Cheat GPT
How OpenAI challenges our integrity and authenticity
Copy Cat
In the 2014 Oscar winning movie “Her”, Joaquin Phoenix plays Theodore Twombly — a depressed and soon to be divorced middle aged man who falls in love with an artificially intelligent personal assistant named Samantha (voiced by Scarlett Johansson). 10 years after the movie’s release, Sam Altman’s OpenAI released ChatGPT 4.0 with a voice named Sky. The issue was, Samantha from the movie and ChatGPT’s Sky sounded uncannily similar — to the point where Johansson’s lawyers took legal action against the company for utilizing the actress’ voice without her consent. OpenAI has recently dropped the voice and apologized to the actress.
Johansson’s voice is not the only intellectual property OpenAI has allegedly “stolen”. Late last year, the New York Times sued OpenAI and its colossal backer Microsoft for copyright infringements in utilizing millions of their online articles as a knowledge base to train their large language models. The lawsuit brings about very complex issues regarding the definition of “fair use” and whether or not OpenAI’s technology is illegally “memorizing” and regurgitating copyrighted material from the New York Times’ news archives.
Concerningly, the $80 billion AI startup behaved as the guilty party in both occasions. Just as they took down Scarlett Johansson’s alleged voice from their product and apologized to the actress, OpenAI has erred on the side of the accuser (newspaper companies) when it comes to their alleged copyright infringements of online text materials. Though they have not reached a deal with the New York Times, the company has entered a deal with Rupert Murdoch’s News Corp to license the media conglomerate’s online news materials for their large language model data set. This will give OpenAI legal and valuable access to Murdoch's wide array of intellectual properties including all archives from The Wall Street Journal, The New York Post and much more.
By taking down Johansson’s alleged voice from ChatGPT 4.0 and entering licensing deals with newspaper companies, OpenAI is effectively admitting their past wrongdoings and attempting to make amends with their accusers. Whether it be copying somebody’s voice, words, or ideas, the company has gained an unfortunate reputation as being a copycat.
Liar Liar
Unfortunately, it is not just the company itself who has been engaging in disingenuous behavior. Their customers are doing the same. OpenAI’s revolutionary technology seems to have provided a powerful and anonymous global breeding ground for a behavior that we may loosely call “cheating”.
According to Google, cheating means to: “Act dishonestly or unfairly in order to gain an advantage, especially in a game or examination”. The launch of ChatGPT in late 2022 created a pandemic of unoriginal thought — especially in academic institutions and specifically in the act of essay writing. In short, it birthed a massive population of student cheaters. Millions of students utilized the technology to write their essays for them, leaving education institutions perplexed on how to best tackle the issue.
Today, tools/software do exist to assist humans to differentiate between sentient versus non-sentient writing. But they can only provide a probabilistic judgement instead of a definitive one. As large language models become more sophisticated, rest assured that it will become increasingly harder to distinguish between sentient and non-sentient pieces of literature.
Outside of academia, the clash between sentient versus non-sentient writing can be felt very strongly in the entertainment industry. Hollywood writers were probably the earliest population of workers to vehemently protest against the advent of large language models. Although ChatGPT’s story writing ability is believed to not yet be on par with experienced human writers, the threat of potential unemployment and obsolescence is nevertheless felt very strongly in the Hollywood writing community.
Whether it be in academia or in the entertainment industry, literary authenticity and originality are being increasingly challenged by this much more efficient AI technology. If left unchecked, the entire writing profession may be at stake. But there is a general sense that humanity still has some time on its hands before monumental shifts in the labor market occurs.
The long-term fear is not in the eventual loss of the human touch in our literary works — but in our inability to distinguish between human and non-human literature. At some point, one can imagine a world where AI generated and human made art are effectively indistinguishable. Some may even argue that we are already living in such a world. Or that at the very least, ideals of maintaining the integrity and authenticity of human made art is soon becoming a thing of the past.
To this day, it is still unclear whether or not OpenAI really did steal Scarlett Johansson’s voice for their latest ChatGPT model. Some say that the company merely used a voice very similar to Johansson’s — but not necessarily hers. In a world of deepfakes and generative AI — anything can be digitally manufactured on a whim. A voice. A song. A picture. And indeed, this very article.
