Milestones in artificial intelligence

From intelligent personal assistants to home robots, technology once thought of as a sci-fi dream is now embedded into everyday life. But this leap from dream to reality didn’t happen overnight.

There is no one ‘eureka’ moment in a field as vast as AI. Rather, the technology we enjoy today is a result of countless milestones in artificial intelligence, delivered by countless forgotten people across a countless range of projects.

So, let’s pay homage to some of that work. Here’s a look back at just a few of the significant milestones in artificial intelligence that make our modern lives possible.

1943 – The first ANN

1943 saw the publishing of Warren McCulloch and Walter Pitts’ A Logical Calculus of the Ideas Immanent in Nervous Activity. In it, the two discuss the possibility of networks of artificial neurons performing logical functions.

Fast forward to today, and artificial neural networks (ANNs) work as McCulloch and Pitts conceived in the 40s. ANNs are used in modern AI functionality to make ‘learning’ possible. As such, their introduction is one of the first major milestones in artificial intelligence.

1948 – First autonomous robots

Another exciting AI milestone came in 1948 when William Grey Walter created Elmer and Elsie. These two robots were the first to work autonomously — independent of a human. They could navigate their way around obstacles using light and touch.

1955 – Official term and academic recognition

Perhaps one of the most impactful milestones in artificial intelligence is the creation of the term itself. John McCarthy, computer scientist and AI ‘founding father’, coined the term ‘artificial intelligence’ in 1955.

He then went on to organise the Dartmouth University conference, which took place in the summer of 1956. It was this conference that marked the introduction of AI as a recognised field of academic study.

1964 – The first chatbot

The next big AI milestone we’ve pulled out centres on chatbots. Before SIRI, before Alexa, came ELIZA, the first face of conversational AI.

Built between 1964-1966, ELIZA is one of the first natural language processing programs. ELIZA marks the start of the pursuit of AI chatbots. That is, computers that can understand and talk to us. (Despite Joseph Weizenbaum, her creator, often reminding people that she wasn’t intelligent.)

1969 – Backpropagation

Today, backpropagation is a core part of AI. It essentially allows a neural network to learn from mistakes. While this might not sound thrilling, it means that AI can be trained to perform better over time. In turn, this allows AI to become increasingly better at making decisions.

The idea of backpropagation, then, is another of the milestones in artificial intelligence. The idea was conceived in 1969, and became popular in 1986.

1970 – First ‘intelligent’ robot

Though autonomous robots had been around for decades by 1970, it wasn’t until the creation of ‘Shakey’ that a robot could actually reason through its actions.

Unlike its predecessors, Shakey the Robot did not need to be instructed on each individual step of a complex process. Rather, it could analyse commands and break them down. So, Shakey represents a weighty AI milestone: the first physical robot controlled by artificial intelligence.

1978 – Voice-activated technology

If you were a child in the late 70s or early 80s, you likely would have encountered the Speak & Spell. By being able to “speak” to children, this language learning game was able to teach both the correct spelling and pronunciation of a word.

And this marked the first electronic duplication of the human vocal tract on a single silicon chip. Which, in turn, prepared us for something that has only recently exploded: voice-activated tech.

1981 – Commercialised AI

The early 80s saw the rise and fall of ‘expert systems’. As the name suggests, this is a computer system with similar decision-making abilities to a human expert.

Expert systems promised complex problem-solving for businesses, and enjoyed widespread commercial viability. Indeed, two-thirds of Fortune 500 companies used expert systems during their heyday. Since they didn’t quite live up to the hype, however, this heyday proved brief.

1989 – Chess victories – defeating masters

Games are an important part of AI research. They help track the progress of AI and bring new insights and additional attention to the technology.

Chess programs had beaten novice players in the late 1950s. But it wasn’t until the late ‘80s that AI chess programs started to challenge — and defeat — master chess players. Specifically, chess master David Levy was beaten by a computer for the first time in 1989. This marked another AI milestone.

1996 – Chess victories – defeating the world champion

In 1996, IBMs Deep Blue achieved the first single-game victory against reigning world chess champion Garry Kasparov. (Though it lost the match.)

In the following year, an updated version of Deep Blue defeated Kasparov in a 6-game match, winning two, losing one, and drawing three. This victory over humans is one of the most prominent milestones in artificial intelligence.

1998 – Widespread introduction – Furby and machine learning

In 1998, Furby hit shelves in time for Christmas. Over 40 million units sold in the first three years of release. But what makes Furby an AI milestone?

Furbies displayed fake intelligence, appearing to learn language over time. They introduced the general population to the idea of machines that could learn, and talk to you. In other words, they gave a fuzzy face to the concept of AI entering homes.

2001 – A.I. Artificial Intelligence

In 2001, Stephen Spielberg’s A.I. was released. The film tells the story of David, a humanoid boy robot that can love and express emotion like a human child.  

Though sci-fi depictions of artificial intelligence in film were nothing new by 2001, A.I. is novel for frequently switching to David’s perspective. Rather than yearning for his own advancement, David yearns for love and belonging. As such, the film sparked sympathetic considerations on how AI might fit into the world around us.

2010 – Jeopardy! win

Chess wasn’t enough for AI. As such, in 2010 and 2011, IBM’s Watson took on the game show Jeopardy!

Watson defeated two former Jeopardy! champions. In doing so, Watson demonstrated the potential for NLP coupled with the processing speed of a computer.

2011 – Voice assistant

2011 also saw the release of Siri, Apple’s voice-controlled virtual assistant. Siri marks one of the most widely loved milestones in artificial intelligence to this day.

Siri’s use of speech recognition and NLP was nothing short of ground-breaking for everyday users. It’s also the first (advanced) version of a technology that’s since become widespread, with Alexa and Google in our homes, and Siri in our pockets.

2016 – Winning at Go

Back to the games, 2016 brought an AI victory that became one of the breakthrough runners up that year. It was a public display of how far AI had come.

This was the year that AlphaGo defeated Lee Sedol, Go world champion, four games to one. Go is largely considered one of the hardest games in the world to master. This victory demonstrated that AI was (and is) capable of handling a seemingly intractable problem.

And more milestones in artificial intelligence than we can feasibly list

AI has a rich history full of major events. Indeed, compiling a list of every milestone in artificial intelligence would be a job for AI itself. So, this list is only a fractional part of the vast body of work behind our AI-enriched lives.

And, as the technology continues to mature, we foresee a proliferation of AI milestones in the future.

But what they are remains to be seen. Perhaps we will see general intelligence, or the singularity. Only time will tell.

Useful links

ELI5: what is an artificial neural network?

A history of automation: The rise of robots and AI

The awkward interplay of man vs machine