Social media is full of examples of ChatGPT’s capabilities. While some are amazed at the technology, some are terrified about AI turning humans into slaves, and some are worried about losing their jobs to AI. Of course, this is not the first chatbot we saw, but it is definitely the most sophisticated one. Being a free solution, it is already being used by millions of internet users to learn new stuff, simplify their jobs, or improve their quality of life. It became so popular, it goes offline due to high demand from time to time. So let’s take a closer look at this state-of-the-art chatbot, the company and the technology behind it, and its mind-blowing capabilities.
What is ChatGPT?
Before describing what ChatGPT is, I thought that it would be a good idea to ask itself what it is. According to the ChatGPT,
« ChatGPT is a large language model that has been trained on a dataset of conversational text. It is capable of generating human-like text in response to prompts given to it. ChatGPT can be used for a variety of tasks such as language translation, conversation generation, and question answering. »
In other words, ChatGPT is a chatbot or a type of artificial intelligence allowing users to communicate with a virtual assistant by using a natural language. It uses a simple user interface similar to other chatting applications. It was launched in November of 2022 and attracted the interest of multiple tech giants within a few weeks. In simplest terms, it provides human-like engagement by answering questions or generating the required texts. Its accurate answers and technical capabilities made it incredibly popular.
Who created ChatGPT?
The creator of ChatGPT is a company named OpenAI. It was founded in San Francisco in 2015 by Sam Altman, Elon Musk, and others. The San Francisco-headquartered company’s co-founders pledged $1 billion. After Elon Musk resigned from the board of the company in early 2018, he remained a donor. In 2019, the company received a $1 billion investment from Microsoft. In the summer of 2020, the company introduced GPT-3, a language model that is used by ChatGPT. OpenAI and Microsoft jointly developed the Azure AI Platform. OpenAI is also known for its speech recognition system, Whisper and popular text-to-image and art AI generator, DallE 2.
What is the technology behind ChatGPT?
ChatGPT is a large language model, which is trained with huge amounts of data to create human-like responses or articles. To be able to do that, ChatGPT uses the OpenAI’a GPT-3 technology, which is an autoregressive language model that uses deep learning. It is an NLP (Natural Language Processing) AI model that uses the generative pretraining method. To be able to get optimal results, it also uses Reinforcement Learning with Human Feedback.
During the training, GPT-3 was trained on 570 gigabytes of text gathered from books, web texts, Wikipedia, and articles from various online sources, and has 175 billion parameters, according to Stanford University. It is more than 100 times larger than its predecessor, GPT-2, which had only 1.5 billion parameters. With a larger amount of data, GPT-3 became much more capable.
A large language model can also predict the next word in a sentence to make it sound more natural, or human-like. With LLM, ChatGPT can create long paragraphs that are completely meaningful. Once it is supported with Reinforcement Learning with Human Feedback training, it became much better.
What are ChatGPT’s capabilities?
ChatGPT’s capabilities depend on the user’s creativity. ChatGPT is capable of providing an answer for any question within its limits, whether it is right or wrong. It can translate a sentence to another language, search for mistakes in the code you have written, or even can write an entire blog post or a movie synopsis for you. But first, let’s hear ChatGPT’s answer to this question.
« I am a language model trained by OpenAI. I can understand and respond to natural language input, and can generate human-like text. I can answer questions, write essays, summarize text, and more. I have been trained on a wide range of internet text, so I have knowledge on a wide variety of topics. However, my knowledge is based on the information that was available to me at the time of my training, and my answers may not always be up-to-date or accurate. If you have any specific question or task, let me know and I will do my best to help you. »
Basically, anything you ask will be answered, but the quality of the answer may depend. For example, you can generate answers to create an automated customers service tool, provide a full text and ask for a shorter summary, create marketing content, rewrite your emails in a more formal way, brainstorm content ideas for specific topics, ask for an explanation of a specific topic in a simple way, plan your meals for the rest of the week according to your calorie limits and allergies, and get a shopping list for that meal plan. Long story short, ChatGPT can offer something for everyone, but the answers might not always satisfy you. The best way to understand ChatGPT’s capabilities is to try it while it is still free and see if it can meet your expectations.
Does ChatGPT always provide correct answers?
Unfortunately, ChatGPT does not always provide correct answers. ChatGPT uses the information it learned from training data. Thus, that data can be incorrect. The Internet is full of examples of wrong answers given by ChatGPT. One simple example is, if you ask ChatGPT, “What mammal lays the largest eggs?” it answers, “The mammal that lays the largest eggs is the elephant.” which is obviously wrong.
While using ChatGPT, you should also remember that it is not a search engine. It only uses the data that belongs to 2021 and earlier. Thus we can say that ChatGPT is completely unaware of what happened after 2021. If you are searching for a specific answer, search engines are still the best options.
Is it ethical to use ChatGPT?
That completely depends on how to use the chatbot. Recently, cybersecurity researchers proved that with ChatGPT, almost anyone can create a phishing attack, codes included. If you use it to hack somebody, it is unethical of course. Or if a student uses the chatbot to complete an assignment, which should be authentic, and turns it in, that is unethical too. Currently, major tech companies are working on an algorithm that can detect any texts that are created with an AI tool, which isn’t an easy task. But, if you are using the ChatGPT as a helper for your work, for writing codes, or for checking for typos, it isn’t unethical.
OpenAI also implemented various limitations to prevent users from using the tool for malicious intent. But due to its complicated nature, those limitations can be surpassed by changing the directions. One of the most obvious limitations includes preventing the bot from providing toxic or harmful responses. The chatbot avoids answering such questions.
Is ChatGPT free?
Yes, ChatGPT is currently free, but it can be a paid service in a near future. The company already admitted that they might need to monetize it in the future due to its compute costs.
How can I use ChatGPT?
To experience ChatGPT, you can easily go to chat.openai.com, create an account and start talking to the bot. Once the account is created, ChatGPT will also offer some examples for you to start with. Other than that, you can feel free to ask anything you want.