|
|
|
|
|
|
|
|
|
|
Is Superintelligent AI Ready to Take Over?
|
|
|
|
|
|
|
|
|
|
|
Illustration: Jarred Briggs
|
|
|
|
|
|
Good morning. When did everyone go from talking about "artificial general intelligence," where AI performs at human level ability, to "superintelligence," where the technology outperforms humans in, well, everything? More importantly... are we cool with that?
Nevertheless, leaders in the field are pushing forward to hit this milestone. This pursuit helped motivate Meta Platforms to pour millions into ScaleAI and land its CEO, Alexandr Wang, on the superintelligence team. And OpenAI CEO Sam Altman, in a post last week, said we were closer to superintelligence than ever.
But before humans just pack it in, WSJ Tech Columnist Christopher Mims finds researchers who "aren’t buying all that talk."
|
|
|
|
|
In areas other than coding and mathematics, the latest models aren’t getting better at the rate that they once did. And the newest reasoning models actually hallucinate more than their predecessors.
“The broad idea that reasoning and intelligence come with greater scale of models is probably false,” says Jorge Ortiz, an associate professor of engineering at Rutgers, whose lab uses reasoning models and other cutting-edge AI to sense real-world environments. Today’s models have inherent limitations that make them bad at following explicit instructions—the opposite of what you’d expect from a computer, he adds.
It’s as if the industry is creating engines of free association. They’re skilled at confabulation, but we’re asking them to take on the roles of consistent, rule-following engineers or accountants.
|
Analysis by the WSJ's Christopher Mims
|
|
|
|
|
|
|
|
|
|
|
Content from our sponsor: Deloitte
|
|
|
Databricks VP of AI: Take Generative AI From Useful to Transformative
|
|
Extracting increased business value from generative AI may start with understanding the technology’s limitations, says Naveen Rao of Databricks. Read More
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
AWS, Amazon’s cloud-computing arm, has trained over 400,000 people in Australia since 2017 to develop digital skills Photo: Kazuhiro Nogi/Agence France-Presse/Getty Images
|
|
|
|
|
|
Amazon plans to pour some $13 billion into data center infrastructure in Australia between through 2029, WSJ reports. Over the past few weeks, Amazon has announced plans to spend at least $20 billion in Pennsylvania and $10 billion in North Carolina to expand its AI infrastructure, as well as a $5 billion investment in Taiwan to launch a new cloud services region on the island.
|
|
|
|
|
Apparel maker H&M is deploying AI to strengthen its operations and improve customers’ experience when shopping across its physical and digital channels, WSJ reports.
|
|
|
|
|
Anne Wojcicki, the co-founder and former CEO of 23andMe, is poised to regain control of the DNA-testing company after a nonprofit she controls topped a prior bid, WSJ reports.
|
|
|
UK telecom giant BT plans to cut more than 40,000 jobs by the end of the decade, CEO Allison Kirkby tells the FT. And there could be more as AI improves, she said.
|
|
|
|
|
|
|
|
Everything Else You Need to Know
|
|
|
|
A man suspected of shooting two state Democratic lawmakers was arrested late Sunday after authorities tracked him into an area with crops and woodlands in rural Minnesota, authorities said. (WSJ)
Israel is reshaping the Middle East on its own terms and forcing the Trump administration to play catch-up as it ramps up attacks against Iran. The moves could upend global markets and remake geopolitics. (WSJ)
Nations are eyeing the G-7 summit in Canada this week as an opportunity to strike trade deals with President Trump—or at least build momentum to keep talking and ease tensions over tariffs. (WSJ)
The U.S. labor market is holding steady despite extraordinary economic upheaval. But it is a bad time to be a job seeker—especially if you are young. (WSJ)
|
|
|
|
|
|
|
|
|
|
|
|
