Written by Julius K. (KIS'19)
Edited by Jenny C. (KIS'19)
━━ February 26, 2018 ━━
As the world becomes increasingly dependent on the use of technology, the demand for computing systems to function autonomously has been created. This marveled technology, also known as artificial intelligence, can be seen as the ultimate goal in technological advancement. As a result, all major corporations such as Google, Samsung, and Amazon have prepared to invest all of their time and focus on developing the technology. Since artificial intelligence development first grabbed the attention of the whole world, there have been countless debates regarding catastrophic consequences this could lead to. Of the many debates, one of the most famous debates would be the Twitter battle between Facebook founder Mark Zuckerberg, and Tesla and SpaceX founder Elon Musk. Most of the projected outcomes that stemmed from such debates followed the scenarios of Terminator, Avengers: Age of Ultron, I, Robot, or any other films that deal with intelligent robots taking over the world. While it is true that such terrifying scenarios are possible if the technology were fully developed, the reality is that given the current state of AI technology, most of us won’t be seeing or meeting fully intelligent robots in our lifetimes. But before we can take a deep breath and celebrate in disappointing relief, we should first take a moment to analyze where we are right now as of in terms of the technology.
━━━━━━━━
Is Artificial Intelligence Actually as Threatening as We Predict?
Whether it’s Google Assistant, semi-autonomous cars, or AlphaGo, all of these well-known technologies have a commonality: they can only operate within their programmed regions. What this means is that Google Assistant can only operate within the range of the phones or laptops it is installed in, the hands-free driving technology in cars such as Tesla can only make maneuvers based on its sensors and pre-installed programming, and AlphaGo can only do exactly what it is known for: play Go and win by calculating every single combination and permutation move possible that its opponent might use. And this AlphaGo can only do so with sufficient energy and supplied to it. This means that all of the current AI devices and technologies that scientists and engineers have developed are very narrow in its functions and capabilities, thus called narrow-AI.
The issue with narrow-AI is the process of how the systems are being programmed to carry out their ordered tasks. Almost all semi-autonomous devices that have been created so far have all been made by programming the robots or systems beforehand and having the system respond based on what it has been programmed to do. In short, even Siri(the voice assistant for an Apple iPhone) or the well-known Asimo (the humanoid robot developed by Honda), are just advanced versions of the Roomba, the small circular house-cleaning device that runs on sensors. The issue with this is that in order for any of these devices to be operational, this means that those devices must be dealt with by a human first before being functional. In no way can any of the developed technology so far order itself to carry out a completely original task without a human to preprogram it to do so accordingly.The exception to this rule is the
computing systems created by DeepMind, a company acquired by Google in 2014. DeepMind claims that their systems are not programmed and that reinforcement learning is used. But even with DeepMind being an exception to the need of being pre-programmed by humans, the capabilities of their systems are not only limited but extremely inefficient. The energy used to power the AlphaGo on an instantaneous basis in comparison to Lee Sedol (the #1 Go player in the world) was 1 megawatt to 20 watts, which is 50,000 times more than what Lee Sedol used. This is not regarding the amount of energy and resources it took the create the massive, RV-sized machine. So when even the most advanced artificial intelligence system is compared to the human brain, the AI system is lacking in both energy and efficiency. Regardless of energy inefficiency and the vast amounts of literal, physical space that current AI technology requires, in order for a fully autonomous AI system, the practice of programming a system must be an act of the past. In other words, the matter and material makeup of the machine and system must be programmable, flexible, and self-sustaining just like the human brain.
But for the benefit of the doubt, let us put ourselves in the position of when the perfect artificial brain becomes a reality. Many fear that even when ordered to do good, it will use destructive tactics to do so. An example of this would be if an AI system were ordered to maximize the profit of a medication company, it might infiltrate government securities and release propaganda to initiate a war. The motivation for AI to do this would be simply because it lacks soul and emotion, the components that perhaps makes us human. However, it is most likely that by the time such technology is under control to the point where we can order it to perform certain tasks, there will not only be one artificial intelligence brain out there and as a result of competition, the AI systems could eliminate one another in a process similar to natural selection. It is likely that there are many other systems that are used by other companies to maximize profit, and with the computer being able to think just like a human, it will begin to compete with other existing AI programs. As a result, the Skynets and Ultrons of the real world may not even have an objective to eradicate human life or enslave it; they might even eventually self-destruct by competing for each other to find the superior among themselves.
Another reason as to why we might be able to breathe easy when intelligent robots plan to take over the world is the material makeup of such robots. Because “they” are likely to create themselves from mechanical machines and computer chips, if we are able to expose their bodies either to freezing temperature or above 85 degrees Celsius, they will malfunction as the electronic devices within them are not able to work at extreme temperatures. Of course, if the robots have an air conditioning system or a heater within them, they will be able to defend themselves at an extreme temperature to an extent. Nevertheless, the general idea in defending ourselves for these robots is exposing them to extreme temperatures. And, of course, we have the comedic solution of “spraying water” at these robots, a notorious idea proposed to Elon Musk by a Twitter user.
Now, it would be absurd to say that we as a species would be completely fine even with the uprise of true AI. It would also be ridiculous to say that neither the potential problems and issues of an AI uprising are the only possible catastrophic consequences, and nor are they inevitable. There will be new problems that spawn non-stop with the lightning thinking speeds of these robots, which we humans may not be able to keep up with. So the predictions throughout this paper are merely that — predictions. In the far future, when the news that the first fully autonomous AI created is released, don’t sleep too easy. It is highly likely that problems will arise immediately. But the biggest immediate threat that we face as a species is not the robots, nor are they any celestial object. The biggest threat we face as species is us — humans.
━━━━━━━━
Bibliography:
-Ilic, Andrijana. “Artificial Intelligence: Where Are We Now And What The Future Of AI Will
Look Like?” Out of the Box Science, Out of the Box Science, 28 Apr. 2017,
- Liptak, Andrew. “James Cameron Is Rumored to Oversee Terminator Sequels, but It's Not so
Simple.” The Verge, The Verge, 22 Jan. 2017,
Pictures:
━━━━━━━━
#spectra #technology #engineering #AI #artificialintelligence