What is ChatGPT?

We have already talked about virtual reality and augmented reality, then we looked at the metaverse, now it’s time to address another topic that is all the rage: artificial intelligence and, more specifically, ChatGPT.

In recent months, the Internet has been receiving a wave of AI-generated text, art, and audio. The latest is the creation of audios with artists singing songs they have never sung. Can you imagine hearing whatever you want in the voice of your favorite artist? You don’t have to imagine anymore, because that is exactly what they are doing. If you’ve never seen it, here’s Ariana Grande singing Happier Than Ever by Billie Eilish.

ChatGPT is in the text field and, since its launch last year, it has become increasingly well-known. The tool is already revolutionizing the way we interact with AI and is raising pertinent questions about how these technologies can impact our lives. So, come understand more about it!

What is it and how to use it?

ChatGPT is a natural language model trained by OpenAI that allows you to interact with people through text conversations. It is based on GPT-3, one of the most advanced language models in the world. It has been trained on large amounts of Internet text to generate human responses to questions and conversations. It is able to answer questions on a wide range of topics, including history, geography, science, and others.

Not only does it answers questions, but ChatGPT can also generate texts, translations, summaries, poems, articles, music, and basically any textual form you request. It can even do mathematical calculations and code!

To access, simply go to chat.openai.com, create an account, and start chatting – you can ask any question you like and you will receive an answer within seconds. You can currently use it for free, but there is also a paid subscription option, ChatGPT Plus, for $20 per month. It gives you the ability to access at peak times without having to deal with “at capacity” messages, plus benefits like faster responses and new features.

Other than ChatGPT, these are some options for text-generating chats that use AI: YouChat, Jasper, and Chatsonic.

Differences to other chatbots and to search engines

You may be wondering if ChatGPT is the same as other chatbots, such as those for virtual assistance, and the answer is no. These chats don’t use artificial intelligence, they just take keywords to generate pre-programmed answers, while ChatGPT can establish conversations with the user, adapting to new information and contexts.

Search engines, on the other hand, search the Internet for results. ChatGPT, as we have just mentioned, is designed for conversation, and it does not have the ability to find new data on the Internet, relying only on what has already been learned.

Other AI models

DALL-E 2

Also created by OpenAI, DALL-E 2 is an AI system that can create realistic images and art from a text description.

Whisper

Another creation of OpenAI, Whisper is “an automatic speech recognition (ASR) system trained on 680,000 hours of multilingual and multitasking data collected from the web.”

Midjourney

Like DALL-E 2, Midjourney creates images from text descriptions.

Stability AI

Stability AI’s goal is to make open source tools and models available, together with AWS, to researchers, students, startups and companies.

Problems and limitations

The first problem is that the answers generated are not 100% reliable, and this is made clear by the bot itself. The tool is error-prone and can pass on data that sounds true even when it is not, so all information needs to be checked. To tell you the truth, this goes for anything you see on the internet – always check the facts!

Stack Overflow, a question and answer site in the programming field, has temporarily banned ChatGPT answers for this very reason. Moderates said in a post, “The primary problem is that while the answers which ChatGPT produces have a high rate of being incorrect, they typically look like they might be good and the answers are very easy to produce”.

Moreover, the chat finds it difficult to answer questions depending on how they are structured, and can give meaningless answers. One big limitation is that the data was collected until 2021, so it doesn’t know any information after that year. For example, if you ask who is the current president of Brazil, it will not know how to answer correctly.

What does it mean for the future - will AI replace humans after all?

Perhaps the biggest question that the rise of AI brings up is how society will use these tools and whether they will replace human labor. There is no denying that they are here to stay, so we have to try to understand what the consequences will be for our reality.

On the one hand, ChatGPT can help with daily tasks and optimize time spent on them. However, one cannot ignore the problem of spreading wrong information, in addition to the cybersecurity risks.

Google already plans to launch its own chatbot, called Bard, as well as more than 20 AI products, according to The New York Times. The biggest concern, however, is for the brand’s reputation in the face of these past data error and security issues.

The education system also finds a valid concern in the chat: plagiarism. With how easy it is to generate texts, you can do an entire job with ChatGPT. Noam Chomsky, American linguist, philosopher, activist, author, and political analyst, has criticized the chat, referring to it as “high-tech plagiarism” and “a way of avoiding learning“.

The profession that is most affected by the chat is that of copywriters and textual content creators. If a text can be produced in a matter of seconds based on simple prompts, companies may end up choosing to use the tool instead of professionals in the field.

Graphic design also has a possible threat in AI – not with ChatGPT, but with other models that generate art, such as DALL-E 2.

This text, for example, was written by an actual person, but you can already see that it didn’t have to. In fact, some of it was made by the ChatGPT – can you figure it out? If you want to test your ability to identify AI-generated text, you can go back and reread, but if you are just curious, I’ll tell you: it was the one who described itself in the first paragraph of the topic “what it is and how to use it”.

But can AI really replace people? We believe not, at least for the time being. Only someone real can do a fact-check, for example. Creativity and empathy cannot yet be reproduced by a machine; the nuances of human conduct are perceived and understood only by another human being.

There is still a debate about the ethics of artificial intelligences. In a copyright case, GettyImages sued Stability AI for illegal use of its copyrighted images in training the Stable Diffusion algorithm, another model for generating images by text description.

So the idea is to use technology to our advantage, as an assistant in the execution of our tasks. Thinking about how to use artificial intelligence in an ethical and moral way is essential for it to be an ally and not a problem in the future.

Want to know more about ChatGPT? Listen to the episode of the podcast O Assunto that talks about it!