OpenAI ChatGPT went viral
The most advanced AI chatbot launched on Nov 30th 2022. Some say it’s better than Google, others worry it’s problematic.
ChatGPT could be a good debugging companion; it not only explains the bug but fixes it and explain the fix 🤯 pic.twitter.com/5x9n66pVqj
— Amjad Masad ⠕ (@amasad) November 30, 2022
What kid is ever doing homework again now that ChatGPT exists pic.twitter.com/oGYUQh3hwh
— Liv Boeree (@Liv_Boeree) December 1, 2022
“A new chatbot that’s captivated the internet can tell you how to code a website, write a heartfelt message from Santa Claus and talk like a Valley girl. But it’s also proven to be potentially as problematic as it is entertaining”, noted by nbcnews.
“The dialogue format makes it possible for ChatGPT to answer followup questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests,” the website states.
What is ChatGPT?
ChatGPT is based on GPT-3.5, a language model that uses deep learning to produce human-like text. However, while the older GPT-3 model only took text prompts and tried to continue on that with its own generated text, ChatGPT is more engaging. It’s much better at generating detailed text and can even come up with poems. Another unique characteristic is memory. The bot can remember earlier comments in a conversation and recount them to the user.
So far, OpenAI has only opened up the bot for evaluation and beta testing but API access is expected to follow next year. With API access, developers will be able to implement ChatGPT into their own software.
What are the limitations?
But while many people were in awe of the abilities of the bot, some were also quick in spotting its limitations. ChatGPT is still prone to misinformation and biases, which is something that plagued previous versions of GPT as well. The model can give incorrect answers to, say, algebraic problems.
And due to the fact that it appears so confident in its super-detailed answers, people can easily be misled into believing those are true.
OpenAI understands these flaws and has noted them down on its announcement blog: “ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers. Fixing this issue is challenging, as: (1) during RL training, there’s currently no source of truth; (2) training the model to be more cautious causes it to decline questions that it can answer correctly; and (3) supervised training misleads the model because the ideal answer depends on what the model knows, rather than what the human demonstrator knows.”
Limitations aside, ChatGPT still makes for a fun little bot to interact with. You can try it out from its official website after signing up for the same.
