Business

From OpenAI to Nvidia, researchers agree: AI agents have a long way to go

Published

4 months ago

August 5, 2025

Welcome to Eye on AI! AI reporter Sharon Goldman here, filling in for Jeremy Kahn, who is on holiday. In this edition…General Services Administration approves OpenAI, Google, Anthropic for federal AI vendor list…Consequences of AI spending boom on U.S. economy…Clay AI raises $100 million at $3.1 billion valuation.

Only in the Bay Area does spending a Saturday geeking out about AI agents—alongside 2,000 students, researchers, and tech insiders crammed into UC Berkeley—feel like a totally normal weekend plan. As I picked up my badge at the day-long Agentic AI Summit and watched the line snake through the student union lobby, it felt less like an academic conference and more like Silicon Valley’s version of a buzzy New York brunch spot.

This was certainly due to the speaker lineup, which was stacked with top AI researchers and scientists, including Jakob Pachocki, chief scientist at OpenAI; Ed Chi, VP of research at Google DeepMind; Bill Dally, chief scientist at Nvidia; Ion Stoica, cofounder at Databricks & Anyscale, as well as a UC Berkeley professor; and Dawn Song, a pioneering UC Berkeley professor focused on AI security.

The popularity might have been due to the buzzy topic—AI agents, generally defined as an AI-powered system that can complete tasks, mostly autonomously, using other software tools. Think not only suggested a vacation itinerary, but also booking the flight and making the hotel reservation.

As my colleague Jeremy Kahn said in a recent article, “This kind of automation is a perennial C-suite fever dream. Over the past decade, companies embraced ‘robotic process automation,’ or RPA. This was software that could automate repetitive tasks, such as cutting and pasting between database programs. But traditional RPA systems are inflexible and unable to deal with exceptions, and can usually handle only one narrow task.” Agentic AI is meant to be both more flexible and powerful, adapting to business needs.

In a January 2025 blog post, OpenAI CEO Sam Altman said, “We believe that, in 2025, we may see the first AI agents ‘join the workforce’ and materially change the output of companies.”

But despite the hype, the overall message at the Agentic AI Summit was cautious and grounded: Agents may be the buzziest trend in AI right now, but the tech still has a long way to go, they said. AI agents, unfortunately, aren’t always reliable. They may not remember what came before.

Google DeepMind’s Chi, for example, stressed the gap between what agents can do in curated demos versus what’s still needed in real-world production environments. Pachocki highlighted concerns around the safety, security, and trustworthiness of agentic systems, particularly when they’re integrated into sensitive applications or operate autonomously.

“I still don’t think agents have really lived up to their promise,” said Sherwin Wu, head of engineering at OpenAI API. “Certain more generic cases have worked, but my day-to-day work doesn’t really feel that different with agents.”

While today’s agents may not currently live up to the massive hype (consider Salesforce CEO Marc Benioff’s recent claim that a shift to digital labor means he will be the “last CEO of Salesforce who only managed humans”), the speakers at the Agentic AI Summit still had plenty of optimism to share. Databricks’ Stoica expressed enthusiasm about infrastructure improvements that are making it easier to build agentic systems. Nvidia’s Dally suggested that continued hardware advances will enable more powerful and efficient agent behavior. Several pointed out “narrow wins” in specific domains, like coding.

Today’s AI agents may still have growing pains, but given the crowded UC Berkeley ballroom, the industry maintains its eye on the prize: AI agents that can reliably operate in the real world. The payoff, they believe, will be well worth the wait.

With that, here’s more AI news.

Sharon Goldman
sharon.goldman@fortune.com
@sharongoldman

AI IN THE NEWS

U.S. agency approves OpenAI, Google, Anthropic for federal AI vendor list. Reuters reported today that the General Services Administration, which is the U.S. government’s central purchasing arm, added OpenAI’s ChatGPT, Google’s Gemini, and Anthropic’s Claude to a list of approved AI vendors in order to accelerate use of the technology by government agencies. The tools will be available to the agencies through a platform with contract terms in place. The GSA said approved AI providers “are committed to responsible use and compliance with federal standards.”

The AI spending boom could have real consequences for the U.S. economy. According to the Washington Post, Big Tech’s record-breaking investment in artificial intelligence—more than $350 billion this year from Google, Meta, Amazon, and Microsoft—is becoming a major economic force, even as the broader U.S. economy shows signs of slowing. While job growth is cooling, this massive AI spending spree is fueling construction of data centers and driving demand for chips, servers, and networking gear—potentially boosting GDP growth by up to 0.7% in 2025. But economists warn the growing reliance on tech giants to prop up the economy is risky: if the AI boom loses steam, the economic fallout could be significant.

AI sales tool Clay raises $100 million at a $3.1 billion valuation. The New York Times Dealbook reported that Clay, which helps sales reps and marketers find new leads and turn them into customers, has raised $100 million at a $3.1 billion valuation.The round was led by CapitalG, an investment arm of Alphabet, Google’s parent company. Other participants included Meritech Capital Partners and Sequoia Capital. It comes around six months after the start-up raised money at a $1.25 billion valuation.

