Last semester’s Project work became increasingly challenging towards the end. I was asking my classmates for help regarding MongoDB and Mongoose. However, I was really getting annoyed by their responses, which were all focused on this new technology called “ChatGPT.”
Despite my initial reluctance, a good friend of mine convinced me to give it a try claiming it to be a beautiful tool.
Believe it or not, I now use ChatGPT more than Google!
Why is that?
First of all, as a computer science student, I was completely amazed when I started using ChatGPT. From giving step-by-step instructions for installing Mission Planner to writing CSS files for styling my HTML pages, ChatGPT assisted me with almost everything.
Talking to ChatGPT is just like having a conversation with another human being. It has knowledge in every domain and gives you answers relative to your previous questions. Although many websites have been using ‘chatbots’ to answer the queries of their users for a long time, OpenAI’s ChatGPT is the next advancement in this technology. When this product hit the market, it was such a huge success that competitors like Google, Amazon, Deepmind and other companies started rapid developments in their chatbots.
For those unfamiliar, ”GPT” in ChatGPT stands for Generative Pre-trained Transformer. GPTs are a family of neural network models that use transformer architecture to engage in conversations and generate human-like texts. [1]
ChatGPT with its advanced searching, compiles the information as requested. This new platform offers a high level of artificial intelligence which generates unique responses using the latest Natural Language Generation (NLG) Technology.
At first, ChatGPT may appear powerful but a detailed analysis will quickly point out its flaws – many of which the tool itself will admit.
As an AI model trained on a diverse range of online content, ChatGPT can be biassed according to the data it learned from. This can replicate societal prejudices and amplify misleading information, thereby contributing to the spread of false narratives.
Another drawback is that the model training only covers the period until September 2021. Therefore it does not include the most recent information on the latest developments, current events, or real-time data.
Chat GPT may be quick to find answers, but those answers may not always be correct. From what I have experienced, ChatGPT is bad at Maths, especially in the concepts of Permutations and Combinations. Thus, relying on Chat GPT exclusively for any task is not advisable.
To test the ability of ChatGPT, it was made to attempt many challenging exams as well. It passed the NEET, with an impressive score in Biology. On the other hand, it failed to crack JEE where it could solve only 11 questions. It could not pass the UPSC civil service test as well.[3]
Although ChatGPT will provide you with the codes you ask for, they may not always be correct. Sometimes, we would have to sit for hours together to debug the codes it generated. People claim ChatGPT will take over the software jobs, but it still has a long way to go.
Apart from this, ChatGPT falls short in one critical aspect—The human touch. While it can generate content by synthesising information online, it cannot think, reason, or offer genuine opinions.
Overall, the advancements in AI have raised ethical concerns regarding accountability and transparency. Since ChatGPT operates based on patterns and algorithms, it is difficult to understand how decisions are made or to hold it accountable for its actions.
While ChatGPT offers impressive capabilities, its limitations must not be overlooked.
Striking a balance between its benefits and drawbacks is essential for experiencing the beauty of this tool.
REFERENCES:
[1] https://openai.com/blog/chatgpt
[2] “Generative AI: a game-changer society needs to be ready for”. World Economic Forum.
[3] ChatGPT attempts JEE Advanced exam. Here’s how much it scored
Leave a Reply