You have probably interacted with Siri, Alexa, Cortana or other chatbots recently, and while they manage to do specific tasks well, you have probably wondered if there would be a more usable/knowledgeable answer engine that not only recognises your previous requests but also understands and answers all of your questions in whatever tone you want it to.
Enter ChatGPT. The artificial intelligence world was taken by storm on Wednesday when generative AI behemoths Open AI launched a free trial version of their latest endeavour.
Chat GPT is an experimental chatbot based on an autocomplete text generator GPT-3.5. It can write jokes (some of which are actually funny), working computer code and college-level essays. It can also guess at medical diagnoses, create text-based Harry Potter games and explain scientific concepts at multiple levels of difficulty.
Let’s say you input the request “write the plot for two episodes of Seinfeld where they discover ChatGPT?” the following comes up:
“In each episode, the friends encounter a problem or situation and turn to ChatGPT for advice. However, ChatGPT’s responses are always comically misinterpreted by the friends, leading to even more chaos and hilarity.
“In the pilot episode, Jerry and Elaine are trying to decide where to go for dinner. They ask ChatGPT for suggestions, but ChatGPT responds with a detailed description of the history of the restaurant industry. Jerry and Elaine take this to mean that ChatGPT wants them to go to a restaurant that’s been around for a long time, so they end up at a stuffy old diner where they are the only customers.
“In the second episode, George is trying to impress a girl he met at the coffee shop. He asks ChatGPT for advice on what to say, but ChatGPT responds with a lengthy monologue on the intricacies of human communication. George, misunderstanding the advice, ends up reciting the monologue verbatim to the girl, who is thoroughly confused and unimpressed.
“Throughout the series, the friends continue to seek out ChatGPT’s advice, but always end up in comedic mishaps because of their misinterpretations.”
Obviously, this level of creativity and understanding has a lot more applicable uses.
While this is just a trial phase of the program, this chatbot may have the potential to replace Google.
It is important to note that it currently doesn’t have access to the internet to pull out answers, rather it is a fancier autocomplete generator that has been trained on data up until 2021.
This means that any prompts about current events is something the bot struggles with.
Currently, it is free to use and can be accessed here. But founders and lead investors Sam Altman and Elon Musk have said that it won’t stay that way for long.
As a side note, there are many things that ChatGPT refuses to do that it terms “inappropriate requests”. These encompass pre-programmed conversations that veer anywhere around lewd and/or racist remarks, as well as illegal activity. However, users have found a way to bypass these restrictions by rephrasing a question as a hypothetical thought experiment or a scenario in a fictional universe.
There have been concerns that certain careers that involve creating content, such as playwrights, professors, programmers, undergraduate student essays and journalists, could become obsolete.
After its release, some academics claimed that the tool could produce responses to exam questions that would earn full marks if submitted by a student. Additionally, programmers have used the tool to quickly solve coding challenges in even the least popular programming languages. The technology’s ability to generate human-like written text has led to suggestions that it could also replace journalists.
But it still has major (albeit infrequent) shortcomings. It is prone to giving out incorrect information confidently and assertively. These systems are trained by analysing patterns in huge reams of text scraped from the web. They look for statistical regularities in this data and use these to predict what words should come next in any given sentence.
This means, though, that they lack hard-coded rules for how certain systems in the world operate, leading to their propensity to generate “fluent bullsh*t”. This was particularly seen when Stack overflow, the leading Q&A platform for software engineers, banned this program as a way to answer difficult coding questions.
Altman has come out and said that this is still an early demo of what is possible. After the feedback round this month, the team would be working on a better GPT framework.
“Soon you will be able to have helpful assistants that talk to you, answer questions, and give advice. Later you can have something that goes off and does tasks for you. Eventually, you can have something that goes off and discovers new knowledge for you,” he told The Guardian.