ChatGPT: latest news, updates and useful tips

Updated 5 months ago on June 07, 2024

In the ever-evolving landscape of artificial intelligence, ChatGPT stands out as a groundbreaking development that has captured the world's attention. From its impressive capabilities and recent advances to the heated debate surrounding its ethical implications, ChatGPT continues to hit the headlines.

  • What is ChatGPT?
  • When was ChatGPT released?
  • How to use ChatGPT
  • How to use the ChatGPT apps for iPhone, Android and Mac
  • Is ChatGPT free to use?
  • Who created ChatGPT?
  • ChatGPT's ongoing controversy
  • Can the results of ChatGPT be detected by anti-plagiarism systems?
  • What are ChatGPT plugins?
  • Is there a ChatGPT API?
  • What is the future of ChatGPT?
  • Alternatives to ChatGPT that you should try
  • Other information about ChatGPT

If you're a tech enthusiast or just curious about the future of artificial intelligence, dive into this comprehensive guide to learn everything you need to know about this revolutionary AI tool.

What is ChatGPT?

ChatGPT is a natural language artificial intelligence chatbot. At the most basic level, this means that you can ask it a question and it will generate an answer. Unlike a simple voice assistant like Siri or Google Assistant, ChatGPT is built on what's called an LLM (Large Language Model). These neural networks are trained on huge amounts of information from the Internet, which allows them to generate entirely new responses, rather than just repeating specific ready-made answers. They are not built for a specific purpose like the chatbots of the past, and they are much smarter.

This is implied in the name ChatGPT, which stands for Chat Generative Pre-trained Transformer. In the case of the current version of ChatGPT, it is based on the GPT-4 LLM. The model behind ChatGPT has been trained on all sorts of web content, including websites, books, social media, news articles and more - all fine-tuned into a language model using both supervised learning and RLHF (Reinforcement Learning From Human Feedback). OpenAI claims that the use of human AI learners is what sets ChatGPT apart.

When was ChatGPT released?

ChatGPT was originally introduced to the public in November 2022 by OpenAI. The initial version was based on the GPT-3.5 model, but the system has undergone a number of iterative improvements since then. The current version of ChatGPT uses the GPT-4 family of models, and GPT-5 is reportedly just around the corner. GPT-3 was first launched in 2020, GPT-2 came out the year before that.

How to use ChatGPT

First, go to chatgpt.com . If you want to keep a history of your previous chats, register a free account. If you wish, you can use the system anonymously, without a login. Users can link their ChatGPT login to their Google, Microsoft or Apple account login. On the signup screen, you'll see some basic rules about ChatGPT, including possible data errors, how OpenAI collects data, and how users can leave feedback. If you're eager to get started, we've put together a collection of top tips on ChatGPT.

OpenAI ChatGPT rules and limitations.

Using ChatGPT is very simple and straightforward: just type a text and wait for the system to respond. You can be as creative as you want and see how your ChatGPT responds to different prompts. If you don't get the result you want, try changing the prompt or giving ChatGPT additional instructions. The system understands context based on previous responses in the current chat session, so you can refine your prompts rather than starting over each time.

For example, if you start with "Explain how the solar system was created", the result will be more detailed, with more paragraphs, than with "How the solar system was created", although both queries will give fairly detailed results. Take it a step further by giving ChatGPT more guidance regarding style or tone, such as "Explain how the solar system was created as a high school teacher."

You can also ask for more specific queries, such as an essay with a given number of paragraphs or a link to a specific Wikipedia page. We got an extremely detailed result for the query "write a four paragraph essay explaining Mary Shelley's Frankenstein".

ChatGPT can automate any number of everyday work or personal tasks, from writing emails and composing business proposals to suggesting fun date ideas or even composing the best man's speech at your buddy's wedding. As long as the request does not violate the site's rules regarding explicit or illegal content, the model will do its best to fulfill it.

Since its launch, people have been experimenting to find out what this chatbot can and can't do, and the results are impressive to say the least. To find out which prompts and follow-ups ChatGPT responds best to, you need to experiment a bit. Just as we've learned to get the information we need from traditional search engines, ChatGPT may take some time to get the best results. It all depends on what you want to get out of it. For starters, try using it to write template blog posts, for example, or even blocks of code if you're a programmer.

Three different screens from the official iOS app.

Our authors also experimented with ChatGPT, trying to see if it could handle holiday shopping or even correctly interpret astrological makeup. In both cases, we found limitations in capabilities, but were very impressed with the results.

