“I am ChatGPT, a large language model created by OpenAI. I have been trained on a vast amount of text data using deep learning algorithms, which allows me to understand and generate human-like language in response to text-based queries. You can think of me as a sophisticated chatbot capable of answering a wide range of questions and engaging in natural language conversations.”
It was launched in November 2022 and signed up 1 million users within 5 days, 100 millions users in 2 months and has continued in increased popularity.
ChatGPT is like a digital genie that can grant unlimited word-based wishes. It works as your personal assistant and researcher that has read 11 billion pages to base their answers on. It generates answers, reports, summaries, software code and documents on almost any question in whatever style and format you wish for.
The naming of "ChatGPT" may suggest simplicity and just for fun use. The name does not suggest the level of sophistication available from it's responses to very difficult and complicated questions and the ability to get it to perform informational tasks for you is amazing. You need to push ChatGPT to it's limits in usefulness as a productivity tool by getting it to answer questions that would take a long time to answer if you searched across webpages using Google and to write documents for you just by giving it very few words and letting it interpolate the rest.
“ChatGPT works using a machine learning technique called deep learning, specifically a type of deep learning called transformer-based models. These models use neural networks to analyze and understand natural language text data and then generate responses based on that understanding.
To be more specific, ChatGPT uses a type of transformer-based model called a generative language model. This means that it is designed to generate text based on a given prompt or input. The model is trained on a vast amount of text data, such as books, articles, and other sources, and it uses this training data to learn patterns and relationships in language.
When a user inputs a question or prompt, ChatGPT’s algorithm processes the text and generates a response based on its understanding of the language. The response is not simply a pre-written answer, but rather a unique response generated by the model based on its learned patterns and relationships in language.
As a language model, ChatGPT is constantly learning and updating itself based on new data and user interactions. This allows it to improve its accuracy and generate more human-like responses over time.”
As a concept it takes the words of 11 billion pages plus the words of your prompt/request and works out what is the most likely first word as a response. Then it works out the most likely second word, third word, fourth word, etc until you’ve got a full response. So it doesn’t “know” anything but it can create the most likely response based on what everyone else has already written. And it does it super quick due to phenomenal processing power and in a way that makes it sound intelligent.
You can sign up to ChatGPT at https://chat.openai.com/
You'll need to register with an email, phone number and password.
Warning: there is no App for it, they are either fakes or scams.
It's free to use online as a chat system on the OpenAI website but you'll need to register and you get better response times if you pay $20/month for "ChatGPT Plus".
People often ask “Can I use the output from ChatGPT licence free?”, “Can I use the output from ChatGPT without copywrite?” or “Can i use chatgpt output for commercial use?” The answers here explain how OpenAI sees it. So who does ultimately owns content generated By ChatGPT?
Here is ChatGPT's answer:
"As an AI language model, ChatGPT generates text output based on its understanding of natural language and its training data. The output itself is not subject to copyright or ownership, and you are free to use it for your own purposes.
However, it's important to note that the use of ChatGPT's technology may be subject to certain terms and conditions set by the application or platform that provides access to the model. For example, some platforms may require attribution or prohibit the use of the model for certain types of content.
Additionally, if you are using ChatGPT's output to create content for commercial purposes, such as marketing materials or advertisements, you should exercise caution and ensure that your use of the output is legal and does not infringe on any copyrights or trademarks."
"As an AI language model, ChatGPT's answers are generated based on its understanding of natural language and the patterns it has learned from its training data. While the model is designed to generate responses that are as accurate and relevant as possible, it is not always guaranteed to be 100% correct.
The accuracy of ChatGPT's answers depends on a variety of factors, including the quality of the input text, the complexity of the question or prompt, and the amount of relevant information available in the training data. Additionally, like any other machine learning model, ChatGPT may be subject to biases or inaccuracies in the data it has been trained on.
It's important to keep in mind that ChatGPT is a language model and not a human expert. While it may be able to provide helpful information or insights on a wide range of topics, its responses should be evaluated critically and should not be considered authoritative or conclusive without further research or verification."
There are multiple instances of ChatGPT being wrong so if it's important information and/or you don't know if it's right when you read it then double-check it before use.
"As an AI language model, I don't have the ability to retain or recall information from previous chat sessions on my own. However, some chatbot platforms or applications that use my technology may have the ability to store previous chat logs and allow users to reference them later.
If the platform or application you are using has this feature, you may be able to access your previous chat logs and reference them during a new chat session. This can be helpful if you want to refer back to previous information or continue a conversation from a previous chat.
It's important to note that the ability to reference previous chats may depend on the specific platform or application you are using and the settings or permissions associated with your account. Additionally, if the information you are referencing is sensitive or personal in nature, you should exercise caution and ensure that you are not violating any privacy or security policies."
