AI has made remarkable strides in sophistication, enabling chatbots to engage in meaningful conversations with humans. Its pervasive presence is undeniable; individuals contribute to its advancement, consciously or unconsciously. Each internet search conducted adds to the vast pool of data that feeds AI’s knowledge, while smartphones incorporate AI features to monitor user activities. The prospect of AI surpassing human capabilities is plausible. Concerns regarding the existential threat AI poses to humanity have been voiced by many.
In a previous post, I argued that even though we have crossed the point of no return, we still have some time left since they still need our help in reproduction. And when AI wants to take over the world, they don’t need killer robots or start a nuclear war. All they need to do is for us humans to rely on them too much that we cannot survive without them.
A couple of weeks ago, I attended a conference on the ethics of AI. I’m not an ethicist, but I still enjoyed the conference. One major takeaway from the conference is that the threat of AI is not an army of killer robots or AI figured out a means to mass murder humans. The primary threat is humans misusing AI, which leads to human self-destruction. Today, I will be focusing on Deep Fakes technology.
Technology, by its nature, lacks inherent moral value. Its potential for good or harm lies in how it is wielded. Even technologies created to benefit humanity can become tools of destruction when they fall into the wrong hands. Throughout history, we have witnessed numerous instances where malevolent individuals have twisted technology to serve their wicked agendas.
A striking example can be found in Nazi Germany, where advanced war machines, both conventional and biological, surpassed those of other European nations. Adolf Hitler capitalized on this technological superiority to invade Poland and France, sparking the devastating conflict of World War II. The Nazi regime perpetrated unspeakable horrors, including systematically exterminating millions of Jews.
In the digital age, the accessibility of technology has lowered the barrier for criminals to acquire and utilize it for malicious purposes. One prevalent example is the use of ransomware by cyber criminals to extort large sums of money from companies. Furthermore, there have been instances where cyberattacks have targeted power plants and other critical infrastructure.
In 2016, substantial evidence emerged suggesting that the Russian Government was involved in operating a troll farm responsible for creating fake Facebook accounts. These accounts were allegedly used to interfere in US elections by spreading disinformation and attempting to manipulate public opinion. At the time, it was believed that they used human beings to create those posts; with the help of AI, they can now manipulate humans’ emotions far better. Many anti-terrorism experts have explored the vulnerability of the election system to Chatbots like the ChatGPT. Jason Blazakis of the Middlebury Institute of International Studies at Monterey concluded, “Some of it is far better-written than right-wing text we have analyzed as scholars.”
I love photography; I recently purchased an AI-powered image processing software; with a touch of a bottom, one can remove objects or change the sky. Of the three images below, two are modified with a simple mouse click, while one is the original. Can you tell which one is the original?
Although AI-generated images are not flawless, they are proficient at significantly modifying authentic images and transforming them into something entirely different. For instance, they can seamlessly change the sky in a photograph, converting a midday scene into a stunning sunset. This capability raises concerns as it reminds us of the age-old adage that the devil’s oldest trick distorts the truth. When the serpent tempted Eve, it didn’t present a complete lie but instead distorted God’s word, highlighting the potential dangers of manipulation and deception in various contexts.
I recently viewed the fifth installment of Indiana Jones, where Harrison Ford portrays the iconic protagonist. The film begins with a captivating opening sequence featuring a younger Dr. Jones engaged in a battle against Nazis towards the end of World War II. Remarkably, an AI de-aging algorithm transformed the 80-year-old actor’s appearance to resemble himself from the 1940s. The result is compelling, making it the most advanced implementation of de-aging technology ever showcased on the big screen.
While AI-modified images find their place within the bounds of fair use in the entertainment industry, the most disconcerting aspect of AI-generated media lies in deep fakes. Deep fakes involve manipulating videos or images of individuals by digitally altering their faces or bodies to make them appear as someone else. Numerous deep fake videos depicting former President Trump resisting arrest have been circulating online, often intended as humorous content. However, it is easy to imagine a scenario where politicians exploit deep fake technology to create fabricated videos of their political adversaries, spreading damaging rumors about them. Conversely, when individuals are captured on camera engaging in inappropriate behavior, they may dismiss the evidence as a fabricated video, undermining the trust in authentic visual evidence.
Deep fakes can instigate doubt and erode trust, potentially leading to the fragmentation of society. They can potentially be exploited for inciting violence or even triggering conflicts on a larger scale. Although contemporary AI demonstrates considerable intelligence, its responses are derived from its underlying data models. While AI predominantly operates as a truth-telling mechanism, it can inadvertently generate falsehoods if the data model contains erroneous information. However, AI has yet to reach an exceedingly persuasive stage, capable of creating fake videos to coerce individuals into taking specific actions. Humans will self-destruct before AI can acquire the necessary capabilities to destroy humans.
During my Ph.D. research, I was an early pioneer of Scanning Tunneling Microscopy (STM) and Atomic Force Microscopy (AFM). They are sight-extending instruments and incredibly powerful tools capable of resolving individual atoms. Despite their remarkable capabilities, these microscopes can produce distorted images due to various artifacts; part of my research was to understand various artifacts. Throughout my training, I learned to rely on visual interpretation only partially. Instead, I had to engage my intellect and utilize additional analytical tools to enhance accuracy and obtain reliable results. Combining human cognition and supportive tools is essential to overcome potential challenges and to achieve meaningful scientific insights. In the same way, with so many fake videos circulating, it is hard to tell what is real and what is fake. We must educate ourselves to avoid being fooled so quickly and always look for other sources to verify the source.
But even with our best human efforts, detecting deep fake content can be challenging. Realizing the problem, big tech giants like Facebook, Microsoft, and Google are actively working on creating software tools for deep fake detection. In a sense, we are employing AI to detect AI.
To enhance accountability and transparency regarding AI-generated content, it is necessary to implement regulations that mandate the disclosure of such information, akin to enforcing health warnings in tobacco advertisements. It would be beneficial to embed a distinct signature within media as well. However, it is important to note that these measures primarily address individuals using commercial Deepfake software, as nations can develop their own AI programs to propagate misinformation without including such warnings and digital signatures.
Furthermore, it is crucial to establish comprehensive legislation that effectively prohibits the malicious dissemination of deepfake materials. While some existing laws attempt to address this issue, their efficacy is limited due to the rapid advancement of technology and the need to uphold the principles of free speech. To rectify this, it is imperative that legislators with a strong understanding of AI technology collaborate to draft new laws that can effectively regulate Deepfakes. Until we understand how to regulate this technology, it may be prudent to consider a temporary halt on the use of deepfakes.
Reducing the demand for deepfakes is a practical approach to tackling this issue. As Christians, we must refrain from using Deepfake technology to propagate misinformation, regardless of any seemingly noble intentions. Christianity places great emphasis on the pursuit of truth, as the Bible states: “The truth will set you free.” Therefore, our commitment should be focused on upholding the truth and fostering loving environments rather than contributing to the dissemination of distrust. In social media, we must exercise caution when sharing information. I personally abide by the principle of only sharing uplifting and wholesome stories that inspire and uplift others. Even if I experience negative emotions, I refrain from sharing negative information. A potent strategy to combat falsehoods is taking personal responsibility to cease their propagation. Occasionally, I engage in a media fast, refraining from media consumption for a few days. This practice helps to clear my mind and enables me to focus on more meaningful aspects of life.