EXPLAINER: What Is ChatGPT, Minus the Technical Jargon?

(SPOT.ph) It’s impossible to not have encountered the news and discourse on ChatGPT in the recent months, but it’s more than possible to get lost in everything that’s been said (or not said).

Since it launched in November 2022, ChatGPT has excited, confused, and scared people—sometimes all at once—and it’s still just in its beta version. Humans have long braced themselves for the rise of artificial intelligence, but it has never reached this level of mainstream before. What gives?

Also read: Artificial Intelligence and What It Might Mean for Your Future

What is ChatGPT?

Chat Generative Pre-Trained Transformer—or ChatGPT for short—is an artificially intelligent (AI) chatbot powered by OpenAI, whose purpose is to pass as a human conversationalist with the knowledge of the Internet as a brain. It can respond to any given prompt in a matter of seconds: think facts, mathematical equations, and hypothetical scenarios. Each answer will be a 100% original text based on the top information online.

What’s fascinating about ChatGPT is that it won’t simply answer in a Q&A format—it will be in the form in which you requested it to be. We’re talking ad copies, programming codes, movie scripts, news reports, and so much more. You can give a prompt with something as random as “write a poem on the dying cryptocurrency market” and ChatGPT will deliver: “As the market tumbles down / Cryptocurrencies hit the ground / Once worth a pretty penny / Now worth less than many.” How about a fictional rap battle between John F. Kennedy and Homer Simpson as a prompt?

Another remarkable feature ChatGPT has is its ability to remember previous prompts given to it in the same conversation, which means it can learn and adapt to the information in real-time. 

Everyone is free to use the unpaid version of ChatGPT, which can only be accessed when there’s little to no foot traffic on the platform. However, ChatGPT is almost always at full capacity. You want in during peak hours? You have to be willing to pay up. Subscribing to its paid version, ChatGPT Plus, will cost U.S. $20 per month, but even that has a waitlist to get in.

Why the panic?

AI chatbots like ChatGPT aren't new, with Meta’s BlenderBot 3 being introduced much earlier in August 2022. However, OpenAI’s ChatGPT is the first of its kind to offer its platform for public use, garnering a million users within five days of its official launch. Its massive success prompted other big players in the tech industry to join in on the fun. Emerging in wake of the ChatGPT craze are Google’s Bard and Bing’s ChatGPT. 

This has the world panicking over what this means for the future, especially in the field of education where students are using ChatGPT to write and pass full-fledged papers that can pass at a university-level. The same culture of fear is plaguing writers and journalists whose livelihood is threatened by the rising potential of ChatGPT.

In as early as 2020, The Guardian published an article entirely written by OpenAI’s earlier AI language generator, back when ChatGPT hadn't existed yet. Another freelance writer claimed ChatGPT wrote a marketing article in less than 30 seconds with a quality he would normally charge a client $600 for. According to CBS News, other industries at risk of being replaced by ChatGPT  are advertising, computer programming, content creation, and legal functions.

Also read: UP Diliman to Offer PhD Program in Artificial Intelligence

What’s wrong with ChatGPT?

Impressive as it is, ChatGPT is still lacking in many ways.

As mentioned earlier, ChatGPT will collect information from top sources on the Internet, but what is mined isn’t always the correct information. This is something that OpenAI admits to: “ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers.” In addition, the current version of ChatGPT will only have information until 2021 at the latest. Everything else that comes after will not be reflected in its responses.

Another thing to note: OpenAI has programmed ChatGPT to refuse prompts that are racist and sexist, as well as ones that ask about illegal activities. However, hackers and other bad actors have managed to set up a restriction-free version of ChatGPT. This is not good news for cybersecurity. Consider how easy it would be for someone to use ChatGPT to craft phishing emails with so little effort. In fact, ChatGPT will even instruct hackers where best to place a phishing link in an email.

What now?

ChatGPT has a lot to improve on, but the fact of the matter is that it can only go up from here. However, many AI experts believe that there’s no reason to fear AI as a whole, especially not a chatbot whose main function is to answer you. Meta’s chief AI scientist Yann LeCun was quoted telling the Big Technology Podcast that “ChatGPT is not a particularly interesting scientific advance” since it just churns “one word after another” based on what one can already find on the Internet. 

Don’t believe LeCun? Take it from the co-founder of OpenAI, Sam Altman, who expressed his confusion over the panicked response ChatGPT is receiving.

“ChatGPT is incredibly limited, but good enough at some things to create a misleading impression of greatness,” Altman wrote in a tweet. “It's a mistake to be relying on it for anything important right now. it’s a preview of progress; we have lots of work to do on robustness and truthfulness.”

ALSO READ:
Singaporean A.I. Predicts When The Coronavirus Will End
Seoul is Using AI and CCTV to Prevent Suicide Attempts on Bridges