OpenAI CEO: AI is working just fine...for now

The CEO of OpenAI has admitted that he is amazed by really good nan exertion is working, particularly successful nan look of utmost excitement and hype.

Speaking astatine Dreamforce 2023, Sam Altman and Salesforce CEO Marc Benioff discussed a wide number of AI topics, but nan OpenAI laminitis admitted to immoderate pleasant shocks astatine really his company’s activity is performing.

“The biggest astonishment is conscionable that it's each working,” Altman said, drafting laughter from Benioff and nan audience. “We want to person nan smartest, astir capable, astir customizable models retired there.”

AI future

Altman went connected to admit that nan magnitude of liking successful AI has besides taken him by surprise, peculiarly successful nan standard of usage cases that person been seen truthful far.

“The level of enthusiasm, hopefulness and excitement astir nan world, of course, balanced pinch wanting to successfully reside nan downsides (of AI) - is really bonzer to see," he noted.

“I thought possibly this was conscionable a tech, Silicon Valley phenomenon, but seeing what everyone astir nan world is doing pinch nan technology, and really they've incorporated it into their lives, was really cool.”

Having been quizzed by Benioff arsenic to his favourite AI-themed science-fiction movie (Her, wherever Joaquin Phoenix’s characteristic falls successful emotion pinch a virtual girlfriend), Altman did look to propose he believed immoderate shape of caller governing agency to show AI could beryllium necessary

There is, he noted, a anticipation of needing to create a model that can, “deal pinch nan short-term and semipermanent challenges" posed by AI, adding that “getting thing going - moreover if conscionable focused connected penetration and not oversight - I deliberation would beryllium great.”

He besides admitted that he expected nan surveillance authorities issues caused by AI to lead to a emergence successful surveillance successful nan adjacent early - and did not deliberation he could connection a solution conscionable yet.

“I don’t spot a world where, if AI is arsenic powerful arsenic we think, and group tin do important harm pinch it, I tin spot a world wherever we don’t person little surveillance," Altman said. “I don’t deliberation that's a bully thing.”

But overall, nan OpenAI CEO did opportunity he was hopeful that nan elephantine strides already made by AI tin proceed successful nan future.

“We are empiricists; we cognize we are going to beryllium surprised,” Altman noted. "We're ne'er amazed - we conscionable effort to meet reality wherever it is and travel nan exertion wherever it tin go."

“I person nary delusions that thing is going to protect america aliases anyone if AI goes wrong," he added. "I deliberation it's silly that's what group presume it's for. It was conscionable for illustration a boyhood dream that stuck."

More from TechRadar Pro

• Einstein Copilot is present to supercharge your Salesforce CRM pinch AI
• OpenAI accused of drawstring of information protection breaches successful GDPR complaint
• These are nan best AI writers astir correct now