EYE ON AI RESEARCH

Google DeepMind’s new Genie 3 ‘world model’ creates real-time interactive simulations. Google DeepMind has unveiled Genie 3, a powerful new AI system that can generate rich, interactive virtual worlds from simple text prompts—making it possible to navigate dynamic environments in real time at 24 frames per second. But while it’s tempting to immediately leap to using the model for the ultimate gaming experience, it’s actually the latest leap in the company’s long-term push toward ‘world models’—or AI systems that can learn how the world works and simulate real-world environments. These are seen as key to training advanced agents and, eventually, achieving artificial general intelligence. Unlike prior video generators, Genie 3 allows users to move through AI-generated environments that stay visually consistent over several minutes—and even respond to commands like “make it snow” or “add a character.” For now, DeepMind is limiting access to Genie 3 to a small group of researchers and creators while it explores responsible deployment and risk.

FORTUNE ON AI

North Korean IT worker infiltrations exploded 220% over the past 12 months, with gen AI weaponized at every stage of the hiring process —by Amanda Gerut

AI is doing job interviews now—but candidates say they’d rather risk staying unemployed than talk to another robot —by Emma Burleigh

These charts show how China is pulling ahead of the U.S. in the race to power the AI future —by Matt Heimer and Nick Rapp

AI CALENDAR

Sept. 8-10: Fortune Brainstorm Tech, Park City, Utah. Apply to attend here.

Oct. 6-10: World AI Week, Amsterdam

Oct. 21-22: TedAI San Francisco. Apply to attend here.

Dec. 2-7: NeurIPS, San Diego

Dec. 8-9: Fortune Brainstorm AI San Francisco. Apply to attend here.

BRAIN FOOD

Could “depth of thought” be key to AI reasoning?

A tiny new AI model is challenging what we know about how models learn to reason: Researchers from Singapore’s Sapient Intelligence recently released the Hierarchical Reasoning Model (HRM), which draws inspiration from the brain’s layered thinking process—and the results have the AI community chattering. Despite being 100 times smaller than ChatGPT and trained on just 1,000 examples (with no internet data or step-by-step guidance), HRM solves tough logic problems like Sudoku, maze navigation, and abstract reasoning tasks that stump much larger models. Instead of mimicking human language, HRM reasons internally—quietly working through problems in hidden loops, much like a person thinking through a puzzle in their head. Its success hints at a radical shift in AI: one where depth of thought might matter more than scale.

Source link

Business

HP’s chief commercial officer predicts the future will include AI PCs that don’t use the cloud

Published

24 minutes ago

December 7, 2025

Jace Porter

Increased focus on “privacy and security” may open the door for AI-enabled devices rather than rely entirely on cloud computing and remote data centers.

“In a world where sovereign data retention matters, people want to know that if they input data to a model, the model won’t train on their data,” David McQuarrie, HP’s chief commercial officer, told Fortune in October. Using an AI locally provides that reassurance.

HP, like many of its devicemaking peers, is exploring the use of AI PCs, or devices that can use AI locally as opposed to in the cloud. “Longer term, it will be impossible not to buy an AI PC, simply because there’s so much power in them,” he said.

More broadly, smaller companies might be served just as well by a smaller model running locally than a larger model running in the cloud. “A company, a small business, or an individual has significant amounts of data that need not be put in the cloud,” he said.

Asian governments have often had stricter rules on data sovereignty. China, in particular, has significantly tightened its regulations on where Chinese user data can be stored. South Korea is another example of an Asian country that treats some locally sourced data as too sensitive to be housed overseas.

Governments the world over, and particularly in Asia, are also investing in local sovereign AI capabilities, trying to avoid relying entirely on systems and platforms housed wholly overseas. South Korea, for example, is partnering with local tech companies like search giant Naver to build its own AI systems. Singapore is investing in projects like the Southeast Asian Languages in One Network (SEA-LION), which are better tailored to Southeast Asian countries.

Asian AI adoption

Asia is HP’s smallest region, but also its fastest-growing. Revenue from Asia-Pacific and Japan grew by 7% over the company’s 2025 fiscal year, which ended in October, to hit $13.3 billion. That’s around a quarter of HP’s total revenue of $55.3 billion. (HP’s other two regions are the Americas; and Europe, the Middle East, and Africa.)

McQuarrie also suggested that there was an opportunity to be “disruptive” in Asia. While many business leaders have been eager to embrace AI, at least rhetorically, actual adoption is proving more difficult. A recent survey from McKinsey reports that two-thirds of companies are still in the experimentation phase of AI.

But McQuarrie believed that AI adoption in Asia could be “just as quick, if not quicker,” than other regions.

Asia seems to be more comfortable with the use of AI, at least when it comes to users. An October survey from Pew found that fewer people in countries like India, South Korea and Japan reported feeling “more concerned than excited” about AI compared to the U.S.

