Connect with us

Published

on

David Silver, leader of the reinforcement learning research group at DeepMind, being awarded an honorary “ninth dan” professional ranking for AlphaGo.
JUNG YEON-JE | AFP | Getty Images

Computer scientists are questioning whether DeepMind, the Alphabet-owned U.K. firm that’s widely regarded as one of the world’s premier AI labs, will ever be able to make machines with the kind of “general” intelligence seen in humans and animals.

In its quest for artificial general intelligence, which is sometimes called human-level AI, DeepMind is focusing a chunk of its efforts on an approach called “reinforcement learning.”

This involves programming an AI to take certain actions in order to maximize its chance of earning a reward in a certain situation. In other words, the algorithm “learns” to complete a task by seeking out these preprogrammed rewards. The technique has been successfully used to train AI models how to play (and excel at) games like Go and chess. But they remain relatively dumb, or “narrow.” DeepMind’s famous AlphaGo AI can’t draw a stickman or tell the difference between a cat and a rabbit, for example, while a seven-year-old can.

Despite this, DeepMind, which was acquired by Google in 2014 for around $600 million, believes that AI systems underpinned by reinforcement learning could theoretically grow and learn so much that they break the theoretical barrier to AGI without any new technological developments.

Researchers at the company, which has grown to around 1,000 people under Alphabet’s ownership, argued in a paper submitted to the peer-reviewed Artificial Intelligence journal last month that “Reward is enough” to reach general AI. The paper was first reported by VentureBeat last week.

In the paper, the researchers claim that if you keep “rewarding” an algorithm each time it does something you want it to, which is the essence of reinforcement learning, then it will eventually start to show signs of general intelligence.

“Reward is enough to drive behavior that exhibits abilities studied in natural and artificial intelligence, including knowledge, learning, perception, social intelligence, language, generalization and imitation,” the authors write.

“We suggest that agents that learn through trial and error experience to maximize reward could learn behavior that exhibits most if not all of these abilities, and therefore that powerful reinforcement learning agents could constitute a solution to artificial general intelligence.”

Not everyone is convinced, however.

Samim Winiger, an AI researcher in Berlin, told CNBC that DeepMind’s “reward is enough” view is a “somewhat fringe philosophical position, misleadingly presented as hard science.”

He said the path to general AI is complex and that the scientific community is aware that there are countless challenges and known unknowns that “rightfully instill a sense of humility” in most researchers in the field and prevent them from making “grandiose, totalitarian statements” such as “RL is the final answer, all you need is reward.”

DeepMind told CNBC that while reinforcement learning has been behind some of its most well-known research breakthroughs, the AI technique accounts for only a fraction of the overall research it carries out. The company said it thinks it’s important to understand things at a more fundamental level, which is why it pursues other areas such as “symbolic AI” and “population-based training.”

“In somewhat typical DeepMind fashion, they chose to make bold statements that grabs attention at all costs, over a more nuanced approach,” said Winiger. “This is more akin to politics than science.”

Stephen Merity, an independent AI researcher, told CNBC that there’s “a difference between theory and practice.” He also noted that “a stack of dynamite is likely enough to get one to the moon, but it’s not really practical.”

Ultimately, there’s no proof either way to say whether reinforcement learning will ever lead to AGI.

Rodolfo Rosini, a tech investor and entrepreneur with a focus on AI, told CNBC: “The truth is nobody knows and that DeepMind’s main product continues to be PR and not technical innovation or products.”

Entrepreneur William Tunstall-Pedoe, who sold his Siri-like app Evi to Amazon, told CNBC that even if the researchers are correct “that doesn’t mean we will get there soon, nor does it mean that there isn’t a better, faster way to get there.”

DeepMind’s “Reward is enough” paper was co-authored by DeepMind heavyweights Richard Sutton and David Silver, who met DeepMind CEO Demis Hassabis at the University of Cambridge in the 1990s.

“The key problem with the thesis put forth by ‘Reward is enough’ is not that it is wrong, but rather that it cannot be wrong, and thus fails to satisfy Karl Popper’s famous criterion that all scientific hypotheses be falsifiable,” said a senior AI researcher at a large U.S. tech firm, who wished to remain anonymous due to the sensitive nature of the discussion.

“Because Silver et al. are speaking in generalities, and the notion of reward is suitably underspecified, you can always either cherry pick cases where the hypothesis is satisfied, or the notion of reward can be shifted such that it is satisfied,” the source added.

“As such, the unfortunate verdict here is not that these prominent members of our research community have erred in any way, but rather that what is written is trivial. What is learned from this paper, in the end? In the absence of practical, actionable consequences from recognizing the unalienable truth of this hypothesis, was this paper enough?”

What is AGI?

While AGI is often referred to as the holy grail of the AI community, there’s no consensus on what AGI actually is. One definition is it’s the ability of an intelligent agent to understand or learn any intellectual task that a human being can.

But not everyone agrees with that and some question whether AGI will ever exist. Others are terrified about its potential impacts and whether AGI would build its own, even more powerful, forms of AI, or so-called superintelligences.

