Sciences and covers the following sections:
1. The dawn of AI (1950s-1980s)
2. Boom (1980s-2000s)
3. Modern (2010s to present)
1. The dawn of AI (1950s-1980s)
Who was Alan Turing and what does he have to do with AI?
Between 1939 and 1945, British mathematician and computer scientist Alan Turing played a key role in breaking Nazi codes during World War II, which were generated by a machine called Enigma that the German army used to encrypt military communications.
AI as we understand it today do not have AI played a key role in breaking Enigma. During World War II, the concept of AI and related fields such as machine learning and deep learning had not yet been developed. However, the Enigma cracking effort involved several important pioneers of modern AI and computer science. The research at Bletchley Park, led by mathematicians such as Alan Turing, laid the foundation for developments in AI and computer science in the decades that followed.
Turing suggested that since humans use available information and reasoning to solve problems and make decisions, why can't machines do the same? This was the logical framework for his 1950 paper “Computing Machinery and Intelligence”, in which he discussed how to build intelligent machines and how to test their intelligence.
Despite writing this groundbreaking paper, Turing was never able to implement his ideas for some reason: on a computer. Computers before 1949 lacked key features that we take for granted today. shop I couldn't execute any commands, Execute That is, if you tell them what to do, they won't remember what you did. Also, in the 1950s, computers were very expensive (about $200,000 per month to lease) and only large corporations could afford them.
In the same 1950 paper, Alan Turing proposed what would become known as the Turing Test, which he initially called the “Imitation Game,” to test a machine's ability to exhibit intelligent behavior that is equal to or indistinguishable from human behavior. This concept has since become a key benchmark in the field of artificial intelligence.
In this test, a human judge converses in natural language with two other parties, one human and one machine posing as a human. Both the human and the machine attempt to convince the judge that they are also human. If, after a period of conversation, the judge cannot reliably distinguish between the machine and the human, the machine is said to pass the Turing test, demonstrating a level of intelligence and conversational ability comparable to that of a human.
The goal of the Turing Test was not necessarily to replicate true human intelligence, but to determine whether a machine could exhibit behavior indistinguishable from humans in a text-only conversation: if a machine could convince humans through a natural language conversation that it was not a machine, then it could be considered to have achieved a degree of intelligence, at least for the purposes of conversation.
To date, no AI system has consistently achieved a success rate of 50% or higher on the blind Turing test, which is the bar that an AI system can claim to have passed with confidence. Despite significant recent progress, human-level artificial general intelligence (AGI) is still elusive. It is worth noting that the Turing test itself has been subject to criticism and debate regarding its validity as a measure of machine intelligence or consciousness. Some argue that passing the Turing test does not necessarily indicate true intelligence or understanding.
Full testing aims to achieve human-level conversational ability in all situations, which remains an ongoing challenge. Progress towards this benchmark will continue with new, more powerful generative models that we will discuss in the coming days.
Logic theorist
An important computer program developed in 1956, about five years after Turing's paper, by Allen Newell, John Shaw, and Herbert Simon, was Logic Theorist. It was created to mimic human problem-solving abilities and is considered by many in the field to be the first AI program ever created. This work was Dartmouth College Artificial Intelligence Summer Research Project The hosts are Marvin Minsky and John McCarthy, the people who actually coined the term artificial intelligence.
Although the event did not achieve its goal of agreeing on a standard approach to AI, it marked the starting point for the next 20 years of AI research.
Between 1957 and 1974, AI flourished as computers became faster, cheaper, and more accessible, with better storage capacity. Additionally, machine learning algorithms improved, and people became better able to choose the right algorithms for their tasks. In 1970, Marvin Minsky expressed an optimistic view in Life magazine, suggesting: At the inner In 3 to 8 years, we will develop machines with the intelligence of an average human.Despite this optimistic outlook and fundamental progress, achieving the ultimate goals of natural language processing, abstract thinking, and self-awareness remained a distant goal.
Having overcome the early challenges of AI, numerous obstacles became apparent, mainly centering on severe limitations in computing power. Computers could not store vast amounts of information or process it quickly. For tasks such as communication, understanding the meaning of large numbers of words and their combinations was essential. Hans Moravec, a doctoral student under McCarthy, said computers were millions of times too weak to demonstrate intelligence. As patience waned, so did funding, and research progress stagnated for a decade.
2. Boom (1980s-2000s)
The 1980s was a period known as the “AI boom,” when research breakthroughs and increased government funding led to rapid growth and interest in AI. During this period, AI experienced a resurgence, driven by two main factors: an expanding algorithmic toolkit and increased financial support.
John Hopfield and David Rumelhart played key roles in popularizing “deep learning” techniques that enable computers to learn from experience. At the same time, Edward Feigenbaum Expert SystemsA program that replicates the decision-making process of human experts. The program asks experts in a field how they would respond in a given situation, and once it has learned this for nearly all situations, non-experts can then take advice from it.
