what we know about OpenAI’s next chatbot

ChatGPT ended a huge 2023 with a series of huge updates, but this year could be even bigger for the artificial intelligence chatbot.

In recent months, hype has been building around a new and more powerful version of the technology that underpins ChatGPT.

This so-called large language model is called GPT-5. It is seen as a stepping stone towards general artificial intelligence, or a machine that can think like a human.

Here’s what we know about the successor to ChatGPT.

What is GPT-5?

GPT-5 is the follow-up to GPT-4, OpenAI’s most advanced chatbot for which you need to pay a monthly subscription.

GPT stands for pre-trained generative transformer, which is a type of large language model that can create text and content similar to human images. In the case of ChatGPT, the AI ​​chatbot can answer questions conversationally.

OpenAI CEO Sam Altman told the Financial Times in November that his company was working on GPT-5, although he did not reveal a timeline for its release.

Altman previously quashed rumors that the firm was training the new AI model in March and June.

However, things changed in the autumn. Altman allegedly talked about GPT-5 and GPT-6 during a speech at an alumni gathering of his former venture capital firm Y Combinator in September, according to two people who attended the event.

Both AI models were “in the bag,” said Omar Shams, founder and CEO of Mutable AI.

Another attendee, Iba Masood, founder of Y Combinator-backed AI startup, said Altman also discussed how GPT-5 and 6 were better than their predecessors.

What can GPT-5 do?

According to Masood, Altman said that GPT-5 and 6 “would do a more reliable job, with better personalization, with multimodal outputs.”

The story continues

Multimodal AI is trained and uses content including images, audio, video and numerical data. OpenAI says that GPT-4 is multimodal, meaning it can accept text or image inputs but can only output text.

Altman said the next ChatGPT still fell short of general artificial intelligence, according to Masood and Shams.

Showing off his smarts, the OpenAI boss told the FT that GPT-5 would need more data to train it. The plan, he said, was to use publicly available data sets from the internet, as well as large-scale proprietary data sets from organizations. The last of those would be long writing or conversations in any format.

However, Altman said it was difficult to predict the model’s new abilities and skills until training had begun.

GPT-5 versus GPT-4

So how can it beat GPT-4? Mainly, it will have to surpass GPT-4 Turbo, the next-generation model that OpenAI released in November to paying subscribers.

The company’s most advanced AI chatbot has knowledge of global events up to April 2023, compared to 2021 for GPT-4; it can analyze even longer clues of up to 128,000 characters or about the length of a 300-page book; it is better to follow the instructions; and can automatically switch between tools, including the Dall-E 3 image generator and the Bing search engine, based on user requests.

Both OpenAI and several researchers have tested the chatbot on real-life exams. GPT-4 was shown to have a good chance of passing the extremely difficult chart financial analyst (CFA) exam; scored in the 90th percentile of the bar exam; they got the SAT reading and writing section; and was in the 99th to 100th percentile on the 2020 USA Biology Olympiad Semifinal Exam.

When will GPT-5 be available?

As of this writing, OpenAI has not announced a launch date for GPT-5. It looked like he didn’t even start training the AI ​​model as recently as November.

It’s also unclear whether it was affected by the turmoil at OpenAI late last year. On November 17, Altman was fired by the company’s board of directors. After five days of turmoil that symbolized the dueling views on the future of AI, Altman was back at the helm with a new board.

Curiously, some ChatGPT users recently claimed that the bot told them it was running on a new AI model called GPT-4.5 Turbo, but that turned out to be an error.