GPT-5: What you need to know about OpenAI's next generation of chatbots

According to two sources who spoke to Business Insider, the new AI model called GPT-5, is expected to launch this summer. Ahead of the launch, several businesses are said to have tried out a demo of the tool, allowing them to test its upgrade capabilities.

The technology is part of OpenAI's future mission of artificial general intelligence (AGI), or systems that are smarter than humans.

This science fiction foundation predicts an AI that can think for itself, thereby creating many AI models like it without the need for human supervision. Depending on who you ask, such a breakthrough development could either destroy or accelerate human progress.

As the race to build the most powerful AI systems becomes increasingly fierce, here's everything you need to know about GPT-5.

What is GPT-5?

GPT-5 is the next version of  GPT-4,  OpenAI's fourth-generation  chatbot  that you pay a monthly fee to use.

GPT stands for Generative Pre-trained Transformer, which is an AI engine built and improved by OpenAI to power different versions of ChatGPT. Like the processor inside your computer, each new version of a chatbot runs on a completely new GPT with more capabilities.

OpenAI's ChatGPT app allows users to ask questions by voice (OpenAI)

In the case of GPT-4, the AI ​​chatbot can give human-like responses, even recognizing and generating images and voices. Its successor, GPT-5, is said to offer better personalization, fewer errors, and handle a wider range of content, including video.

Other companies like Google and Meta have released their own GPT under their own names, all of which are collectively known as the major language model.  Based on the human brain, these AI systems are capable of generating text as part of a conversation.

GPT-5 is being trained?

The latest report states that OpenAI has started training GPT-5 as it prepares to release the AI ​​model in the middle of this year. According to Business Insider, after the training process is complete, the system will go through many stages of safety testing.

As part of this process, the bot will be “ red teamed ,” a technique that includes internal and external testers putting it through its paces and providing feedback on its strengths and weaknesses.

The report speculates that GPT-5's training may have begun recently, based on a recent tweet from an OpenAI official.

In January, one of the tech company's top researchers hinted that OpenAI was training a much larger GPU than usual. This revelation follows a separate tweet from OpenAI's co-founder and president, detailing how the company has expanded its computing resources.

GPU, short for graphics processing unit, is like a computer that helps an AI model find connections between different types of data, such as linking an image to a corresponding text description.

https://twitter.com/gdb/status/1750558864469299622

For his part, Sam Altman confirmed that his company was working on GPT-5 on at least two separate occasions last fall.

According to two people who attended the event, the first time was during a talk at the alumni reunion of his former venture capital firm Y Combinator last September. Mr. Altman said that GPT-5 and its successor, GPT-6, are “ready” and outperform their predecessors.

https://twitter.com/IbaMasood/status/1706009980200235142

In November, he made its existence public, telling the Financial Times that OpenAI was working on GPT-5, although he did not reveal its release date.

More recently, a report said that OpenAI's boss has come up with a bold plan to buy the large number of GPUs needed to train larger AI models.

To overcome the  supply shortage that is hindering technological innovation, Mr. Altman wants to mobilize up to $7 trillion to accelerate chip production with the help of a global network of investors, governments government and energy suppliers, according to  the Wall Street Journal.

<!-- wp:heading /-->

What can GPT-5 do?

In February, the head of OpenAI spoke about GPT-5 at the World Government Summit in Dubai. He said the next version of ChatGPT will be smarter than previous versions.

This is a bigger deal than it sounds, because what makes these models so magical is that they're general. So, if it's a little bit smarter, it's a little better at everything", Mr Altman explained.

Speaking about its power, Mr Altman told the FT that GPT-5 will need more data to train. He said the plan is to use publicly available datasets from the internet  , along with large-scale proprietary datasets from institutions. The last of these will include long-form articles or conversations in any format.

Speaking on Bill Gates' Unconfuse Me podcast in mid-January, Mr. Altman said: "Multimodality will definitely be important. Which means speech in, speech out. Images. Eventually video. Clearly, people really want that. We’ll be able to push that much further but maybe the most important areas of progress will be around reasoning ability."

He continued: "Currently, GPT-4 can only reason in extremely limited ways. There is also a need for reliability. If you ask most questions about GPT-4 10,000 times the result is The results are pretty good. But there will be one in 10,000 of those questions that it doesn't always know, and you want to get the best response from all 10,000 times, so increasing reliability will be important."

GPT-5 vs. GPT-4

So how can it beat GPT-4? Essentially, it will need to surpass GPT-4 Turbo, the next generation that OpenAI released in November for paid subscribers.

The company's most advanced AI chatbot has data through April 2023, compared to 2021 for GPT-4; it can parse long prompts of up to 128,000 tokens or roughly the length of a 300-page book; it would be better to follow the instructions; and it can automatically switch between engines, including the Dall-E  3 image generator and the Bing search engine  , based on user requests.

ChatGPT is banned by some schools and has proven itself capable of excelling in exams.

Both OpenAI and several researchers have also tested  chatbots in real-world tests. GPT-4 is proven to be very capable of passing the difficult financial analysis (CFA) exam. It scored 90% on the exam, scored well on the SAT reading and writing section, and placed in the 99% to 100% range in the 2020 US Biology Olympiad semifinals.

When will GPT-5 be ready for use?

At the time of writing, OpenAI has not announced a launch date for GPT-5. It's also unclear whether it was affected by the turmoil at OpenAI late last year. On November 17, Mr. Altman was fired by the company's board of directors. After five chaotic days of mixed views on the future of AI, Mr. Altman returned to the helm with a new board of directors.

https://twitter.com/mattshumer_/status/1736349957685223512

Oddly, some ChatGPT users previously said the bot told them it was running on a new AI model called GPT-4.5 Turbo, but that turned out to be incorrect.

If we look back at OpenAI's GPT release timelines, we can see that the gap between updates is getting shorter and shorter. GPT-1 launched in June 2018, followed by GPT-2 in February 2019, then GPT-3 in June 2020, and the current free version of ChatGPT (GPT 3.5) in December /2022, while GPT-4 will launch just three months later, March 2023. More frequent updates have also appeared in recent months, including a “turbo” version.

So June could be the month that GPT-5 will launch? Or will OpenAI wait until its developer conference in November to announce GPT-5? Let's wait and see!