When it comes to convincing more companies to adopt AI, let alone AI PCs, McQuarrie said the answer was to make AI functions as seamless as possible, so “that it doesn’t really matter whether you understand that you’re embracing AI or not.”

“What we’re doubling down on is the future of work,” McQuarrie said. “The future of work is a device that makes your experience better and your productivity greater.”

“The fact that we’re using AI in the background? They don’t need to know that.”

Source link

Business

Trump administration waives part of a Biden-era fine against Southwest Air for canceled flights

Published

1 hour ago

December 7, 2025

Jace Porter

The U.S. Department of Transportation is waiving part of a fine assessed against Southwest Airlines after the company canceled thousands of flights during a winter storm in 2022.

Under a 2023 settlement reached by the Biden administration, Southwest agreed to a $140 million civil penalty. The government said at the time that the penalty was the largest it had ever imposed on an airline for violating consumer protection laws.

Most of the money went toward compensation for travelers. But Southwest agreed to pay $35 million to the U.S. Treasury. Southwest made a $12 million payment in 2024 and a second $12 million payment earlier this year. But the Transportation Department issued an order Friday waiving the final $11 million payment, which was due Jan. 31, 2026.

The department said Southwest should get credit for significantly improving its on-time performance and investing in network operations.

“DOT believes that this approach is in the public interest as it incentivizes airlines to invest in improving their operations and resiliency, which benefits consumers directly,” the department said in a statement. “This credit structure allows for the benefits of the airline’s investment to be realized by the public, rather than resulting in a government monetary penalty.”

The fine stemmed from a winter storm in December 2022 that paralyzed Southwest’s operations in Denver and Chicago and then snowballed when a crew-rescheduling system couldn’t keep up with the chaos. Ultimately the airline canceled 17,000 flights and stranded more than 2 million travelers.

The Biden administration determined that Southwest had violated the law by failing to help customers who were stranded in airports and hotels, leaving many of them to scramble for other flights. Many who called the airline’s overwhelmed customer service center got busy signals or were stuck on hold for hours.

Even before the settlement, the nation’s fourth-biggest airline by revenue said the meltdown cost it more than $1.1 billion in refunds and reimbursements, extra costs and lost ticket sales over several months.

Source link

Business

Trump slams Democratic congressman as disloyal for not switching parties after pardon

Published

2 hours ago

December 7, 2025

Jace Porter

Trump blasted Cuellar for “Such a lack of LOYALTY,” suggesting the Republican president might have expected the clemency to bolster the GOP’s narrow House majority heading into the 2026 midterm elections.

Cuellar, in a television interview Sunday after Trump’s social media post, said he was a conservative Democrat willing to work with the administration “to see where we can find common ground.” The congressman said he had prayed for the president and the presidency at church that morning “because if the president succeeds, the country succeeds.”

Citing a fellow Texas politician, the late President Lyndon Johnson, Cuellar said he was an American, Texan and Democrat, in that order. “I think anybody that puts party before their country is doing a disservice to their country,” he told Fox News Channel’s “Sunday Morning Futures.”

Trump noted on his Truth Social platform that the Democratic President Joe Biden’s administration had brought the charges against Cuellar and that the congressman, by running once more as a Democrat, was continuing to work with “the same RADICAL LEFT” that wanted him and his wife in prison — “And probably still do!”

“Such a lack of LOYALTY, something that Texas Voters, and Henry’s daughters, will not like. Oh’ well, next time, no more Mr. Nice guy!” Trump said. Cuellar’s two daughters, Christina and Catherine, had sent Trump a letter in November asking that he pardon their parents.

Trump explained his pardon he announced Wednesday as a matter of stopping a “weaponized” prosecution. Cuellar was an outspoken critic of Biden’s immigration policy, a position that Trump saw as a key alignment with the lawmaker.

Cuellar said he has good relationships within his party. “I think the general Democrat Caucus and I, we get along. But they know that I’m an independent voice,” he said.

A party switch would have been an unexpected bonus for Republicans after the GOP-run Legislature redrew the state’s congressional districts this year at Trump’s behest. The Texas maneuver started a mid-decade gerrymandering scramble playing out across multiple states. Trump is trying to defend Republicans’ House majority and avoid a repeat of his first term, when Democrats dominated the House midterms and used a new majority to stymie the administration, launch investigations and twice impeach Trump.

Yet Cuellar’s South Texas district, which includes parts of metro San Antonio, was not one of the Democratic districts that Republicans changed substantially, and Cuellar believes he remains well-positioned to win reelection.

Federal authorities had charged Cuellar and his wife with accepting thousands of dollars in exchange for the congressman advancing the interests of an Azerbaijan-controlled energy company and a bank in Mexico. Cuellar was accused of agreeing to influence legislation favorable to Azerbaijan and deliver a pro-Azerbaijan speech on the floor of the U.S. House.

Cuellar has said he his wife were innocent. The couple’s trial had been set to begin in April.

In the Fox interview, Cuellar insisted that federal authorities tried to entrap him with “a sting operation to try to bribe me, and that failed.”

Cuellar still faces a House Ethics Committee investigation.

Source link