ChatGPT: How to use it?
ChatGPT is revolutionary technology because it is trained to learn what humans mean when they ask a question.
The tool is an Artificial Intelligence (AI) that can be very useful for your day-to-day. Introduced by the company OpenAI, an artificial intelligence company that Elon Musk invested in, ChatGPT can answer, with high precision, complex questions through conversation.
Many users are impressed by its ability to provide human-quality answers, inspiring a feeling that it may eventually have the power to change the way humans interact with computers and change the way searches for information are done.
What is ChatGPT?
ChatGPT is a large chatbot (an artificial intelligence for conversations) developed by OpenAI based on GPT-3.5 (an autoregressive language model that uses deep learning to produce human-like text). It has a remarkable ability to interact in conversational dialogue fashion and provide responses that can seem surprisingly human.
An additional layer of training that uses human feedback to help ChatGPT learn the ability to follow instructions and generate satisfying responses for humans is Reinforcement Learning with Human Feedback (RLHF).
Who created ChatGPT?
ChatGPT was created by the San Francisco-based artificial intelligence company OpenAI. OpenAI is famous for its well-known DALL·E, a deep learning model that generates images from text instructions called prompts.
Microsoft is a partner and investor in Open AI, worth $1 billion. They jointly developed the Azure AI platform.
Large Language Models
ChatGPT is a Large Language Model (or, as it is more commonly known, LLM). LLMs are trained on massive amounts of data to accurately predict which word comes next in a sentence.
Increasing the amount of data has been found to increase the capacity of language models. According to Stanford University:
“GPT-3 has 175 billion parameters and was trained on 570 gigabytes of text. For comparison, its predecessor, GPT-2, was 100 times smaller, with 1.5 billion parameters.
This increase in scale drastically changes the behavior of the model — GPT-3 is able to perform tasks it has not been explicitly trained to do, such as translating sentences from English to French, with little or no training examples.
This behavior was mostly absent in GPT-2. Furthermore, for some tasks, GPT-3 outperforms models that have been explicitly trained to solve those tasks, while for other tasks it falls short.”
LLMs predict the next word in a string of words in a sentence and the next sentences – like autocomplete but on an absolutely larger scale. This skill allows them to write paragraphs and entire pages of content, but LLMs are limited because they don’t always understand exactly what a human being wants.
And that’s where ChatGPT comes into play, with the aforementioned Reinforcement Learning with Human Feedback (RLHF) training.
How was ChatGPT trained?
GPT-3.5 was trained on large amounts of code data and information from the Internet, including sources such as Reddit discussions, to help ChatGPT learn dialogue and achieve a human style of response.
ChatGPT was also trained using human feedback (a technique called Reinforcement Learning with Human Feedback) so that the AI learned what humans expected when they asked a question. Training the LLM in this way is revolutionary because it goes beyond simply training the LLM to predict the next word.
A March 2022 research article titled Training Language Models to Follow Instructions with Human Feedback explains why this is an innovative approach:
“This work is motivated by our goal to increase the positive impact of large language models by training them to do what a given set of humans want them to do.
By default, language models optimize for the next-word prediction objective, which is just a kickoff for what we want these models to do.
Our results indicate that our techniques promise to make language models more useful, truthful and harmless.
Making language models bigger doesn’t inherently make them better at following user intent.
For example, large language models can generate output that is false, toxic, or simply not useful to the user.
In other words, these models are not aligned with their users.”
Key Differences From Other Chatbots
What sets ChatGPT apart from a simple chatbot is that it has been specifically trained to understand the human intent on a question and provide helpful, truthful, harmless answers.
Because of this training, ChatGPT can dispute certain questions and discard parts of the question that don’t make sense.
Another research work related to ChatGPT shows how they trained the AI to predict what humans prefer.
The researchers noted that the metrics used to evaluate the natural language processing AI outputs resulted in machines that scored well statistically, but did not line up with what humans expected.
Researchers’ Explanation
The researchers explained the problem as follows:
“Many machine learning applications optimize for simple metrics that are just rough proxies for what the designer intended. This can lead to issues such as YouTube recommendations promoting click-bait.”
So the solution they came up with was to create an AI that could generate responses that were optimized for what humans preferred. To do this, they trained the AI using datasets of human comparisons between different responses, so that the machine would become better at predicting what humans judged to be satisfactory responses.
What are the limitations of ChatGPT?
Limitations on Toxic Responses
ChatGPT is specifically programmed not to provide toxic or harmful responses. Therefore, it will avoid answering these types of questions.
Quality of Answers Depends on Quality of Questions
An important limitation of ChatGPT is that the quality of the output depends on the quality of the input. In other words, expert guidance (prompts) generates better responses.
Answers Not Always Correct
Another limitation is that the programmers have trained ChatGPT to provide answers that look right to humans, so the answers can mislead humans that the information is correct.
