AI Can Destroy Human World: Truth or Myth?
Artificial Intelligence - or AI - has been developing at a rapid rate these last few years. It has practical applications in our everyday lives, and it is more present than we are aware of. This begs the question whether we should be worried if AI has the ability to destroy the human world or not. Are we looking at an Age of Ultron-like future, or perhaps the future does not seem so grim after all?
What’s the Prognosis?
Opinions on this topic vary. One of the greatest minds of all times, Stephen Hawking, has raised his concerns about the possibility of AI being the end of the world as we know it. On the other hand, many technological geniuses, such as Bill Gates, have stated that there is no room for panic and that AI can only make human lives better.
It is a fact that computers are becoming more and more intelligent, and this is due to the fact that they can store and process larger and larger amounts of data. However, computers do not have the intuition that humans have, which is one key difference between the former and the latter. For instance, humans are able to learn social pragmatics with ease, since it is in their nature, but computers still struggle with it. In other words, it might be said that it is highly unlikely that AI will possess the skillset which defines humans and human intelligence.
AI and Thought Manipulation
When discussing the influence of AI over humans, and its overall presence in the world, there is one fact of which we should be aware: our thoughts can be manipulated by machines. Has this ever happened to you: you are having a casual talk with a friend of yours on Facebook or Twitter, and 10 minutes later you are faced with an advertisement on the same topic that you and your friend talked about?
Targeting and Retargeting Technique
Yes, you might be getting a little bit too paranoid, but you are probably not far from the truth. For instance, Amazon keep detailed records of products users have searched for, then use the data with the help of AI software in order to target and retarget those users with advertisements on other websites which users decide to visit.
This targeting and retargeting technique would likely result in a purchase by the users. After all, humans are susceptible to external influence. Professionals say this is only one of the instances where people seldom notice the presence of AI on the internet. However, it is highly unlikely that an Age of Ultron scenario would happen out of that.
Advantages of AI
Owing to its intelligence, AI can be used to track criminal activities on the internet, such as bank frauds, financial scams, harassment, child abuse, organizing terrorism, etc. In other words, AI can be very useful if developed in the proper manner.
For instance, more than a year ago, Facebook has announced their campaign for fighting terrorism by employing a team of experts who would create AI that eradicates any form of behavior related to terrorism. It is a slow and painstaking process, but they already report palpable results - the AI software uses image recognition to eliminate content which has been banned previously; it analyzes text and is able to determine on its own whether that text has previously been used for unlawful activities; it is able to calculate the next unlawful occurrence based on social ties; etc. And since Facebook still reigns supreme in the world of social media, it is evident how AI helps reduce, if not completely eliminate, harmful behavior.
Drawbacks to AI
But there is an obvious drawback to AI - it is not hack-proof. Cybercriminals are becoming more innovative by the day and it is possible that they may hack and control the AI system in order to do criminal deeds. Ever since AI had become the newest new thing in the world of technology, there have been numerous hackings of what seemed to be a perfect AI software: bypassing spam filters, bypassing facial recognition, faking voice commands, etc.
And it does not take too much - in 2017 a security company called Endgame demonstrated how they used the shareware OpenAI to create intelligent malware that security software systems were unable to detect. By making minor modification in the original code, they were able to create intelligent malware that appeared harmless to all security engines. And let us not forget the experiment conducted in 2016 by ZeroFOX, when two scientists trained AI software to learn how to phish from the data posted on social media and the software turned out to be better at phishing than an actual human. It all boils down to the claim that people themselves would be responsible for their own downfall should they decide to use AI in adverse ways.
Regarding the topic of whether AI can think and act for itself, the answer would be: most likely not. Obviously, this statement might not refer to some distant future, but it will surely be applicable in the next couple of decades. This is so because the systems of the AI have predefined parameters which aid the AI to stay on track and debug.
A problem could be created only by a human factor, that is, only if somebody tampers with the AI system parameters. After all, despite the fact that AI and computers, in general, are gaining increasingly more powerful processing capabilities, the human brain is still unmatched. Computers might be able to store more information and can process data faster, but they are intrinsically flawed - they lack the general human intelligence: intrapersonal and interpersonal intelligence, creativity, naturalistic intelligence, linguistic intelligence, etc.
Is AI the way towards the future? Yes. Do we, as humans, have the capacity to pave that way to a possible hell on Earth? Most definitely. All in all, it can be said that AI can transform our lives in many aspects, and in a positive manner. And while all great things come with great responsibilities, as long as we are careful, there is nothing to be afraid of.