Following the August 10 update, you can now use custom instructions in ChatGPT. This will allow you to customize the AI chatbot's response to your actions so you can tailor it to your needs. However, you can't ask anything: OpenAI has taken precautions to "create a safe and useful general-purpose artificial intelligence." This means that any questions containing hate, sexism, racism or discrimination will generally remain off-limits.

Don't take at face value everything ChatGPT (or any other chatbot, for that matter) tells you. When ChatGPT first launched, it was very prone to "hallucinations". The system would repeat erroneous data as fact. This problem has become less common as the model is constantly being improved, but errors still happen. Trust but verify!

Furthermore, because of the way OpenAI trains its underlying big language models - whether it's GPT-3.5, GPT-4 and GPT-4o, or the upcoming GPT-5 - ChatGPT won't be able to answer your question without the help of an internet search if it's about something that happened recently. For example, GPT-3.5 and 3.5 Turbo can't answer questions about events after September 2021 without an internet search because the data the model was originally trained on was acquired before that "knowledge cutoff date." Similarly, GPT-4 and GPT-4 Turbo have a December 2023 cutoff date, and GPT-40 (despite being released more recently) has an October 2023 cutoff date.

While ChatGPT may not remember your entire history, it will remember what you have discussed with it in previous chats. Logged-in users can access chat history from the navigation sidebar on the left side of the screen and manage them by renaming, hiding, or deleting them as needed. You can also ask ChatGPT questions based on previous conversations right in the chat window. Users also have the option to use ChatGPT in dark or light mode.

ChatGPT is more than just a word wizard. Users who pay for a $20/month subscription to ChatGPT Plus or a $30/month subscription to ChatGPT Business get access to the Dall-E image generator, which converts text prompts into realistic generated images. Unfortunately, this feature is not currently available for the free tier. Regardless of subscription status, all users can use images or voice for the prompts.

How to use ChatGPT apps for iPhone, Android and Mac

ChatGPT app running on an iPhone.

ChatGPT is available through the OpenAI website and as a mobile app for iOS and devices. The iOS version was an instant hit as soon as it appeared on the App Store, with half a million downloads in less than a week.

If you can use chatGPT on the web, you can use it on your phone too. Logging in or registering through the app is almost identical to the web version, and almost all of the features available on desktop have been ported to the mobile versions. The app also allows you to switch between GPT-3.5, GPT-4, and GPT-4o. The clean interface displays your GPT conversation in a simple manner, hiding your chat history and settings behind a menu in the top right corner.

The ChatGPT iPhone app answering the question "how long are canned tomatoes good for."

Some devices go beyond just the app. For example, Infinix Folax is an Android phone that integrates ChatGPT into the entire device. Instead of an app, the phone replaces the typical smart assistant (Google Assistant) with ChatGPT.

There's even an official ChatGPT app for the Mac that you can use for free. The app is capable of all sorts of things that take the Mac's artificial intelligence capabilities to the next level - and you don't even have to wait for macOS Sequoia later this year.

Is ChatGPT free to use?

A laptop opened to the ChatGPT website.

Yes, ChatGPT is completely free to use, albeit with some limitations. Even with a free account, users have access to the GPT-3.5 and GPT-40 models, although the number of requests for the more advanced model is limited. Upgrading to a paid subscription significantly increases this query limit, provides access to other generative AI tools, such as Dall-E image generation, and the GPT store.

Of course, it is not free for OpenAI to continue using it. Initial estimates indicate that OpenAI spends about $3 million per month to keep ChatGPT running, which is about $100,000 per day. The April 2023 report indicates that the cost of operation is closer to $700,000 per day.

In addition to the cost of the servers themselves, some disturbing information and allegations have come to light regarding what else has been done to secure the model from producing offensive content.

Who created ChatGPT?

OpenAI, a San Francisco-based artificial intelligence research lab, created ChatGPT and released the first version of the LLM in 2018. The organization began as a nonprofit designed to collaborate with other institutions and researchers, and was funded by big names such as Peter Thiel and Elon Musk, the latter of whom left the company after an internal power struggle to found rival firm xAI.

OpenAI went commercial later in 2019 and is now led by CEO Sam Altman. It runs on Microsoft's Azure system infrastructure and is powered by Nvidia GPUs, including the new supercomputers announced this year. Microsoft has also been investing heavily in OpenAI since 2019, expanding its partnership with the AI startup in 2021 and then in 2023, when Microsoft announced a multi-billion-dollar round of investment, including naming its Azure cloud as OpenAI's exclusive cloud provider.

