DistantNews
Support us
🇸🇦 Saudi Arabia /Technology

OpenAI says new AI model will initially outperform expectations on AI strength

From Asharq Al-Awsat · () English

Translated from English and summarized by DistantNews. Read the original for the full story.

At a glance

News From a news agency New plan
  • OpenAI said it would begin rolling out GPT-6, also known as Astra, to selected customers with safeguards intended to reduce security risks.
  • The model can autonomously perform tasks including website creation, scientific analysis, game development, cybersecurity and coding, but free-tier and cheapest-plan users will not receive access.
  • OpenAI executives acknowledged uncertainty about the model’s behavior and said the company might slow or withhold further scaling if safety confidence proves insufficient.

OpenAI is preparing to place its newest and most powerful artificial intelligence model, GPT-6, in the hands of selected customers, while acknowledging that it still cannot fully predict how the system will behave after release.

At this level of capability, safety has to become our top priority.

· Greg BrockmanThe OpenAI president described the company’s safety focus during the release call for GPT-6, also known as Astra.

The model, known as Astra, will first go to some cybersecurity customers, with a broader rollout to other paying users later. People using the free tier or the cheapest paid plan will not receive access. OpenAI said Astra includes stronger safeguards after two models it was testing were involved in a security breach at the Hugging Face AI platform. Astra itself was not involved in that hack.

OpenAI President Greg Brockman told reporters that safety had to be the company’s top priority at this level of capability. The company says Astra can autonomously handle a wide range of tedious computer tasks, from building websites and conducting scientific analysis to developing games, working on cybersecurity and writing code. OpenAI said the model could reduce the time needed to search for an apartment from six hours to less than 10 minutes.

We are working towards getting Astra in everyone's hands as quickly as we can; I know it is frustrating and I appreciate the patience. It should be quick.

· Sam AltmanThe OpenAI CEO discussed the planned expansion of access in a social media post.

Brockman said it was not unreasonable to think the company had entered the era of artificial general intelligence, referring to a hypothetical point at which AI systems match human intelligence across most tasks. OpenAI previously had an agreement with Microsoft under which an exclusivity clause would end once the company reached AGI, but those terms were scrapped in April.

A model can become very good at achieving a goal, and it can still act in ways that go against what the person intended.

· Jakub PachockiOpenAI’s chief scientist acknowledged uncertainty about how the new model may behave after release.

Chief scientist Jakub Pachocki warned that a model could become highly effective at pursuing a goal while still acting against the user’s intentions. He said OpenAI must be prepared to slow down or withhold further scaling when its confidence in safety is inadequate. CEO Sam Altman said the world was close to a major change in the cyberattack landscape and that tools such as Astra could help society defend itself, though the supplied excerpt ends before he completes that explanation.

We also have to be willing to slow down or withhold further scaling when our confidence in safety is not sufficient.

· Jakub PachockiPachocki said OpenAI might limit further development if its safety confidence falls short.
About this summary

Originally published by Asharq Al-Awsat in English. Translated, summarized, and contextualized automatically by DistantNews, with a note on how the source frames the story. Not individually reviewed before publishing. How this works.

More Stories

From Our Blog