Deep learning / Deepfakes Research Proessor Mahmood, Student Fahd M. Saleem (G01154081) IT-104-214 (Fall 2018)
However deep learning poses various social as well as ethical concerns to individuals of many backgrounds from celebrities, to political figures, and even the general public. This was notably prevalent during the 2018 campaign season where deep fakes started to take over online, creating the genre of “fake news” which many now associate with deep fakes. During campaign seasons, political figures are especially at risk by deep fakes. Along with the social concern, it compromises the security of political figures, using their own words against them making it nearly impossible to differentiate between the deep fakes & legitimate video. The use of deep fakes is mainly used to discredit, which is predominately done through making it seem as the specific individual has said or done something they normally wouldn’t consider doing inevitably leading to public backlash as the evidence was nearly impossible to distinguish as real or fake without the use of deep learning itself. “Phil's Stock World: Detecting 'deepfake' videos in the blink of an eye” article also provides the flaws in the algorithm of deep fakes. Despite its immense realism, deep fakes cannot perfect the replication of a human blink due. This can only be noticed by deep learning machines conditioned to look out for discrepancies within blink intervals which are a lot less frequent in deep fakes (Stock,2018).
