Gen Alpha or Gen Android? The Case for Keeping OpenAI’s Voice Robotic
Scarlett Johansson has accused OpenAI of creating a voice called “Sky” for ChatGPT that sounds eerily similar to hers. This is despite her declining an offer from the company’s CEO, Sam Altman, to be the voice behind the AI system. Johansson claims that OpenAI pursued a voice that closely resembled hers without her consent, leading her to hire legal counsel and demand transparency from the company regarding the process used to create the “Sky” voice. See here for NPR’s publication of her letter.
OpenAI has since agreed to remove the voice and posted a clarification of how they hired the voice talent for the AI-voice assistants showcased at their GPT4o event, which sparked the controversy.
Hollywood’s Struggle with AI and Creative Rights
The incident involving Scarlett Johansson and OpenAI is a stark reminder of the power of AI to create convincing imitations of actors’ voices and likenesses — with or without their consent. This issue was a central catalyst for the Hollywood strike, as actors, scriptwriters, and other industry professionals demanded protection against the potential exploitation of their creative work and the looming threat of job displacement posed by the unchecked advancement of AI-driven content creation.
Hollywood, after all, is a business and will fund projects that are box office successes. This has led to formulaic approaches to filmmaking, focusing on sequels, remakes, and cinematic universes. While AI technologies like OpenAI’s language models may not eliminate jobs within the industry, they could enable film production companies to hire less-skilled individuals who can leverage these tools (trained on a treasure trove of content) to churn out hit movies.
Granted, the impact of AI on the livelihoods of people in Hollywood and elsewhere is a crucial issue that demands attention. However, there is a broader societal challenge brewing in the background, one that transcends the immediate concerns of the entertainment industry.
Why did OpenAI opt for an AI voice that mimics Scarlett Johansson instead of C-3PO?
While it might be tempting to dismiss the choice of a non-robotic voice as a trivial matter, such a view fails to recognize the significant consequences of blurring the lines between AI and human voices and the potential ramifications of anthropomorphizing artificial intelligence.
For those unfamiliar with Her, it is a dystopian sci-fi flick directed by Spike Jonze and starring Scarlett Johansson as the voice of an artificial intelligence (AI) operating system. Released back in 2013, the movie follows the lonely and depressed Theodore Twombly (Joaquin Phoenix), who falls in love with his AI assistant, Samantha, and develops a romantic relationship with her. Spoiler alert: The AIs “determine” that they are better off without humanity and choose to abandon it.
From my perspective, the movie offered a broad critique against a “supersized” form of social media. In the movie, the company behind the AI-operating system takes the concept of algorithmic addiction to an extreme, bypassing the need for user-generated content entirely. Instead, users become directly addicted to the algorithm itself, forming a deep emotional attachment to the AI persona.
This scenario highlights the dangers of anthropomorphizing AI and blurring the lines between human and artificial intelligence. By creating an AI that mimics human interaction so closely, the company in the movie effectively “skips a step” in the usual social media addiction process. Rather than relying on the dopamine-inducing feedback loops associated with likes, comments, and other forms of user engagement, the “Her” algorithm fosters a direct, personal connection with the user.
Social Media Addiction and Its Consequences
Today, we see the challenges of social media’s dopamine-induced feedback loops. Four years after the movie’s release, Chamath Palihapitiya (Facebook’s former VP of Growth) confessed to feeling “tremendous guilt” about being involved with tools that would create social divisions. In 2020, we saw the release of Social Dilemma, which featured many industry insiders decrying the addictive nature of the technology. In 2023, thirty-three states, including California and New York, filed a lawsuit against Meta Platforms Inc., alleging that the company deliberately designed addictive features on Instagram and Facebook, which contributed to the youth mental health crisis by harming the mental well-being of young users.
Jonathan Haidt released The Anxious Generation in 2024 (and was also featured in Social Dilemma). In the book, he examines the troubling trends in adolescent mental health, mainly focusing on the sharp increases in anxiety and depression starting around 2012. Based on the data from the American College Health, he charts a significant rise in mental health diagnoses among U.S. undergraduates. Anxiety diagnoses have increased by 134% and depression by 106% since 2010. Other mental health issues that have increased since 2010 include ADHD (72%), bipolar disorder (57%), anorexia (100%), substance abuse (33%), and schizophrenia (67%).
