Starting as a promising technology in 1950s but failing to deliver for many years, Artificial Intelligence (AI) has made extraordinary progress in the past few years. It improves our lives with many applications, yet it also causes loss of jobs and threats to our security.
Origin of Artificial Intelligence: Simulation of Human Intelligence
The term “Artificial Intelligence” was first coined by an American computer scientist John McCarthy in the summer of 1956 at Dartmouth. He organized a conference there, where a group of professors and scientists brainstormed and laid the foundation of a branch of computer science known as Artificial Intelligence (AI).
The goal of AI was to develop “intelligent” computers that were able to understand and perform works just like human. AI computers were expected to see and feel things, to speak, to learn and reason, to make decisions and solve problems.
Computers or machines that can mimic human intelligence in carrying out a broad range of tasks were said to possess “Artificial General Intelligence” (AGI)
Evolution of Artificial Intelligence
For the following decades after Dartmouth conference in 1956, various research and experiments were carried out, under funding from governments, to pursue the ambitious AI goals, but most works did not produce fruitful results.
From 1990s, more progresses were made in AI research, assisted by advanced technologies (powerful computing, large databases etc) and developments of more productive mathematical tools (machine-learning programming, logic reasoning and control software etc).
Artificial Intelligence has matured sine the beginnings of 21st century. Products using AI are helping us in industries and commerce as well as our daily lives. Notable products are robots, smart manufacturing, customer and market management, decision making tools and personalized voice assistants.
Intelligent computers challenged humans in public contests and won. IBM’s Watson computer defeated ‘Jeopardy’ champions in televised quiz shows in early 2011 and Google’s AlphaGo computer beat Master Lee in a Go match held in March 2016.
Narrow Artificial intelligence versus General Artificial Intelligence
The AI systems we used today are programmed to carry out specified tasks. They demonstrate a more focused approach known as Narrow AI or alternatively known as Artificial Narrow Intelligence.
Google’s AlphaGo was designed and trained to play the Go game. Robots are made to do specific assembly works in the factories or some routine chores for our homes. These systems will not only complete the specified tasks, but also usually do them better than humans.
Narrow AI is different from General AI where the approach is to build a platform that can stimulate human intelligence and can complete a broad range of tasks. That general AI platform is like a true thinking machine. General AI was the original goal of AI in 1950s and is still the ultimate goal for some scientists and researchers.
Some researchers believe that the success of Narrow AI will lead to General AI. The narrow AI working on natural language processing, speech and image recognition etc can lead to smarter machines that will eventually have the reasoning and decision making capabilities comparable to humans.
Causes for fear of Artificial Intelligence
General AI is still many years away. Even before the arrival of true thinking machines, people are feeling the threats of Artificial Intelligence, mainly in two areas. First, the automation systems brought by narrow AI will replace us in the job market. Second, the smart machines will run out of control and harm us.
For the first threat, history showed that mechanization created more jobs than destroyed. Automation allowed workers to produce more, reduce the price of goods, increasing demands and requiring more workers. New jobs are also created in designing and overseeing the operation of intelligent machines.
The second threat has a bigger impact on our survival. It comes back to the basic principle that the AI system, whether it is narrow or general, must be designed with the right purposes to serve humans. That is why there are strong petitions from celebrities and community groups that weapon AI should be banned from development.
On the question of control, AI systems are designed and constructed by engineers and scientists, who should include safety measures to prevent the systems to run out of control. AI systems can learn and become smarter to play tricks, but their learning are trained by scientists. Again, scientists can influence the actions of these machines by training them to learn the right things to do.
Guided co-existence with Artificial Intelligence
AI systems are assisting humans in every aspect of our work and lives, so that we can grow and flourish. To keep these systems under control, we should understand their operation and monitor their interaction with us to ensure that their existence enhances human values.