Article Review 1 The Danger of Human Assessment of AI Deepfakes:
William Whitehurst
February 22, 2025
The Danger of Human Assessment of AI Deepfakes
Abstract
In this paper I will review the cybersecurity article “Testing human ability to detect ‘deepfake’ images of human faces.” Deepfakes threaten our society and need to be addressed. If it is not addressed soon then we will soon see the consequences.
Introduction
When it comes to the dangers of AI there are many. One of the bigger concerns is deepfake images or videos. “‘Deepfakes’ are computationally created entities that falsely represent reality (Bray et al., 2023).” We will start by going over the article Testing human ability to detect ‘deepfake’ images of human faces and how deepfakes are dangerous to society.
Main body
In the article “Testing human ability to detect ‘deepfake’ images of human faces,” the focus was to see how good humans are at detecting AI generated images. In this experiment they got 280 participants from a website called Prolific and were split into four groups using an online survey. The participants were offered £6/hr, and if their performance were in the top 50% of participants they would get a 50% bonus. Each participant was given 20 deepfake images from a pool of 50, and 50 real images. They were asked to determine which ones were fake and to rank their confidence. Overall, there was a 62% accuracy. “While performance was marginally better for those in the experimental conditions, a one-way analysis of variance [F(3276) = 1.95, p > 0.1 ns] suggested that there were no statistically significant differences between groups. Simply put, for overall performance, none of the intervention conditions led to a significant improvement in participants’ accuracy at detecting deepfake from non-deepfake images (Bray et al., 2023).” One of the biggest things they found was there was misplaced confidence for each image even they got it wrong.
Deepfakes are a serious threat to society. This article has given us a good understanding of our lack of misplaced confidence and ability to detect deepfakes. They can be used for political gain, monetary gain, and there can even be a psychological impact on people. What I mean by psychological impact is that people might not know what is real and what is fake causing them to lose trust in technology. Deepfakes can also be used against poor/poverty-stricken people who do not have access to education about this kind of technology. It can also be used against people who might not understand the language of that country. These are serious issues because politicians and corporations can use this against these people. Even though the article showed a 62% detection rate overall, deepfakes are only going to get better and better.
Conclusion
In conclusion, AI deepfakes are a serious threat to society. We need to start thinking of how we can prevent deepfakes. They will only get better, and more people will start to use them for their own benefit. When it comes to media we need to have reliable sources and deepfakes can give us false information. We need to be vigilant and think of the ethics of technology and how it is shaping society.
References
Bray, S. D., Johnson, S. D., & Kleinberg, B. (2023, June 23). Testing human ability to detect ‘deepfake’ images of Human Faces | Journal of Cybersecurity | Oxford academic. Testing human ability to detect ‘deepfake’ images of human faces. https://academic.oup.com/cybersecurity/article/9/1/tyad011/7205694
Article Review 2 Cyberbullying on Social Media:
William Whitehurst
April 8, 2025
Cyberbullying on Social Media
Introduction
Cyberbullying is not this new concept for us. We can see how it has gotten worse over the years through social media use. It will continually get worse as more people start to connect online. Especially with younger generations and even continuing with current generations.
Main body
In the article “Cyberbullying on Social Meida: Definitions, Prevalence, and Impact Challenges” there are five questions they answer. How is cyberbullying defined in literature? What factors contribute to cyberbullying, and how can they be measured? What is the impact of personality traits on the likelihood of engaging in or being victimized by cyberbullying? How do direct involvement and indirect experiences influence the prevalence of cyberbullying? How does cyberbullying affect individuals, and what measures can be taken to mitigate its effects? To answer these questions, they review the relevant literature to cyberbullying on children, adolescents, and adults. “The following databases were used to identify articles: Scopus, Google Scholar, and Science Direct (Ray et al., 2024).” They used key words such as cyberbullying, social media, children, adolescents, young adults, and adults. Also using these words in combination with each other. They reviewed 71 papers out of 4,000 from the years 2007 to 2022. They used all kinds of papers such as empirical, case studies, reviews, and theoretical from all over the world. “We have identified and highlighted tow critical gaps: the lack of extensive research on cyberbullying among adults, and the inconsistencies in the definitions of cyberbullying across various studies (Ray et al., 2024).”
Cyberbullying is a major concern for us a whole. Cyberbullying can have major impacts on a person’s mental health. With kids getting technology at a younger age this will affect how children develop psychologically. Being either a victim or perpetrator of cyberbullying could lead you to committing cybercrimes. This is another reason that we should be worried. Soon it will not just be adults committing these kinds of crimes, it will also be kids too. There are some kids even today that do this, but there will be a spike in younger kids committing cybercrimes. With mass cyberbullying this will influence societies and cultures around the world. We could see people going from attacking people online to doing it in person. A new culture would form, and it could be worldwide. This would not be the kind of worldwide culture we want to form.
This article highlights how big cyberbullying is, and how little research is done on it. This article shows us that we need to do more research on cyberbullying. We need to better define what cyberbullying is and try to find ways to prevent it. Marginalized groups are more likely to be victims of cyberbullying. There are many reasons for that, and it can lead to a disconnect in society. This makes them more likely to also become perpetrators of cyberbullying and cybercrimes.
Conclusion
In conclusion, this article gives us sufficient evidence that more research needs to be done on cyberbullying. With more kids connecting online at younger ages this becomes a huge concern. It will affect how we develop psychologically leading people to become cyberbullies or commit cybercrimes. Our societies will be changed, and we could see a world culture of animosity and hatred form. The marginalized groups are going to be the first to feel the blow from this huge wave of cyberbullying.
References
Ray, G., McDermott, C. D., & Nicho, M. (2024, December 19). Cyberbullying on Social Media: Definitions, Prevalence, and Impact Challenges. Journal of Cybersecurity. https://academic.oup.com/cybersecurity/article/10/1/tyae026/7928395?searchresult=1