Ian Hogarth, an entrepreneur turned angel investor, told CNBC that he hopes reinforcement learning isn’t enough to reach AGI. “The more that existing techniques can scale up to reach AGI, the less time we have to prepare AI safety efforts and the lower the chance that things go well for our species,” he said.

Winiger argues that we’re no closer to AGI today than we were several decades ago. “The only thing that has fundamentally changed since the 1950/60s, is that science-fiction is now a valid tool for giant corporations to confuse and mislead the public, journalists and shareholders,” he said.

Fueled with hundreds of millions of dollars from Alphabet every year, DeepMind is competing with the likes of Facebook and OpenAI to hire the brightest people in the field as it looks to develop AGI. “This invention could help society find answers to some of the world’s most pressing and fundamental scientific challenges,” DeepMind writes on its website.

DeepMind COO Lila Ibrahim said on Monday that trying to “figure out how to operationalize the vision” has been the biggest challenge since she joined the company in April 2018.

Continue Reading

Technology

Joby Aviation says it is doubling production at its air taxi manufacturing hub

Published

on

By

Joby Aviation says it is doubling production at its air taxi manufacturing hub

JoeBen Bevirt, founder and CEO of Joby Aviation, stands near an electric air taxi by Joby Aviation at the Downtown Manhattan Heliport in Manhattan, New York City, U.S., November 12, 2023.

Roselle Chen | Reuters

Joby Aviation is ramping up its manufacturing capabilities in the U.S. as it races to roll out air taxi service in 2026.

The electric vertical takeoff and landing (eVTOL) maker said Tuesday that it’s launching production at its remodeled components facility in Dayton, Ohio, and plans to double capacity at its Marina, California, manufacturing hub.

“Reimagining urban mobility takes speed, scale, and precision manufacturing. Our expanded manufacturing footprint in both California and Ohio is preparing us to do just that,” said product chief Eric Allison in a release.

Shares jumped more than 7%, building on a 16% year-to-date gain.

Joby Aviation and competitors such as Archer Aviation and Eve Air Mobility are aiming to roll out eVTOLs worldwide that can ease traffic congestion in crowded city centers, but they are awaiting regulatory approval.

The company is currently in the process of gaining Federal Aviation Administration approval for its vehicles.

Read more CNBC tech news

Last month, Joby Aviation shares popped on news that it delivered its first eVTOL to the United Arab Emirates, with plans to launch service in the region next year. The company agreed to an exclusive six-year deal to roll out air taxi service in Dubai last February.

Joby said the new facilities will create hundreds of new full-time jobs and underscore its commitment to fostering American innovation. At full capacity, the 435,500-square-foot California factory will manufacture as many as 24 aircraft annually.

The electric air transport company also said the opening coincided with the flight of its sixth aircraft.

Engineers from Toyota will help ramp up aircraft production to 500 annually at the Ohio facility. The companies inked a $500 million deal last year.

Shares of Joby and its competitors have ballooned in value this year as interest in the technology gains steam.

In June, President Donald Trump signed an executive order that included the creation of an air taxi testing program.

WATCH: Joby Aviation CEO on UAE delivery: This is a huge milestone for us

Joby Aviation CEO on UAE delivery: This is a huge milestone for us

Continue Reading

Technology

Commerce Secretary Lutnick says China is only getting Nvidia’s ‘4th best’ AI chip

Published

on

By

Commerce Secretary Lutnick says China is only getting Nvidia’s ‘4th best’ AI chip

Howard Lutnick, U.S. Secretary of Commerce speaks during the Pennsylvania Energy And Innovation Summit 2025 at Carnegie Mellon University in Pittsburgh on July 15, 2025.

David A. Grogan | CNBC

Commerce Secretary Howard Lutnick on Tuesday said the Trump administration reversed course on allowing Nvidia to sell its AI chips to China because the U.S. company will not be giving over its best technology.

Lutnick made the remark speaking with CNBC’s Brian Sullivan, saying that Nvidia wants to sell China its “4th best” chip, which is slower than the fastest chips that U.S. companies use.

“We don’t sell them our best stuff, not our second best stuff, not even our third best,” Lutnick said.

Nvidia said Monday night that it would soon resume sales of the H20 chip to China after the Trump administration signaled that it would grant the chipmaker necessary export licenses.

Lutnick said that the administration said that the renewed sale of H20 chips to China was linked to a rare-earths magnet deal. Lutnick said it was in U.S. interests to have Chinese companies using American technology so they continue to use an American “tech stack.”

“The fourth one down, we want to keep China using it,” Lutnick said. “We want to keep having the Chinese use the American technology stack, because they still rely upon it.”

Similarly, Nvidia CEO Jensen Huang has said in recent weeks that the U.S. should continue selling his chips to China so Chinese companies don’t invest in homegrown infrastructure. Huang on Sunday also said that the Chinese military wouldn’t use Nvidia chips anyway, and previously signaled that China’s Huawei is a legitimate competitor.

