OpenAI’s ambitious new artificial intelligence project, GPT-5 (codenamed Orion) faces significant hurdles. The project is behind schedule and incurring substantial costs, with its ultimate success uncertain. Some even question whether sufficient data exists to achieve the desired level of intelligence.
The project, designed to be a major leap forward from the technology powering ChatGPT, has been in development for over 18 months. Microsoft, OpenAI’s key partner and largest investor, had anticipated the release of this new model around mid-2024, according to sources familiar with the situation.
OpenAI has undertaken at least two extensive training runs for Orion, each involving months of processing vast datasets to enhance its intelligence. However, sources close to the project reveal that each attempt has encountered fresh obstacles, with the software failing to meet researchers’ performance expectations.
Insiders suggest that, at best, Orion shows marginal improvements over OpenAI’s existing technology. Critically, these gains don’t justify the immense operational expenses. Estimates, based on both public and private data regarding various training components, indicate that a single six-month training run can rack up approximately half a billion dollars in computing costs alone.
Analysts project that tech giants could invest a staggering $1 trillion in AI projects in the coming years. Much of this expectation rests on the shoulders of OpenAI, the company that spearheaded the AI revolution.
The company’s $157 billion valuation, assigned by investors in October, is largely based on Altman’s forecast that GPT-5 will mark a “significant leap forward” across a wide range of subjects and tasks.
Dear CHATGPT users, openai is collecting personal data so use it with your own risk, check what italy has to say, and make sure you are not sharing you personal or your gf/family photos or vidoes, unless you want to be seen, such as on social media which are collected, just read the terms and conditions