Many users have found that ChatGPT can give partially incorrect answers, including some totally incorrect ones. For example, one user reported getting a response that Elephants are the mammals that lay the largest eggs. However, it is possible that this is due to the inability of the AI to differentiate hypothetical scenarios from real scenarios.
OpenAI Explains the Limitations of ChatGPT
OpenAI’s announcement offered this caveat:
“ChatGPT sometimes writes answers that seem plausible but are incorrect or meaningless.
Fixing this problem is challenging because:
(1) during RL training, there is currently no source of truth;
(2) training the model to be more cautious causes it to decline questions it can answer correctly; and
(3) supervised training misleads the model because the optimal response depends on what the model knows, not what the human demonstrator knows.”
Is ChatGPT Free?
Currently, the use of ChatGPT is free but there are already rumors that the paid version of ChatGTP may come out soon and will cost 42 dollars per month.
Currently, the chatbot is open for users to try it out and provide feedback on answers so that the AI can better answer questions and learn from its mistakes.
The official announcement states that OpenAI is looking forward to receiving feedback on the bugs:
“While we’ve made efforts to get the model to reject inappropriate requests, it sometimes responds to harmful instructions or exhibits biased behavior.
We’re using the Moderation API to warn or block certain types of unsafe content, but expect there to be some false negatives and positives for now.
We look forward to collecting user feedback to help in our continued work to improve this system.”
How to Use ChatGPT?
1. Access the ChatGPT website in your browser
However, it is possible that the service is down due to the number of users, but do not be discouraged! Normally, within a few minutes, the service is back up and running.
2. Sign up for the service
Currently, the service is free, but you need to login with a username and password. If you prefer, you can sync the chat with your Google or Microsoft account.
3. Start Chatting!
Write in the chat field (marked in red) your command or question – it can be in a lange number of languages.
4. Wait for ChatGPT’s response
In the left menu, you can create a new conversation (“New Chat”) or select a previous chat. In that way, you don’t mix up your subjects.
Will Language Models Replace Google Searches?
Google itself has already created an AI chatbot called LaMDA. The Google chatbot’s performance was so close to a human conversation that a Google engineer claimed that LaMDA was sentient.
Given how these large language models can answer so many questions, is it unlikely that a company like OpenAI, Google or Microsoft will one day replace traditional search with an AI chatbot?
Some Twitter users are already declaring that ChatGPT will be the next Google.
The scenario that a Q&A chatbot could one day replace Google is a daunting one for those who make their living as search marketers. This sparked discussions in online search marketing communities such as the popular Facebook SEOSignals Lab, where someone asked whether searches could shift from search engines to chatbots.
Having tested ChatGPT myself, I have to agree that the fear that search will be replaced by a chatbot is not unfounded.
The technology still has its limitations, but we can already envision a hybrid future where search engines team up with chatbots to search for information.
But the current ChatGPT implementation appears to be a tool that will, at some point, require purchasing credits to use.
How Can I Use ChatGPT?
ChatGPT can write code, poems, songs and even short stories in the style of a specific author.
Experience in following instructions elevates ChatGPT from a source of information to a tool you can call upon to accomplish a task.
This makes it useful for writing articles on just about any topic. ChatGPT can work as a tool to generate article outlines or even entire novels.
It will provide an answer to just about any task a written text can answer.
What does ChatGPT consider What is ChatGPT?
To join in the fun a bit, I asked ChatGPT to write an article about itself. That was the answer.
“ChatGPT is a language model developed by OpenAI. It has been trained using a large amount of text data available on the internet, which makes it capable of generating naturalistic and intelligent responses to a variety of natural language questions and tasks.
Thus, one of the key features of ChatGPT is its ability to understand and respond to contextual questions. This means that he can understand the meaning of a question and provide an accurate answer, even if the question is not worded exactly. In addition, it is also able to generate text with fluidity and cohesion, which makes it useful for tasks such as text generation, automatic summarization and automatic translation.
Another advantage of ChatGPT is its scalability. As it has been trained on a large amount of data, it can be easily adapted for a variety of specific tasks and domains. That means it can be used to chat with customers, help write reports, answer technical questions, and much more.
Furthermore, ChatGPT is also easy to use. It can be accessed through an API, which means that developers can easily integrate it into applications and systems. This has allowed it to be used in a variety of scenarios such as virtual assistants, communication platforms and education applications.
In summary, ChatGPT is a powerful and versatile language model developed by OpenAI. It is able to understand and answer contextual questions, generate text with fluidity and cohesiveness, and be easily adapted for a variety of specific tasks and domains. It’s also easy to use, allowing it to be used in a variety of scenarios.”
Conclusion
As mentioned earlier, the developers created ChatGPT as a tool that the public will eventually have to pay to use. More than one million users have registered to use ChatGPT in the first five days since it opened to the public.
Did you like our content?
Stay tuned on Apptuts and share this article with your friends!
