Artificial Intelligence

[gfp927z]

>>> GPT-4 has arrived. Here’s what to expect from the new AI.


The Washington Post

by Rachel Lerman

March 2023


https://www.msn.com/en-us/news/technology/gpt-4-has-arrived-here-s-what-to-expect-from-the-new-ai/ar-AA18CTF2?OCID=ansmsnnews11


OpenAI, the company behind the viral chatbot ChatGPT, released a highly anticipated new version of its artificial intelligence system Tuesday.

The firm pledges that GPT-4 will superpower its chatbot — and that it has the ability to generate answers not just from written prompts, but also from images.

Popularized by chatbots that can interact with users in a conversational tone, AI technology has taken the world by storm in the past several months. The technology can be creative and useful, but researchers also warn about the ethical implications of its capabilities — the systems have been shown to illuminate biases and dispense misinformation.

GPT-4 has arrived. It will blow ChatGPT out of the water

GPT-4 has been eagerly awaited by the AI community for years, and it could mark a turning point in how AI is used in everyday life.

So what exactly is GPT-4, and how will it be used?

What is GPT-4?

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.

OpenAI calls GPT-4 a “large multimodal model” because it can accept text and images and respond in text.

GPT-4 will be available in a limited format on ChatGPT Plus, a paid version of the company’s chatbot. It will also be available for businesses to incorporate into other products, after they make it off the waitlist.

Microsoft also announced on Tuesday that its Bing chatbot is already using a version of GPT-4 to power the bot.

AI can now create any image in seconds, bringing wonder and danger

What can GPT-4 do that ChatGPT can’t?

ChatGPT burst into public view in November and instantly became a sensation. The conversational chatbot can take prompts from users and generate stories, essays, computer code, a back-and-forth dialogue or nearly whatever you ask it to do. But its answers are not always correct or appropriate.

ChatGPT is built based on a large language model called GPT-3.5, an earlier version of the technology announced Tuesday.

OpenAI said that GPT-4 can place in the 90th percentile of test-takers for the Uniform Bar Exam, the certification test for lawyers. It’s also 82 percent less likely than GPT-3.5 to respond to queries for “disallowed content,” the company said, making it safer.

In a video OpenAI released with its announcement Tuesday, the company said GPT-4 can accept longer text inputs than its predecessors — taking in and generating up to 25,000 words, compared with 3,000 words for ChatGPT. It is trained to be safer and more factual, OpenAI said.

“It’s a system that can make dreams, thoughts, ideas flourish in text in front of you,” an OpenAI employee said in the company’s video announcement.

The system can also answer questions based on what an image depicts, OpenAI said. But that capability won’t immediately be publicly available.

GPT-4 is designed to be better at answering complicated questions, the company said.

What is ChatGPT, the viral social media AI?

What 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.)

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.

OpenAI’s technology went viral last year even before ChatGPT when it opened its Dall-E image generator for anyone’s use.

Big Tech was moving cautiously on AI. Then came ChatGPT.

Why am I hearing so much about AI and chatbots all of a sudden?

AI technology has been in the works for decades, and you’ve been using some version of artificial technology embedded in software, search systems and smart speakers for years. You encounter AI when you unlock your phone with facial recognition, run a Google search or rely on spell-checking software.

But ChatGPT’s public debut in November stunned many people with how far the technology had advanced and the ease at which it appears to interact with users in plain-language text.

Since then, many other prominent companies have unveiled more details about their AI ambitions. Microsoft released a chatbot within search engine Bing that it developed with OpenAI. Google said it has its own bot, known as Bard. And Facebook parent Meta has been working on similar technologies.

ChatGPT’s wild success with the public has accelerated the AI arms race among tech giants, prompting pressure within the companies to move faster on the technology than they had previously planned.

<<<



---