The recent artificial intelligence (AI) project by OpenAI called GPT-5 is said to be long overdue, over budget, and its chances of success remain very elusive. Citing the Wall Street Journal (WSJ) this Friday (Dec 20), GPT-5 codenamed Orion may lack enough global data to make it as advanced as initially intended.
Development on GPT-5 has been ongoing for over 18 months, with the model considered a major upgrade to the technology behind OpenAI’s ChatGPT. Sources told WSJ that Microsoft, OpenAI’s largest investor, had hoped for a mid-year release.
OpenAI really had a couple of broad training runs that lasted for months spent gathering data to make Orion more intelligent. However, it was noted that new challenges were coming up at each test, and usually, the software did not match researchers’ expectations.
Nonetheless, Orion has not progressed sufficiently even under optimal conditions to absolve the new model of its operating charges. The Wall Street Journal estimates that a run of training would cost about half a billion dollars for a six-month duration.
PYMNTS made requests for comments from OpenAI but has not received any reply as of yet. WSJ raised the point about all these very high expectations that OpenAI has had placed on it because of the launch of ChatGPT in 2022, with analysts now saying tech companies might spend as much as $1 trillion on AI projects over the next several years. OpenAI’s valuation of $157 billion in October was evidence of the optimistic assumptions regarding CEO Sam Altman’s prediction that GPT-5 would bring to “significant advancements” in many domains and tasks.
Earlier this year, PYMNTS discussed GPT-5’s potential release, suggesting it could mark a pivotal moment in AI development or be merely another incremental upgrade. With high expectations, experts are analyzing how the next-generation model might impact industries from healthcare to finance.
Cash Merrill, Founder and CTO of AI software company Zibtek, told PYMNTS:
“For commerce, the implications of more advanced LLMs [Large Language Models] like GPT-5 are enormous. We could see major improvements in customer service bots, providing more coherent and contextually relevant interactions without human intervention. In digital marketing, content creation could become more sophisticated and personalized, enhancing engagement strategies. Additionally, advanced LLMs can streamline operations like contract analysis and risk assessment by processing and analyzing large volumes of text-based data with high accuracy.”
