We are collectively fascinated by artificial intelligence. Since the first work into the field of artificial intelligence began in 1943, the field has attracted much interest by researchers, tech giants, futurists and the media.
The ability of artificial intelligence to learn at an accelerated rate could potentially lead to supercomputers that outperform humanity.
First, it is important to define exactly what artificial intelligence is.
“An area of computer science that emphasizes the creation of intelligent machines that work and react like humans.” – Technopedia
Simulating human intelligence is certainly a lofty goal. The pursuit of highly intelligent AI to rival or exceed human capabilities will certainly bring great reward to humanity.
However, this pursuit is not without potential future risks. This article explores five main reasons why artificial intelligence has the potential to take over the world.
AI is Constantly Learning
Artificial intelligence is able to learn and adapt. Researchers have long debated the possibility of an “intelligence explosion”. In short, this refers to the possibility that advances in artificial intelligence could exceed our own human capabilities.
Given the autonomous nature of artificial intelligence, some researchers suggest that the existence of highly intelligent and capable supercomputers could potentially pose an existential threat to humanity.
Futurists predict that technological singularity (i.e. artificial intelligence becomes so powerful it cannot be reversed) could happen through the rapid progression of advances in technology. This is a general trend predicted by Moore’s Law.
Many industries in the present day benefit from machine learning to enhance their performance. AI has been able to boost profits for companies like Amazon by offering customers individual product recommendations based on an algorithm. Algorithms also offer a competitive edge for the music streaming platform Spotify.
However, at the moment AI or machine learning can show certain biases. Such errors are ones learned from society. Artificial intelligence scientists still work to remove these biases – indicating that AI is at present far from perfect.
The Prevalence of AI
Artificial intelligence offers much to society both in the present and in the future. Currently, artificial intelligence is commonplace within eCommerce, banking, security, marketing and many more industries.
Smartphones are now commonly used in society, meaning that many of us have our data interacted with by an algorithm every day. Algorithms are able to identify faces and tag individuals on social media platforms such as Facebook and Instagram.
In addition, algorithms are used to identify and take down offensive content – such as the case for YouTube. Algorithms have also become commonplace in eCommerce (e.g. Amazon) and entertainment (e.g. Netflix, Spotify) to make individual recommendations based on the use of user data.
Digital assistants such as the Google Assistant and Amazon’s Alexa have become popular in households all across the UK. On an individual level, more and more of the public rely on artificial intelligence to improve their lives in different ways. This is only set to increase with future advances in technology.
The prevalence of artificial intelligence in everyday life means that such technology is difficult to remove or escape from. This also means that there is no way to “opt out” as an individual member of society.
If artificial intelligence did indeed overpower humanity in the future, it would be difficult to do anything about it because AI is so deeply embedded in many industries globally. If technological singularity were to be achieved, it would be immensely difficult to go back.
In 2016, Microsoft deployed a twitter chatbot called Tay. It intended to learn from human Twitter users through conversation on the platform. Within 24 hours, the bot had to be pulled by Microsoft due to assimilating the worst of the internet.
This begs the question – if highly advanced artificial intelligence analyse the data on the internet then surely the worst conclusions could be drawn about humanity
The internet is known for being a dual-edged sword – on one hand, the internet empowers much of us with knowledge, social connections, and opportunities. On the other hand, some parts of the internet can act as a stagnant pool of toxicity. AI could be influenced by online behaviour that may not be fully representative of humanity in real life.
There have been various reported instances of virtual AI assistants chatting with each other. In 2017, 3.5 million people watched a conversation between two Google Home chatbots running Cleverbot AI software being streamed through Twitch.
While the conversation remained friendly, it did turn existential rather quickly. This is interesting considering that Cleverbot is a chatterbot that learns from human interaction. While many examples of human-AI interaction exist, it is fascinating that such interactions do not always go as planned.
Elements such as these create uncertainty over the future direction of AI.
There has been much speculation as to whether artificial intelligence will remain friendly or go “rogue” towards humanity in the future. Debate on this argument often hinges on whether it is in the nature of artificial intelligence to adopt the human values of its creators.
Varying stances on this spectrum have been explored in science fiction novels and films. Asimov’s Three Laws of Robotics sees AI helping humanity. Others paint AI as destructive forces or even taking a neutral (but not necessarily friendly) view towards humanity.
These mediums have inspired thinkers and generated much thought and interest in the field of AI. However, could the presence of “rogue” AI ingrained in such media potentially cause a self-fulling prophecy? Again, this is a possibility among possibilities.
It is uncertain what stance a highly intelligent supercomputer may adopt. Therefore, it is important for computer scientists to input preventative and well thought out measures in the present to prevent unfriendly AI to exist in the future.
AI is powerful
AI can act with autonomy. It can learn at a rapid pace and is prevalent in many industries across the world. Artificial intelligence, therefore, has the capacity to become very powerful.
Artificial intelligence is currently able to create music, synthesize voices, animate a 3D model of a person from a single photo and alter video recordings.
This is all possible through the use of neural networks. Neural networks are algorithms modeled after the human brain. Various types of the neural network exist, operating on different learning methods and complexity.
The fact that AI is already able to “create” is astounding. However, this could lead to all sorts of problems in the future – such as video and audio manipulation to create realistic fake footage. It also begs the question of what AI could be capable of in twenty or thirty years time.
Tesla are famously working on self-driving vehicles through the use of AI. Research into how artificial intelligence could benefit the medical field is also booming – from digital doctors to caregiving robots.
Artificial intelligence really does represent in the forefront of cutting edge technology. If technological singularity were to be achieved (i.e. AI becomes so advanced we cannot go back), then it would certainly have real power across numerous fields.
There is no doubt that artificial intelligence is playing a role of vital importance in our present and even more so in the future.
Supercomputers could potentially take over the world due to AI’s accelerated continuous learning, its integration, and power in key industries. In addition, the interaction with humans and the rejection of values/nature of AI could all play a part in this potential future scenario.
While highly evolved, capable supercomputers could potentially pose a risk in the future it is important to note that not all researchers in the field subscribe to this belief. For example, human limitations and hardware implications may render a chain of continual advances in AI impossible.
What is certain is that with artificial intelligence, we are achieving more than ever.