A new term called ChatGPT is making waves in the IT industry. On December 1, the system went live for public beta testing. After only one week of operation, ChatGPT has surpassed 1 million users. There are rumours that the platform would eventually replace Google search and, more importantly, people.
A conversational chatbot, ChatGPT. It was created by OpenAI, an independent research organisation formed by Elon Musk. Elon Musk and Sam Altman, the company’s current CEO, co-founded it in 2015.
Returning to ChatGPT, it has been taught to follow a prompt’s instructions and give a thorough response. Users only need to enter their question, and the chatbot will respond. But how does it differ from other chatbots that use artificial intelligence (AI)? The developers claim that ChatGPT, unlike other AI chatbots, has the ability to respond to follow-up queries, acknowledge mistakes, refute false premises, and reject unsuitable requests.
ChatGPT is a free service according to OpenAI, but only while the research preview is in effect. This indicates that using the platform is as simple as going to the OpenAI website and clicking the Try ChatGPT button. To utilise the ChatGPT, you can either create an account or log in with your OpenAI credentials. Additionally, the business has provided a sample for reference on the website.
According to OpenAI, this model has been refined and made more conversational using Reinforcement Learning from Human Feedback (RLHF). We used supervised fine-tuning to train an initial model, which involved human AI trainers acting as both the user and the AI assistant in discussions. We provided the trainers with access to sample written recommendations to assist them in creating their responses “it states.
As previously stated, ChatGPT is only free to use during the research phase. Sam Altman, the CEO of OpenAI, has already made hints that the business would eventually make money off the platform. The computational expenses are eye-watering; at some point, we will have to monetize it.” “When questioned if the service would always be free, he responded in a tweet.
The newest member of the Generative Pre-Trained Transformer (GPT) family is ChatGPT. It is the most recent AI technology for automatically generating text, to put it simply. However, it is not without flaws or restrictions. On its website, OpenAI acknowledges that ChatGPT occasionally generates plausible-sounding but mistaken or absurd responses. The model also utilises specific words and phrases far too frequently. The chatbot is also attentive to the wording of the input. For instance, it might know the response to a question asked in one style, but if the question is asked in a slightly different way, the model might not know the response.
We all use Google excessively to look up the answers to our daily inquiries. Can we trust ChatGPT to provide the same. Even though the model uses the most recent technologies, it has several drawbacks. Professionals are highlighting the platform’s drawbacks. For instance, UC Berkeley researcher Steven T. Piantadosi demonstrated in a series of tweets how ChatGPT contains biases and that its filters are easily circumvented using straightforward techniques. Therefore, the basic answer to the question, “Can ChatGPT replace Google search or humans?” is no. at least right now.