ChatGPT's ongoing controversy

While ChatGPT is an extremely capable digital tool, it cannot be called reliable. AI is notorious for making mistakes or "hallucinating" when it comes up with an answer to a question that is unknown to it. At the very beginning of the paper, a simple example of how unreliable it can be is the misidentification of the Prime Minister of Japan.

Many people are concerned not only about the bugs, but also about how this human-like generative AI could affect the future of the internet. Thousands of technology leaders and prominent public figures have signed a petition demanding that its development be slowed down. It was even banned in Italy due to privacy concerns, along with complaints from the FTC - though that decision has now been reversed. The FTC has since reopened an investigation against OpenAI over its handling of consumers' personal data.

In addition, JPMorgan Chase has threatened to restrict the use of artificial intelligence chatbots for employees, especially for creating emails, something that companies like Apple have also banned. After Apple announced at WWDC 2024 that it would integrate OpenAI technology into its mobile and desktop products, Tesla CEO and loser Elon Musk similarly threatened to ban the use of any device with the software in his business, from iPhones to Mac Studios. Other high-profile companies, including Samsung, Amazon, Verizon, and even the US Congress, have banned the use of ChatGPT internally.

In addition, there are concerns that generative AI such as ChatGPT could lead to many job losses - up to 300 million worldwide, according to Goldman Sachs. In particular, this issue has come into focus during the screenwriters' strike in Hollywood, who want to ensure that scripts written by artificial intelligence do not take jobs away from screenwriters.

In 2023, many people trying to use ChatGPT received a "at capacity" notification when trying to access the site. This is probably behind an attempt to use unofficial paid apps that have already flooded app stores and tricked thousands of people into paying for a free service.

Because of how much it costs to run ChatGPT, it looks like OpenAI limits access when the servers are "at capacity". The wait can drag on for hours, but if you are patient, it will work out eventually. Of the many problems ChatGPT faced, the "at capacity" bugs were the biggest obstacle keeping people from using the service anymore. In some cases, demand was so high that the entire ChatGPT site was shut down for a few hours for maintenance several times over the course of several months.

Numerous controversies have also arisen due to people using ChatGPT to perform tasks that should probably be handled by a real person. One of the worst cases is creating malware, something the FBI recently warned about ChatGPT being used for. More surprisingly, Vanderbilt University's Peabody School has come under fire for creating an email about mass shootings and the importance of community.

There are privacy concerns as well. A recent GDPR complaint says ChatGPT violates users' privacy by stealing their data without their knowledge and using it to train an artificial intelligence model. According to one user, ChatGPT was even allowed to generate Windows 11 keys for free. Of course, this is not how ChatGPT was supposed to be used, but it is significant that it was able to be "tricked" into generating keys at all.

Can the results of ChatGPT be detected by anti-plagiarism systems?

Teachers, school administrators, and developers are already finding various ways to get around this problem and ban the use of ChatGPT in schools. Others are more optimistic about how ChatGPT can be used for learning, but plagiarism will undoubtedly continue to be a problem in education in the future. There are some ideas about how ChatGPT could "watermark" its text and solve this plagiarism problem, but for now, detecting ChatGPT is incredibly difficult.

ChatGPT launched an updated version of its own plagiarism detection tool in January 2023, hoping it will defuse criticism about the way people use the text generation system. It uses a feature called "AI text classifier," which works on a principle familiar to other plagiarism-detection software. However, the tool is still under development and remains "imperfect," according to OpenAI. Since GPT was introduced in April 2024, third-party developers have also begun releasing their own offerings, such as Plagiarism Checker.

What are ChatGPT plugins?

It's a feature that no longer exists. The announcement of ChatGPT plugins caused quite a stir in the developer community, with some calling it "the most powerful developer platform ever created." Artificial intelligence enthusiasts have compared it to the surge of interest in the iOS app store when it first appeared, greatly expanding the capabilities of the iPhone.

Essentially, developers will be able to create plug-ins directly for ChatGPT to open it up to the entire Internet and connect directly to APIs of specific applications. Among the examples given by OpenAI are applications capable of performing actions on behalf of a user, real-time information, and access to knowledge-based information.

However, in 2024, OpenAI changed its plans for plugins, abandoning them and replacing them with GPT applets. OpenAI's GPT applets were released along with the GPT-4o submission, they are small interactive JavaScript applications created by GPT-4 and available on the ChatGPT website. These applets are a variety of tools designed to perform specific, often one-off tasks, such as acting as calculators, schedulers, widgets, image applications, and text conversion utilities.

Is there a ChatGPT API?

ChatGPT versus Google on smartphones.

