In recent years, Generative AI has emerged as a double-edged sword, with its capabilities spanning from creative content generation to controversial deepfake technology. While deepfakes have garnered negative attention for their potential misuse (e.g., the spread of misinformation, fake news, or stories that intend to produce harm to others), they also hold great promise for various domains, one of which is Human-Computer Interaction (HCI). So, in this blog post, we’ll delve into the concepts of Generative AI and deepfakes and then explore their potential for HCI, particularly in the realm of deepfake personas or deepfake user personas.
Rise of Generative AI
Generative AI, the offspring of machine learning and neural networks, has revolutionized the way we approach data, image, and text generation. This technology has made it possible for machines to generate human-like content (i.e., content that is indistinguishable from that created by humans), and its application has raised a lot of discussion and concern.
Controversy of Deepfakes
Among the myriad applications of Generative AI, deepfake technology has gained a negative reputation. Deepfakes involve using AI algorithms to create realistic-looking but entirely fabricated content, such as videos or images, where *real* individuals appear to say or do things they never did. This technology, though impressive from a technical standpoint, has raised significant ethical and security concerns. It has been exploited for disinformation campaigns, identity theft, and more, leading to a tarnished reputation for Generative AI.
Potential of Deepfakes for HCI
Despite the negative reputation surrounding deepfakes, they hold immense potential for HCI. HCI is a multidisciplinary field that explores the interaction between humans and computers, with a focus on improving user experiences. Deepfakes, when used responsibly and ethically, can contribute positively to HCI.
One fascinating application of deepfake technology in HCI is the creation of deepfake personas or deepfake user personas. These are simulated characters or user profiles that can be employed in various digital interfaces to enhance user experiences. The idea is that we do not create fake accounts of real people but realistic accounts of “fake” people! By “fake”, we mean that personas by definition are fictitious people that represent real user groups. So, using deepfake technology in the persona context is not harmful in the sense that it would mislead us into thinking that *real* people that we know (e.g., politicians, celebrities) would say things that they did not say. Instead, the deepfake user persona is created by analyzing collective data from the user or customer base, through which we create segments and then proceed into giving the segments a name, face, and appearance. When we apply deepfake technology in this process, the outcome is deepfake personas.
Here are some ways in which deepfake personas can be beneficial:
1. Personalized Assistance: Deepfake personas can be used as virtual assistants with human-like characteristics, making interactions with technology more engaging and intuitive. The general immersiveness of deepfakes — meaning they are more vivid than static persona profiles — is believed to engage end users of systems and thus deepfake personas can create value as a part of positive user experience. However, it is not yet fully understood in which ways deepfake user personas can enhance the persona user experience. More research is needed!
2. User Testing: Designers and developers can use deepfake personas to simulate user interactions and gather valuable insights during the development phase of software or applications. As deepfake user personas are ideally based on real data, what they say can be trusted for making design choices; however, for this to be the case, the validation of the personas needs to be meticulously done. The validation techniques for deepfake personas are not yet fully developed and a lot more research is needed here.
3. Language Learning: Deepfake personas can serve as conversational partners for language learners, providing a realistic environment for practice and improvement. The ability to have “real” conversations with fictitious but extremely realistic people can stimulate people to learn a language more effectively than alternative methods. Again, this remains to be shown by empirical research.
4. Entertainment and Gaming: In the world of entertainment and gaming, deepfake personas can elevate storytelling by creating lifelike characters and enhancing immersion. Developers have already created AI mods for popular games, enhancing the gameplaying experience and “breathing new life” to old games. In the future, AI personas created using deepfake technologies are likely to become more prominent actors in virtual worlds, ranging from gaming to other use cases. For example, we could imagine scenarios where a movie character’s story arch ends in the movie, leaving the audience question what happened to them. Using deepfake technology, fans could create their own alternative universes and storylines for known characters, with deepfake technologies making these appear highly realistic.
While the journey from Generative AI to deepfakes may have been fraught with controversy, it’s essential to recognize the potential these technologies hold, particularly in fields like HCI. Deepfake personas, when used responsibly and with ethical considerations in mind, can pave the way for exciting and innovative applications that improve our interactions with technology. As we continue to explore the capabilities of Generative AI, it’s crucial to strike a balance between innovation and responsible use to unlock the full potential of these transformative technologies. Many of the benefits remains theoretical and potential at this point, calling for substantial research and development efforts in the field of deepfake personas.