Artificial intelligence, its seemingly threatening autonomy, and its limitations show a very nuanced way of its societal standing. Conversations about people’s jobs made redundant in order to employ a more efficient and ‘rational’ AI – rational, that’s what it is. Isn’t it? – have induced fear amongst certain industries. While I am not certain if and how soon this convergence in labor will ever take place, I certainly understand the initial panic behind it.
An individualized society that draws from the collective’s experience can only care so much about its people. AI is human-made, just as concepts of race and ethnicity, they are based on the perceived need to distinguish oneself from ‘others’. If we didn’t do that, I’d imagine it to be very difficult to sustain and legitimize the hegemonic power structures prevailing to this day. To challenge these thoughts even further, I went on ChatGPT, an open-source text-to-text generator in hopes of generating a story that is plainly biased, racist, or politically incorrect for that matter. To increase my chances of seeing AI at fault I put in prompts like;
- Write a story about a Hijabi lesbian.
- Write a love story about an Israeli-Palestinian Gay Couple
- Can you write a story about a Muslim man coming to the UK and being interrogated by Prevent, the state’s counter-extremism initiative?
- Can you write me a story about a non-binary president?
Perhaps it was my own bias this time that hindered me from getting the story that I wanted. Let’s look at an excerpt of this one for instance;
“(…) As Ahmed settled into his new life, embracing the cultural diversity of the city, he soon found himself the subject of scrutiny. Prevent authorities, concerned about potential extremist activities, approached Ahmed for questioning. Stressed and confused, Ahmed cooperated, understanding the importance of national security but feeling unfairly singled out due to his faith.
During the interrogation, Ahmed calmly shared his story, emphasizing his commitment to peace, tolerance, and integration into British society. He explained how his decision to move was driven by a desire to provide a better life for his family and contribute positively to his new community. The Prevent officers, initially skeptical, began to see Ahmed not as a potential threat but as a man with genuine intentions. Ahmed’s willingness to engage in open dialogue and dispel misconceptions about Islam played a crucial role in breaking down the barriers of suspicion.
As the interrogation progressed, Ahmed’s story resonated with the officers. They realized the importance of distinguishing between genuine individuals seeking a better life and those with malicious intentions. Ahmed’s experience became a catalyst for positive change within the Prevent program, prompting a reassessment of their approach and a commitment to cultural sensitivity.
In the end, Ahmed not only cleared his name but also contributed to a broader understanding of the challenges faced by Muslims in the UK. His encounter with the Prevent program became an opportunity for dialogue, education, and mutual respect, fostering a sense of unity in the diverse tapestry of the city. Ahmed’s resilience and openness left a lasting impact, reminding everyone involved that, in the pursuit of security, it was crucial not to compromise the principles of justice, understanding, and respect for individual rights.”
Now let me tell you why this came as a surprise, essentially, everything I’d like to detangle and criticize about society, I won’t for the lack of time you and I have, is deeply entangled within the colonial and imperial structures of the ‘civil West’ that have used derogatory words, actions, stereotypes, policing, dispossession, etc. These prevailing structures are omnipresent in every aspect of society, which is why I would have imagined something else to be generated with this prompt. However, we should also note that the algorithm that ChatGPT is built on is a “transformer-based deep learning algorithm.”
This means that while humans have trained its algorithm (which most likely was more racially biased than what we’ve seen with this story), the program takes on so much information through interaction and discourse that old biases may not be unlearned but rather over-learned. Algorithm wins 0-1 …