“The idea is the Chinese are more than capable of building their own,” Lutnick said. “You want to keep one step ahead of what they can build, so they keep buying our chips.”

The reversal is a major win for Nvidia. Huang had previously said that the Trump administration’s decision to require a license for the H20 chip in April “effectively closed” the China market. Nvidia said that it could have sold $8 billion in H20 chips in the current quarter before sales were stopped.

Commerce Sec. Howard Lutnick on Indonesia trade deal: No tariffs there, they pay tariffs here

The administration reversed its decision after President Donald Trump met with Huang in Washington last week.

“You want to sell the Chinese enough that their developers get addicted to the American technology stack,” Lutnick said. “That’s the thinking.”

The H20 chip was introduced in 2022 in response to Biden administration export controls. It’s based on the same underlying technology as Nvidia’s Hopper-generation chips, which are sold in the U.S. as finished systems using H100 or H200 chips.

The U.S. chipmaker took some features out of the H20 in order to sell it to China, including fewer graphics processing unit cores and lower bandwidth connecting separate parts of the chip. But the success of the DeepSeek R1 model suggested that there were many Chinese companies that were just fine with the slowed-down chips. The China-specific H20 is behind Nvidia’s Blackwell chips, the H100 and the H200, Lutnick said.

Nvidia says that it releases new artificial intelligence chips every year and that serious AI developers should always try to get the latest and greatest versions because the technology is improving so quickly.

The best AI chips broadly available from clouds and system makers today are called Blackwell, and come as a GB200 chip with a paired central processing unit as well as B100 and B200 versions. Nvidia also makes a range of Blackwell-based chips for gaming and graphics that can be used for AI, but they’re generally weaker than the biggest chips designed for data centers.

A successor, called Blackwell Ultra, is only now starting to be installed in data centers, and it’s expected to ramp in volume over the next year. In 2027, Nvidia will release “Vera Rubin” chips.

Continue Reading

Technology

It’s a huge week for crypto in D.C. But the industry may not get everything it wants

Published

on

By

It's a huge week for crypto in D.C. But the industry may not get everything it wants

The U.S. Capitol building in Washington, D.C., U.S., June 27, 2025.

Elizabeth Frantz | Reuters

It’s “Crypto Week” in Washington.

The cryptocurrency industry is set to notch a major win this week if the House can pass two bills that would set up a long-lobbied-for regulatory framework for digital assets.

The stablecoin bill, known as the GENUIS Act, has already passed the Senate and looks set to become the first standalone crypto measure signed into law should the House do the same.

But the real prize for the industry is a wider and more complex bill on market structure called the CLARITY Act, which faces a more difficult path to President Donald Trump‘s desk.

Seeking CLARITY

The CLARITY Act sets the rules for when an asset is considered a security and overseen by the Securities and Exchange Commission versus when it’s considered a commodity that is overseen by the Commodity Futures Trading Commission, or CFTC.

The act is likely to pass the House on Wednesday, given the bipartisan support when the bill cleared two committees. But the path in the Senate is murky, as Democrats could withhold their support over concerns about how Trump and his family are benefiting from crypto.

The Trump family’s growing crypto empire includes $TRUMP and $MELANIA meme coins, a stablecoin, and a decentralized finance firm called World Liberty Financial, among other ventures.

Some lawmakers who backed the narrower stablecoin bill did so with the hopes of seeing the wider market structure package address conflicts of interest.

“President Trump’s crypto corruption distorts the digital asset marketplace,” said Sen. Raphael Warnock, D-Ga., who voted for the stablecoin bill. “Writing a bill with a corruption caveat for the president sends a clear message — that Congress is not serious about addressing corruption, which we know undermines investors’ faith in capital markets.”

Pushing it to pass

Coinbase attempted to literally sweeten the deal on the CLARITY Act for lawmakers with an advertising push that included handing out about 5,000 chocolate bars around D.C.

The candy wrappers cited a Morning Consult poll that found about “1 in 5” Americans own crypto.

Coinbase, Ripple and other crypto companies are lobbying Congress to put their concerns aside and back the market structure package, anticipating that more regulatory certainty will encourage more investment in crypto.

“When consumers buy and sell and trade these digital assets, they want to know what they’re getting and they want to know that they’re using a reputable intermediary,” Coinbase Vice President of U.S. Policy Kara Calvert told CNBC. “And what this bill does is provide that construct to do that.”

Read more CNBC tech news

The Senate is set to introduce its own market structure bill this month that is expected to differ slightly from the House version.

Senate Banking Chair Tim Scott, R-S.C., is working with Sen. Cynthia Lummis, R-Wyo., and others on the measure.

Other Democrats are planning to work with Republicans on a bill, including Sen. Kirsten Gillibrand, D-N.Y., who worked on previous market structure bills with Lummis.

“We have a lot of work to do, and we’re going to work on a bipartisan basis over the next month,” she told CNBC in a brief interview in the Capitol.

GENIUS and the Fed

Continue Reading

Trending