If the text is visible on one screen then just select it and copy and paste as normal. If it's over one screen length then you should select as much as you can from the start of the text you want, then use the Shift key and the up/down/left/right arrows to change the selection before copying and pasting.
To copy a table into Excel you will need to select the whole table plus an extra line of text above and/or below the table before copying and pasting into Excel then removing the extra lines in Excel.
"It's worth noting that generating a long piece of text may not necessarily result in a high-quality or coherent essay. While I can generate text that is grammatically correct and logically structured, there are limits to my ability to generate text that is cohesive, engaging, and persuasive, which are important qualities of a well-written essay."
Use the prompt of “Analyze the following text for style and tone of voice. Apply that exact style and tone of voice to all your future responses.” then paste in some typical text you use.
"ChatGPT has been trained on a vast amount of text data from a variety of sources, including books, articles, websites, and other online sources. The specific sources of the data used to train ChatGPT vary depending on the version of the model, as well as the training dataset used.
The original version of ChatGPT, GPT-1, was trained on a large dataset of web pages, while subsequent versions of the model, such as GPT-2 and GPT-3, have been trained on much larger and more diverse datasets. These datasets include a wide range of sources, such as books, news articles, academic publications, and social media posts.
It's important to note that the sources of the data used to train ChatGPT are not selected by humans, but rather by algorithms that are designed to gather and process large amounts of text data automatically. This means that the model is trained on a diverse and comprehensive dataset of text, but it also means that there may be some biases or inaccuracies in the data that the model learns from."
"As an AI language model, ChatGPT has been trained on multiple versions and datasets, each with its own specific training period. The specific cutoff date for each version of the model depends on the dataset used and the time period during which the data was gathered and processed.
For example, the original version of ChatGPT, GPT-1, was trained on a dataset of web pages collected in 2015, while subsequent versions of the model, such as GPT-2 and GPT-3, have been trained on much larger and more diverse datasets collected over a longer time period.
The most recent version of ChatGPT, GPT-3, was trained on a dataset that includes web pages collected up until April 2023. This means that the model's training data includes information up to that point in time, but may not include very recent or up-to-date information. Additionally, it's important to note that the model's understanding of specific events or developments may be limited by the data available in its training dataset."
The latest data currently processed is from April 2023.
Google is soon to be launching its competing AI chat system to ChatGPT, to be called "Bard"
Microsoft is planning on merging ChatGPT into its Bing search engine.
"ChatGPT is a product of OpenAI, a research organization that focuses on developing cutting-edge artificial intelligence technologies in a safe and beneficial way. OpenAI was founded in 2015 by a group of technology leaders, including Elon Musk, Sam Altman, and others, and has since become a leading research institution in the field of artificial intelligence. However, it's important to note that ChatGPT is a language model and not a physical entity that can be owned by anyone."
GPT-4 is more sophisticated but slower than ChatGPT 3.5. If you're doing something that requires standard responses then use 3.5 as it's quicker and won't make much difference. However, if you're needed more complex or creative answers then use version 4.
ChatGPT 4 also now allows creation of images directly from within the prompt (if you select the Dall-e option) and analysis of files eg PDF and Excel (if you select the Code Interpreter option).
There are a lot of fake or scam versions of Apps for ChatGPT so be very careful not to use them. There is now an official Android version at https://play.google.com/store/apps/details?id=com.openai.chatgpt&hl=en&gl=US and one for iOS here: https://apps.apple.com/app/openai-chatgpt/id6448311069
OpenAI now has a special page with guidance and examples for teachers and lecturers with example prompts:
Can you have a private area of ChatGPT where your data is kept confidential and not integrated into the model?
Yes, you can use the Enterprise version to do this. It keeps your data private and separate from everyone else's data so it doesn't get shared or revealed to other people or your competitors:
Yes, it's called "Custom Instructions" and you go there by pressing on the account/menu button and selecting Custom Instructions. You then have two places to enter instructions to be used on all new prompts.
(1) "What would you like ChatGPT to know about you to provide better responses?" eg "Use UK English. Located in the UK."
(2) "How would you like ChatGPT to respond?" eg explain the style, format, background, expertise to include every time.
|Open new chat
|Ctrl Shift O
|Cmdl Shift O
|Focus chat input
|Copy last code block
|Ctrl Shift ; (semicolon)
|Cmd Shift ; (semicolon)
|Copy last response
|Ctrl Shift C
|Cmd Shift C
|Set custom instructions
|Ctrl Shift I
|Cmd Shift I
|Ctrl Shift S
|Cmd Shift S
|Ctrl Shift Backspace
|Cmd Shift Backspace
|Ctrl / (forward slash)
|Cmd / (forward slash)