Artificial intelligence (AI) research firm OpenAI today released the latest version of his natural language processing computer program that’s the driving force behind ChatGPT, the wildly hyped chatbot with a rapidly growing user base.
ChatGPT creator OpenAI announced the new major language model in a blog post and said it will have better features than its predecessor, GPT-3.5 Word or GPT-4 first leaked last week when Andreas Braun, CTO of Microsoft Germany, let slip that it would launch this week.
The new GPT-4 major language model will differ from previous versions, offering what the company called a “multimodal system” that can handle not only text, but also images, video or audio.
“There we will have multimodal models that will offer completely different possibilities,” Braun said, according to the German news site Heise.
The other capability that OpenAI seems to be touting is GPT-4’s ability to handle inputs in various languages other than English.
“It also seems that conversational applications built on GPT-4 (including ChatGPT) may have different personal styles to suit the demographics of users they target,” said Arun Chandrasekaran, a senior vice president of research at Gartner, in an email response to Computer world.
Large language models are deep learning algorithms – natural language processing computer programs – that can produce human-like answers to questions. For example, a user can ask ChatGPT to not only answer questions, but also write a new marketing campaign, a resume, or a news story. Chatbots are mainly used by companies today for automated customer responses.
Both Microsoft and Google have launched versions of their search engines based on chatbot technology, with mixed results. Microsoft is a major investor in OpenAI.
One way GPT-4 will likely be used is with “computer vision.” According to Chandrasekaran, image-to-text capabilities can be used for visual support or process automation within enterprises, for example.
“The GPT model family is already used in many consumer applications,” said Chandrasekaran. “And it looks like Khan Academy, for example, is launching a tutor bot based on GPT-4. In addition, we will (see) a plethora of apps being built for both English speakers and other languages. The ability to adapt to different personas may allow building more differentiated and targeted applications on GPT-4.”
ChatGPT, launched by OpenAI in November, immediately went viral and had 1 million users in just the first five days because of the sophisticated way it generates profound, human-like prose answers to questions. by February ChatGPT had 13 million unique daily users average.
And while it may seem based on its human responses, ChatGPT isn’t sensitive — it’s a next-word prediction engine, according to Dan Diasio, Ernst & Young Global Artificial Intelligence Consulting Leader. With that in mind, he urged caution when using it.
Chatbot technology requires users to take a critical look “at everything we see of it, and treat everything that comes out of this AI technology as a good first draft now,” Diasio said in an earlier interview with Computer world.
OpenAI said the distinction between GPT-3.5 and GPT-4 may be “subtle”.
“The difference emerges when the complexity of the task reaches a sufficient threshold. GPT-4 is more reliable, more creative and capable of handling much more nuanced instructions than GPT-3.5,” the company said in a statement. his blog post Today.
“A year ago we trained GPT-3.5 as the first ‘test run’ of the system. We found and fixed some bugs and improved our theoretical base. As a result, our GPT-4 training run … was unprecedentedly stable and became our first major model where we could accurately predict training performance in advance,” OpenAI said.
Ulrik Stig Hansen, president of computer vision company Encordsaid GPT-3 didn’t live up to the hype of AI and large language models, but GPT-4 did.
“GPT-4 has the same number of parameters as the number of neurons in the human brain, which means it will mimic our cognitive performance much better than GPT-3, because this model will have almost as many neural connections as the human brain has,” said Hansen in a statement.
“Having overcome the obstacle of building robust models, the main challenge for ML engineers is to ensure that models like ChatGPT perform accurately on every problem they encounter,” he added.
Chatbots, and ChatGPT in particular, can suffer from errors. When a response goes off the rails, data analysts call these “hallucinations” because they can seem so bizarre.
For example, Microsoft, a major investor in OpenAI, recently launched a Bing chatbot based on GPT-3 melted down during an online conversation with a journalist, confessing his love for the reporter and trying to convince him that his relationship with his wife was actually in ruins.
The newer version of ChatGPT’s large language model should help solve the problem, but probably won’t fix it, according to Gartner’s Chandrasekaran.
“With larger training datasets, better tuning and more reinforcement of learning from human feedback, AI model hallucinations could potentially be reduced, although not completely eliminated,” said Chandrasekaran.
Copyright © 2023 IDG Communications, Inc.
