CHATGPT : THE MASTERMIND THAT MARKS A BRAND-NEW ERA

Timmy Zhang

Tentatively tapping the enter button, you expect the AI to emit awkward sentences resembling a 4-year-old’s writing; the result catches you off-guard. This AI presents a full-length description of a galloping unicorn with radiant colors. Here is ChatGPT, a revolutionary chatbot that answers questions based on a blend of data collection and machine learning.

The learning model that ChatGPT employs makes it powerful. At first, human AI trainers provided ChatGPT with prompts like “Explain who Taylor Swift is” and then spoon-fed it the desired answers. Like a simple-minded toddler, the initially unskilled ChatGPT produced different responses, which the AI trainers ranked. Then, after many trial-and-errors, AI trainers encouraged ChatGPT to walk on its own; it independently answered questions using Proximal Policy Optimization, an algorithm that can be easily tuned and implemented. With such training, ChatGPT demonstrated how practice makes perfect; its witty responses to questions gathered a staggering 100 million active users in January.

This AI might be the key to boosting workspace productivity and quality of education. ChatGPT can tackle some arduous yet challenging tasks with its critical thinking and data-collection proficiency. Professor Erik Brynjolfsson, director of Stanford University’s Digital Economy Lab, comments that the future will not be “AI replacing lawyers” but replacing “lawyers who don’t work with AI” with those who do. Indeed, ChatGPT can accurately extract information from large datasets of case laws and jurisdictions, and it outsmarts Google by providing instant, concise answers. Regarding education, ChatGPT can grade essay exams, especially those in the liberal arts, and evaluate school admission applications that require multiple variables analysis. In both ways, ChatGPT assists workers with the most brain-intensive tasks by solving smaller but still challenging problems. Despite its “big brain,” ChatGPT does have its downsides. First, this AI will sometimes provide responses that, while highly logical, make no practical sense. For example, when you ask, “What will you do if a flood puts me in imminent danger?” it will elaborate a paragraph about evacuation protocols rather than shouting, “I WILL RUN!” It also cannot answer questions outside the datasets it has studied: if a person asks about trending video games, ChatGPT, assuming it has only studied datasets about cats, will likely respond with kittens playing with catnip. In addition, individuals can manipulate this AI for malicious purposes. For example, ChatGPT can be used for political lobbying by commenting on newspapers, blog entries, and social media posts to spread false information, acting like internet bots from the Russian Internet Research Agency in the U.S. 2016 election. With its defects, ChatGPT may not be the safe, omnipotent problem-solver that many expect.

Whether you think ChatGPT is a blessing or curse, one fact remains unchanged: from composing poems to programming independently, ChatGPT can stand proudly and proclaim itself the leader of world-changing AIs. Let us open our computers to see how ChatGPT shapes our future.

Bibliography

[1] ChatGPT: Optimizing Language Models for Dialogue, openai.com/blog/chatgpt/.

[2] Schulman, John, et al. Proximal Policy Optimization, openai.com/blog/openai-baselines-ppo/.

[3] Hu, Krystal. ChatGPT sets record for fastest-growing user base - analyst note, 2 Feb. 2023, www.reuters.com/technology/chatgpt-sets-record-fastest-growing-user-base-analyst-note-2023-02-01/.

[4] Kern, Josh. AI in Law: Transforming Legal Practice, 31 Jan. 2023, www.clio.com/blog/lawyer-ai/.

[5] Paresh M Patel , www.pareshmpatel.com/pros-and-cons-of-using-chat-gpt/.

[6] Sanders, Nathan E., and Bruce Schneier. How ChatGPT Hijacks Democracy, The New York Times, 15 Jan. 2023,

www.nytimes.com/2023/01/15/opinion/ai-chatgpt-lobbying-democracy.html?searchResultPosition=3.