GPT-4 Breakthrough : Ambivalence of AI technology

As humans seek more and more convenience, artificial intelligence has constantly changed and developed to meet their needs. As a result, an innovative language model, ‘Chat GPT’ has recently appeared in the world. Chat GPT stands for Generative Pre-trained Transformer and is an AI chatbot specializing in conversations.

Sam Altman, an Open AI Ceo, said, ‘It will reorganize society and is dangerous.’ However, we should first check the various uses of this AI technology rather than anxiety and worry.

Along with the development of Chat GPT, Open AI has released GPT-4, which handles more extensive content and is more upgraded than the original version. GPT-4 adds the ability to recognize, understand, and process images, which allows you to analyze graphs or solve problems. The key takeaway is that GPT-4 can now analyze pictures and text. In addition, GPT-4 has been challenged with various human exams, and the results show that it has no problem explaining test questions and is expected to be very helpful as a learning tutor. Trained on years of test data, GPT-4 is better at solving test questions than any teacher and can analyze them quickly. “GPT-4 is a work in progress and evolving, and we expect to see more uses and applications than we can imagine.” said Open AI CEO Sam Altman.

So what problems arise from AI technologies like GPT-4, which have so much potential?

Indeed, AI technologies like GPT-4 are highly regarded for their utility and commercial use. However, GPT-4 scored well in all tests requiring reading and memory skills but poorly in math and writing tests requiring complex thinking. Also, GPT-4 isn’t current with training data through September 2021, so it cannot provide users with the correct information. 

Next, there are ethical issues with GPT-4. In the case of the uncensored GPT-4(early), there was a problem with giving specific answers to questions about various crimes. In conclusion, while AI technology has surpassed other technologies in terms of its versatility, there are still many challenges for humans in using AI technology. 

Shortly, when incredibly sophisticated AI technologies are introduced to the world, will human life be enriched or become more competitive? What is clear is that we humans will be responsible for the outcome.

OpenAI, GPT-4 Technical Report, https://arxiv.org/pdf/2303.08774.pdf