Haidt attributes the rise in mental illness among youth to two primary factors: the decline of play-based childhood and the rise of phone-based childhood. The former, starting in the 1980s, saw a reduction in unsupervised outdoor play, depriving children of crucial opportunities for development. The latter, beginning in the late 2000s, introduced constant access to social media via smartphones, contributing to increased anxiety and depression among adolescents. (See here for more information.)
What does this all have to do with choosing C-3PO instead of Scarlett Johansson?
The danger of anthropomorphizing AI is that it could be more problematic than social media. Gen Alpha could become Gen Android if they substitute human relationships for algorithmic ones. It could be the relationship equivalent of eating the empty calories from ultra-processed food (UPF). Though it tastes good, and you feel full, many problems await you in the future. We only need to look to the obesity crisis sweeping America to understand how dramatic an impact eating ‘industrially produced edible substances’ can have on society. Since AI agents will be designed to conform to the user’s wishes, they fail to offer the challenges present in human relationships.
Consequently, it will be a burden to deal with the complexities of dealing with people’s emotions, such as anger, betrayal and so on. Generation Android will increasingly adopt synthetic relationships because it is simply too hard to navigate real ones. This will accelerate the problems we see with social media, leading to a generation that is further insulated from the realities of humanity but more comfortable with the algos that conform to their desires. It will facilitate the further atomization of societies that embrace this technology, resulting in alienated populations preferring AI over human interaction.
Maintaining Distance from Algorithms: Digital Minimalism
Consequently, we need to get distance from algorithms. This has been the preferred approach when it comes to social media. Cal Newport, author of Digital Minimalism, recommends taking a break for 30 days and understanding which social media tools are useful to you. He advocates for a structured process called “digital declutter.”
First, identify and take a break from all optional technologies for 30 days. During this period, engage in meaningful and satisfying non-digital activities. After the 30 days, reintroduce technologies selectively and intentionally, evaluating each for its true value and how it supports your goals and values. Newport emphasizes the importance of optimizing technology use, avoiding clutter, and being intentional in your digital habits to achieve a more focused and fulfilling life.
Future Implications: The Importance of AI Transparency
So, if we need an appropriate level of distance from social media, how much will we need from AI that pretends to be human?
And that’s why using a more mechanical voice will help us stay more vigilant and appreciate not just the fact we are dealing with a computer, but also the corporate interests behind the algorithm. Currently, GenAI is sold on a subscription basis and not through ads or other forms of corporate payments. Perhaps they will stick to this monetization approach. However, we should remember that there was a similar idealism with the Internet. Even the founders of Google were against ads at the start. However, the need to fund their massive server farms forced them to become the ads behemoth they are today.
Social media made us the product for advertisers. The question is, what will we be to these AI systems? There’s a danger of trusting these algorithms because they seem human. For instance, if your AI friend recommends buying a hat, how do you know it wasn’t influenced by the company selling the hat?
Certainly, the robot-centric voice is not a panacea. With such an approach, comes the risk of overestimating AI’s ability more broadly (officially termed “automation bias”). Drawing from my experience as an IT auditor supporting financial auditors, I frequently emphasized the crucial role of system testing before placing trust in the outputs. Financial auditors, often unfamiliar with the importance of testing IT controls, were prone to assume that systems-generated reports were inherently accurate and reliable.
Concluding thoughts: Choosing C-3PO Over Scarlett Johansson
Nevertheless, the more pressing concern lies in the potential for being lulled into a false sense of security by this technology and attributing human qualities to it. Therefore, we must do our best to keep our distance and not anthropomorphize the technology. Having a mechanical alternative will help people recognize they are dealing with a computer. This will be especially important when we need to deal with the commercialization of these technologies. We don’t know yet how they will be monetized. If AI follows the same path as Web 1.0 and 2.0, it could be quite problematic.
While this may not be inevitable, it’s crucial to maintain space between ourselves and AI that pretends to be human. I am sure there will be many regulations, best practices and productivity guides on how to do this. However, it is also important to recognize that AI can be a valuable tool when used appropriately and with clear boundaries. By understanding the capabilities and limitations of AI systems, we can harness their potential to enhance our lives and work while still maintaining a healthy distinction between humans and machines. But for now, choosing C-3PO over Scarlett Johansson is just one way to maintain that distance.
