OpenAI's Rocky GPT-5 Rollout Shows Struggle to Remain Undisputed AI Leader

Hindustan Times

a day ago

OpenAI's newest AI model , GPT-5, was supposed to cement the startup's status as the undisputed leader in the AI race. Instead, it has had a tumultuous public launch, frustrating users and prompting Chief Executive Officer Sam Altman to respond.
Users have flooded social media with embarrassing examples of how the chatbot failed to answer simple math questions or accurately draw a map of North America. Others have criticized its colder tone, reminiscing about older models that OpenAI initially killed off. A new limit of 200 questions a week rankled devotees.
Altman on Tuesday promised to imbue GPT-5 with a 'warmer personality,' restored a popular model after OpenAI declared it to be obsolete and introduced the capability for users to decide which kind of query they want to make. The company has learned how much users expect customization, he said in a post on X.
Computing capacity has been a concern during the rollout, prompting the company to re-evaluate which users are priorities. The turbulence shows the challenge OpenAI has in selling 700 million active weekly users on new models while being constantly short on computing power, which is expensive and scarce.
The rollout comes at a delicate moment for OpenAI, which is struggling to keep its lead among rivals who are aggressively going after its talent and pouring billions of dollars into AI research. Competitors such as Anthropic's Claude have surged in popularity among coders and business users. Whether GPT-5 will enable OpenAI to win over such customers remains to be seen.
'We expected some bumpiness as we roll out so many things at once,' he said earlier on X. 'But it was a little more bumpy than we hoped for!'
When GPT-5 was released last week, Jason Pollak, a digital-marketing specialist based in Atlanta, was excited and thought it would be a perfect expansion from GPT-4.
Instead, the model feels 'bland, generic' and less like a creative thinking partner, Pollak said.
Pollak pays for ChatGPT and uses it regularly to work on copy, analyze data or create custom tools. The transition from GPT-4 to GPT-5 has been frustrating, he said.
'It feels like they took the Ph.D. thing a little too seriously to prove how smart it would be,' he said.
GPT-5's release came after a two-year wait in which it experienced a series of delays and setbacks, leading to speculation that either the AI startup had lost its edge or that AI development was more broadly hitting a wall. Altman dismissed those concerns on a call with reporters last week, saying the startup had found new breakthroughs.
Juliette Haas, an account-strategy coordinator at a communications and crisis-management agency, primarily uses the paid version of ChatGPT for brainstorming and to complete administrative tasks such as creating a to-do list.
With the release of GPT-5, she decided to revisit a business-development prompt to figure out which companies or individuals at her firm would require her support. With GPT-4, the response suggested that she build strong industry connections and emphasized the importance of relationship building. GPT-5 delivered a checklist.
'The AI treated finding distressed companies more like a data-science problem rather than understanding the fundamental considerations of relationships and timing,' said Haas.
She finds that GPT-5 doesn't understand context and subtext as much as GPT-4 did, and GPT-5 delivers more 'data-driven solutions.'
Jim Marsh, founder of JMC Strategic Intelligence, a consulting firm, uses the premium version of ChatGPT and has been a user since 2022.
He said he is impressed with the results of GPT-5, finding that it was able to build a database of his contacts without many errors.
The industry has reached a point in AI innovation where 'every release isn't going to be a magic trick,' which is perhaps why there is so much criticism about the newest model, he said.
'There's still this expectation that AI is going to quickly do more for us without us needing to be involved, and that's generally a false belief,' he said.
Write to Ann-Marie Alcántara at [e-mail] and Berber Jin at [e-mail]