ChatGPT 4 is coming, and rumors suggest it might bring massive improvements to the already incredibly impressive language skills of OpenAI’s ChatGPT.

To be clear, ChatGPT 4 is unlikely to be the name of OpenAI’s next product, but we took a bit of creative license and combined the ChatGPT name with the improved AI model that will drive it in the future, GPT-4. Let’s dig into GPT-4, how ChatGPT works now, and when OpenAI might release its next major upgrade.

What is GPT-4?

A Midjourney cartoon of a robot chatting with a woman using a laptop.
A Midjourney cartoon where a robot chats with a woman using a laptop. Midjourney render prompted by Tracey Truly

GPT-4 is a new language model being created by OpenAI that can generate text that is similar to human speech. It will advance the technology used by ChatGPT, which is based on GPT-3.5. GPT is the acronym for Generative Pre-trained Transformer, a deep learning technology that uses artificial neural networks to write like a human.

Billions of parameters are used to adjust the output of ChatGPT’s simulated neurons to achieve the desired result. The enormous amount of detail helps a computer come closer to the interconnectedness of a human brain, which also contains billions of neurons. OpenAI hasn’t confirmed any details about GPT-4 but does acknowledge that it is in progress. Rumors, however, are plentiful.

For example, as far back as August 2021, Wired reported that industry experts speculated that GPT-4 would feature 100 trillion parameters. Building an AI with more parameters doesn’t necessarily guarantee better performance; however, and could affect responsiveness.

Other rumors suggest better computer code generation and the ability to generate images as well as text from the same chat interface. An AI that can generate video is also expected.

Handling text, images, and video is known as a multimodal model. Machine learning expert Emil Wallner tweeted that GPT-4  might have this capability.


– GPT-4 is coming, but improved Codex is the focus
– The next Codex release will be significant
– GPT-4 will have a similar model size but better trained
– DALL-E will become public
– They bet that multi-modal models will beat pure text models on text generation

— Emil Wallner (@EmilWallner) September 6, 2021

A hint to possible GPT-4 details came from a price sheet shared by Travis Fischer on Twitter. A new developer product called Foundry will include more than one model, adding two “DV” models that can handle more words at a time as input (the max context).

OpenAI has privately announced a new developer product called Foundry, which enables customers to run OpenAI model inference at scale w/ dedicated capacity.

It also reveals that DV (Davinci; likely GPT-4) will have up to 32k max context length in the public version. 🔥

— Travis Fischer (@transitive_bs) February 21, 2023

GPT-3.5 works with up to 2,048 words, but the two DV models, which might be GPT-4, have four times (8K resolution) and 16 times (32K) that capacity, allowing much more in-depth discussions to solve much bigger challenges.

Ultimately, this is all guesswork, and all that we know for certain is that GPT-4 is in progress and that it could bring a significant improvement to the results possible with ChatGPT.

Does GPT-4 have a release date?

A laptop opened to the ChatGPT website.

GPT-4 is coming, and it will probably arrive this year. We don’t have a firm release date yet, though. The New York Times suggested it could arrive as soon as the first quarter of this year. Since we’re in March, that would place the launch in a matter of weeks.

With technology moving so quickly, it might feel like the AI revolution could arrive at any time, bringing disruptive change to many industries and upsetting the balance of capitalism as jobs are lost to computers.

A document that might take a human several hours to write is completed by ChatGPT in seconds. In light of recent breakthroughs, Artificial General Intelligence (AGI) almost seems inevitable.

OpenAI made it clear in a recent blog post; however, that the company won’t be rushing any products out before they are thoroughly tested, stating, “As our systems get closer to AGI, we are becoming increasingly cautious with the creation and deployment of our models.”

OpenAI hasn’t shared any timeline for the release of GPT-4.

Does Bing Chat use GPT-4?

The new Bing chat preview can be seen even on a MacBook.
Photo by Alan Truly

Microsoft states that the new Bing, or Bing Chat, is more powerful than ChatGPT. Since OpenAI’s chat uses GPT-3.5, there is an implication that Bing Chat could be using GPT-4. This hasn’t been confirmed.

Clearly, Bing Chat has been upgraded with the ability to access current information via the internet, a huge improvement over ChatGPT, which can only draw from the training it received through 2021.

In addition to internet access, the AI model used for Bing Chat is much faster, something that is extremely important when taken out of the lab and added to a search engine.

It seems unlikely that will be equivalent to OpenAI’s GPT-4 model. There would be no need for caution if GPT-4 was already publicly available.

GPT-4: an evolution, not a revolution

While GPT-4 will almost certainly be impressive, OpenAI CEO Sam Altman said in a StrictlyVC interview posted by Connie Loizos on YouTube that “people are begging to be disappointed, and they will be.”

StrictlyVC in conversation with Sam Altman, part two (OpenAI)

Altman acknowledges the potential of AGI to wreak havoc on world economies and expressed that a quick rollout of several small changes is better than a shocking advancement that provides little opportunity for the world to adapt to the changes.

In short, GPT-4 will be an evolution, not a revolution. According to Altman, the next version of ChatGPT won’t be an AGI and it won’t have 100 trillion parameters. Those rumors are incorrect.

Editors’ Recommendations

Source link


Leave a Reply

Your email address will not be published. Required fields are marked *