Yes. The API is a way for developers to access ChatGPT and plug its natural language capabilities directly into applications and websites. We've seen it used in a wide variety of use cases, from suggesting parts in a Newegg PC builder to plotting a travel itinerary with just a few words. Many applications have been announced by OpenAI partners that utilize the ChatGPT API. Of the first batch, Snapchat's MyAI is the most prominent example.

OpenAI recently made the ChatGPT API available to everyone, and we've seen a surge in the emergence of tools that utilize the technology, such as Discord's chatbot Clyde or Wix's website builder.More recently, GPT-4 was made available as an API "for developers to build apps and services." Companies that have already integrated GPT-4 include Duolingo, Be My Eyes, Stripe and Khan Academy.

What is the future of ChatGPT?

There's no doubt that the tech world is obsessed with ChatGPT right now, and that process isn't slowing down anytime soon. But the more important development will be how ChatGPT will be integrated into other applications.

GPT-5 is the next major step forward that has been rumored and talked about for the past year. Some claim it will graduate as early as December 2024, paving the way for AGI (general-purpose artificial intelligence). OpenAI CTO Mira Murati has compared it to PhD-level intelligence, while others say it will lead to AI with better memory and reasoning. The timeline, however, seems very uncertain, but it looks like it could launch in 2025.

Outside of GPT-5, many AI enthusiasts and prognosticators are predicting where the technology is headed. Last year, Shane Legg, co-founder of Google DeepMind and chief scientist on AGI, told Time Magazine that he believes there is a 50% chance that AGI will be developed by 2028. Dario Amodei, co-founder and CEO of Anthropic, is even more optimistic, saying last August that "human-level" AI could emerge in the next two to three years. For his part, OpenAI CEO Sam Altman claims that AGI could be achieved within the next half decade.

If you think AI is a big problem, then obviously we are still at the very beginning of the journey.

Alternatives to ChatGPT that you should try

Bing Chat open on a smartphone showing visual results.

ChatGPT remains the most popular artificial intelligence chatbot, but it is not without competition. Microsoft's Copilot is a serious contender, even though Microsoft has invested heavily in this AI startup and Copilot itself uses the GPT-4 model for its responses.

Google's Gemini AI (formerly Google Bard) is another such competitor. This family of multimodal AI models, based on Google's proprietary Transformer architecture, is capable of understanding and generating text, images, audio, video and code. First released in March 2o23, Gemini is available in 46 languages in 239 countries and territories. One of its main advantages is that Gemini can generate images for free, whereas in the OpenAI ecosystem you have to upgrade to ChatGPT Plus.

Anthropic's Claude family of AIs has also emerged as a serious challenger to ChatGPT's dominance. In June 2024, Anthropic announced that its recently released Claude 3.5 Sonnet outperformed GPT-4o and Gemini Pro 1.5 in a range of industry benchmarks and significantly outperformed the older Claude 3.0 Opus by double digits while consuming 50% less power.

Meta, Facebook's parent company, has also spent the last few years developing its own artificial intelligence chatbot based on the Llama family of large language models. In April 2024, the company finally unveiled its chatbot, dubbed Meta AI, and said it uses the latest Llama 3 model to date. The assistant is available in more than a dozen countries and works across Meta's entire suite of apps, including Facebook, Instagram, WhatsApp and Messenger.

Finally, it's long been rumored that Apple is working on its own artificial intelligence system, and at June's WWDC 2024 conference, the company proved its case by unveiling Apple Intelligence. The AI "consists of high-performance big language and diffusion models specialized for your everyday tasks" and is designed to help iPhone, iPad, and Mac users optimize many of their most common everyday tasks across all apps.

For example, the system will self-prioritize certain system notifications to minimize distractions while you focus on a task, and writing assistants will be able to proofread your work, revise it on your command, and even outline the text for you. Apple's AI is expected to start rolling out to users along with iOS 18, iPadOS 18, and Mac Sierra releases in the fall of 2024.

Other information about ChatGPT

Are ChatGPT chats private?

It depends on what you mean by private. All ChatGPT chats are used by OpenAI for further model customization, for which human instructors may be involved. No, this does not mean that a human is looking at every question you ask ChatGPT, but OpenAI cautions against giving ChatGPT any personal information for a reason.

It should be noted that unless you delete your chats, the conversations will be displayed in the left sidebar. Unlike other chatbots, individual chats within a conversation cannot be deleted, although you can edit them using the pencil icon that appears when you hover over the chat. However, when you delete conversations, ChatGPT doesn't forget about them - they simply disappear from the conversation history in the sidebar.

