OpenAI has released a new LLM(Large Language Model), GPT-4 on March 15th. Unlike its previous model, which was mocked for ‘making up’ the historical event that King Sejong the Great threw the MacBook Pro towards his opposing lieges while creating Hunminjeongeum(훈민정음, the Korean alphabet), the latest one improved its Hallucinations by a considerable amount.
Besides the LLM stated above, Artificial Intelligence is leading the fourth industrial revolution by being used in various departments including medical science and astronomy. It does shine like a diamond and sounds like promising us the brightest future. However, we need to take a break from these rose-colored spectacles, since the most significant danger lies underneath the most beautiful-looking places, just like in the movie 〈Metropolis (1927)〉.
〈Coded Bias〉, a documentary directed by Shalini Kantayya, shows the uneasy truth of Artificial Intelligence clearly. This film follows Joy Buolamwini, a researcher working in the MIT media lab, and her international colleagues’ journey that starts from a ‘technical’ error in Facial Recognition technology. Joy finds that Facial Recognition software struggles to recognize her darker-toned skin regardless of multiple trials, however, it recognizes a ‘human face’ right away after she puts on the white-colored mask; thus, it points out the danger of trying to find the ‘correct answer’ based on antiquated, biased data.
For eighty-six minutes, the director’s camera stretches out from Silicon Valley to China to capture the moments of human rights getting violated with the help of Artificial Intelligence. Amazon.com Inc's Artificial Intelligence, AMZN.O, which trained more than 10 years’ worth of human resource data ‘penalizing’ and ‘downgrading’ women candidates’ resumes, Metropolitan Police questioning children under the name of surveillance, and the Chinese government using an active reward system to make citizens obey the CCP’s ideology.
This series of sequences, which inversely reflects the marginalization and prejudice in the information and data, make us ruminate about the quote ‘History is Written by the Victors.’ Considering that the information and data have been processed and weaved by the people with power and privilege, a modern paradigm of web 3.0, based upon them, does not coincide with the original intention of the World Wide Web as the creator Tim Berners-Lee has actively stated several times.
Putting the train of thoughts back to ChatGPT’s lies and Facial Recognition’s racism, the matter of ‘trusting’ Artificial Intelligence continues to the matter of the attitude of a human being who is responding to it. In the interview with The Guardian, Meredith Broussard has already answered this question, by saying that “we can demand that our technologies work well – for everybody – and we can make some deliberate choices about whether to use them.” Therefore, the only thing we can be sure of among these chaotic prospects filled with praises and remorse is that a better future is in the hands of citizens who are willing to make changes with Artificial Intelligence, not Artificial Intelligence itself.
References
1. ChatGPT: What Are Hallucinations And Why Are They A Problem For AI Systems
2. Amazon scraps secret AI recruiting tool that showed bias against women
3. OpenAI’s introduction page of ChatGPT
4. ChatGPT can’t lie to you, but you still shouldn’t trust it
5. ‘Generative AI, you're a bit tricky’: Review on Bing AI, DuckAssist's ‘sourcing’ feature
6. As Web turns 25, its creator, Tim Berners-Lee, calls for a ‘Magna Carta’
7. AI expert Meredith Broussard: ‘Racism, sexism and ableism are systemic problems’
コメント