How AI started out in general
1950s
The first time AI actually gained some traction was in the 1950s when the earliest machines were mostly huge, very powerful computers. Earlier, organizations like NASA employed teams of people (mostly women) to act as human "computers" that would perform complex calculations. Even though ideas of a robot that could think like a human were very new, a lot of researchers were curious to find out wha a computer was capable of.
Foundations 1960s-1970s
In the period between the 1960s and the 1970s the idea of AI grew significantly after the Dartmouth conference, the founding place for AI as a field. The earliest milestone was ELIZA. This was created in 1966 by MIT computer scientist Joseph Weizenbaum. ELIZA is considered the first chatbot and could stimulate therapy by repurposing the answers users gave into questions that prompted further conversation.
AI winter 1974
In 1964 the applied mathematician Sir James Lighthill published a major report on academic AI research stating that researchers had over - promised and under - delivered when it came to the potential intelligence of machines. This report started what was called the AI winter by encouraging people to stop funding AI research. The AI winter continued for another two decades until in the late 1990s when AI received more funding to continue making leaps forward.
1980
A humanoid robot was built in Waseda university in Japan.While building the WABOT-1 humans focused on basic mobility and communication,WABOT-2 was specifically designed as a musician robot.It specialized in playing a electronic organ but can also accompany a human singer.This creation focused on on a more creative robot rather than a less creative robot.
1982
Japan launched the Fifth Generation Computer Systems Project (FGCS).Their aim was to develop computers to handle problem solving.This project was aimed to not give the machine a specific task and tell it a broader or bigger task,so the machine figures out how to finish that task by themselves.
1984
David Rumelhart, Geoffrey Hinton and Ronald Williams publish the paper where they introduce a new algorithm.This new method enables humans to use multilayer networks, basically layers of really good and strong networks to learn about complex patterns.This breakthrough enabled the door for the evolution of deep learning in the 2000s and 2010s.
1987
Apple CEO John Sculley presented the Knowledge Navigator video,which are digital smart agents that process large amounts of data over networked systems.It can also retrieve data,answer questions and display the information it finds on the internet.
AI growth 2000-2019
By the 2000s, (as people started gaining more interest in AI), AI developers were pressured into creating more and more intelligent machines. This led to a period where AI experienced major growth. Below are a few of the examples.
Nasa Rovers
In 2004 Mars (the planet, not the candy bar) was orbiting a lot closer to the Earth than usual. Nasa took advantage of this and launched two rovers on its surface. Spirit and opportunity used AI to navigate the surface of Mars without human intervention.
IBM watson
Many years after IBMs deep blue program, IBM created another competitive AI machine in 2011. This AI system was created to win the US hit game Jeopardy against Ken Jennings and Brad Rutter.
Geoffrey Hinton and the neural networks
The computer scientist Geoffrey Hinton started experimenting with the idea of Neural Networks when he was doing work on his PHD in the 1970s. His ideas only became known in 2012 when he and two of his graduates decided to display their research at the competition ImageNet. This is when the Tech industry saw how far neural networks had developed.
Alpha go
The very old game Go was considered impossible for a computer to learn as there are a lot of potential positions involved.However, google's deepmind lab was able to create a machine that went on to beat Lee sedol, the world's best Go player in a game.
2000
Cynthia Breazeal at MIT developed Kismet,which is designed to interact with humans emotionally and socially.Kismet has many functionalities some of them are camera to machine,microphones and facial features.It responds to human emotions such as happiness,sadness and excitement.This innovation takes social robotics toa whole new level.
AI surge, 2020-present
In recent years the development of generative AI has improved drastically with LLMs like ChatGPT and Gemini. The major improvement part though, is that Gen AI can continuously learn from data unlike earlier models.