Fortunately, OpenAI recently announced the ability to make chats hidden from the sidebar. These "hidden" chats will also not be used to train AI models. You can also opt out of allowing OpenAI to train your models in the settings.

Instead of replacing it, generative AI features are being integrated directly into search. Microsoft started by integrating Copilot directly into its search engine, which put a "Chat" tab right in the Bing search menu. Google, of course, made its big move with AI Overviews, which uses AI-generated answers instead of traditional search results. It was first launched as part of the Search Generative Experience, and began rolling out widely in May 2024.

To be clear, this kind of AI is different from Gemini or ChatGPT. Still, there is no denying that AI will play an important role in the future of search in the near future. Despite all the problems associated with AI reviews, Google seems to be determined to bring it to fruition.

Is Copilot the same as ChatGPT?

Bing Chat shown on a laptop.

While Copilot and ChatGPT are capable of similar things, they are not exactly the same. Copilot, while running on the same GPT-4 model as ChatGPT, is a completely separate product that has been finalized by Microsoft.

Microsoft, as part of its multi-billion dollar investment in OpenAI, initially brought ChatGPT to Bing in the form of Bing Chat. But unlike ChatGPT, Bing Chat required downloading the latest version of Edge at the time.

Since then, Bing Chat has been completely repurposed into Copilot, which seems to have become Microsoft's most important product. It is integrated into Microsoft 365 apps through Copilot Pro, and Copilot+ extends artificial intelligence capabilities into Windows and laptop hardware.

Can you write essays using ChatGPT?

The use of ChatGPT is controversial, with many believing that the power of artificial intelligence will change everything from search engines to novel writing. It has even demonstrated the ability to bring students surprisingly good essay grades.

Writing essays for students is one of the most obvious examples of where ChatGPT can become a problem. ChatGPT may not write this article that well, but it is particularly easy to use for essay writing. Some generative AI tools, such as Caktus AI, are built specifically for this purpose.

Can ChatGPT write and debug code?

Absolutely. This is one of the most powerful features of ChatGPT. As with anything involving artificial intelligence, you'll have to double-check everything it outputs because it won't always understand your code correctly. But it is certainly powerful in both writing code from scratch and debugging it. Developers use it to build websites, apps, and games from scratch - and all of this, of course, becomes even more powerful with GPT-4.

What is the character limit in ChatGPT?

There is no hard limit on the number of characters in ChatGPT. However, the size of the context box (essentially, the length of the prompt) depends on the ChatGPT level you are using. Free level users get only 8,000 characters, Plus and Teams subscribers get context boxes of 32,000 characters, and Enterprise users get as much as 128,000 characters.

What is Auto-GPT?

Built on GPT-4, Auto-GPT is the latest evolution of artificial intelligence technology that has caused a stir in the industry. It's not directly related to ChatGPT or OpenAI - it's an open source Python application that got into the hands of developers all over the internet when it was published on GitHub.

In ChatGPT or ChatGPT Plus, the AI's capabilities are limited to a single chat window. Auto-GPT, at its simplest, makes the AI autonomous. It is tasked with a specific task and then takes the necessary steps to solve it via the Internet, including connecting to applications and software.

According to the official description on GitHub, Auto-GPT is "an experimental open source application that demonstrates the capabilities of the GPT-4 language model. This GPT-4-driven program combines the "thoughts" of the LLM to autonomously achieve any goal you set. As one of the first examples of fully autonomous GPT-4 operation, Auto-GPT pushes the boundaries of what is possible with AI."

The demo provided on the GitHub page is simple - just create a recipe suitable for Easter and save it to a file. What's interesting is how Auto-GPT breaks down the steps taken by the AI to achieve a goal, including the "thoughts" and "reasoning" behind its actions. Auto-GPT is already being used in various applications, and some are calling it the beginning of AGI (Artificial General Intelligence) because of its autonomous nature.

This issue is open to debate. There is a lot of talk around copyright and AI: some argue that generative AI is "stealing" the work of those it has been trained on. This is becoming increasingly controversial in the AI art world. Companies like Adobe are finding ways around this issue by training models only on stock image libraries that already have proper copyrights and legal boundaries.

However, according to OpenAI, you have the right to reprint, sell, and advertise anything created with ChatGPT or ChatGPT Plus. Therefore, OpenAI is not going to sue you.

The broader topic of copyright with respect to generative AI has yet to be resolved by various legislators and interpreters of the law, especially since copyright law as it stands technically only protects human-generated content.

Let's get in touch!

Please feel free to send us a message through the contact form.

Drop us a line at mailrequest@nosota.com / Give us a call over skypenosota.skype