OpenAI, the company behind the viral chatbot ChatGPT, released a highly anticipated new version of its artificial intelligence system. OpenAI said its new GPT-4 artificial intelligence model is safer and better at answering questions than ChatGPT.
OpenAI has released GPT-4, the latest version of its artificial intelligence chatbot ChatGPT. GPT-4. GPT stands for Generative Pre-trained Transformer and this is version 4, hence GPT-4. It is a neural network that is based on OpenAI’s large multimodal model that exhibits human-level performance on various professional and academic benchmarks.
Announcing GPT-4, a large multimodal model, with our best-ever results on capabilities and alignment: https://t.co/TwLFssyALF pic.twitter.com/lYWwPjZbSg
— OpenAI (@OpenAI) March 14, 2023
- Advertisement -
The new model can respond to images – providing recipe suggestions from photos of ingredients, for example, as well as writing captions and descriptions.
GPT-4: new features, visual input, availability, and more
GPT-4 is OpenAI’s most advanced system. GPT-4. GPT stands for Generative Pre-trained Transformer and this is version 4, hence GPT-4. It is a neural network that is based on OpenAI’s large multimodal model that exhibits human-level performance on various professional and academic benchmarks.
GPT-4 is the newest version of OpenAI’s large language model systems, which are trained to predict the next word in a sentence by ingesting massive amounts of text from the internet and finding patterns through trial and error.
1. GPT-4 can analyze and understand images
GPT-4 is able to see and comprehend pictures This adaptable machine learning system has undergone the most notable change in that it is now “multimodal,” which means it can comprehend multiple “modalities” of information. They could read and write, but that was about it for ChatGPT and GPT-3, which were restricted to text. (though more than enough for many applications). However, GPT-4 can be instructed to process images in order to discover pertinent information. Of course, you could just ask it to describe what’s in the image, but more significantly, its comprehension extends beyond that.
Of course, you could just ask it to describe what’s in the image, but more significantly, its comprehension extends beyond that. However, the collaboration with Be My Eyes, an app used by blind and low-vision people to let volunteers describe what their phone sees, is more telling. The example given by OpenAI actually has it describing the joke in a picture of a hilariously oversized iPhone connector.
2. The larger memory of GPT-4
There is a limit to how much these big language models can “remember” during a discussion with a user despite having been trained on millions of web pages, books, and other text data. (one sympathizes). This restriction applied to GPT-3.5 and the previous version of ChatGPT. It was 4,096 “tokens,” or approximately 8,000 words, or four to five book pages.
As a result, once something had moved that far “back” in its attention function, it would start to lose track of objects. The maximum token count for GPT-4 is 32,768; if you’re wondering why the number appears familiar, it’s because it’s 215. This equates to approximately 64,000 words or 50 pages of text, which is sufficient for a complete play or short narrative.
This implies that it will be able to remember roughly 50 pages during a conversation or when writing text. As a result, it will recall the 20 pages of chat you just had or, if you’re writing an essay or a narrative, it might make reference to 35 pages of earlier events.
3. GPT-4 has more language options
English speakers predominate in the field of artificial intelligence, and everything from data to study papers to testing is written in this language. However, it goes without saying that large language models’ capabilities should be made accessible in all written languages.
GPT-4 moves in that direction by proving that it can accurately respond to thousands of multiple-choice queries in 26 different languages, including Arabic, Italian, Ukrainian, and Korean. Although it generalizes well to other languages, it excels at Romance and Germanic languages.
Who is OpenAI, the creator of GPT-4?
The San Francisco-based artificial intelligence lab started in 2015 as a nonprofit, trying to build “Artificial General Intelligence,” or AGI, which is essentially software that’s as smart as humans. It was founded with a combined $1 billion pledge from chief executive Sam Altman, Elon Musk, billionaire venture capitalist Peter Thiel and others. (Musk later parted ways with the organization.)
OpenAI has developed several advanced AI technologies, including the GPT (Generative Pre-trained Transformer) series of language models, such as GPT-2 and GPT-3. The company’s goal is to advance AI in a way that benefits society while minimizing potential risks and negative impacts.
The company wanted to protect against a future in which big tech companies, such as Google, mastered AI technology and monopolized its benefits. The nonprofit’s goal was to build AI software transparently and make its products open-source so more people would be able to access it. Microsoft later invested in OpenAI and released a chatbot earlier this year with technology developed together.
What do you think about GPT-4? How will it enhance your workflow? Please share with us in the comments area below.
