Connect with us

Published

2 years ago

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.

Related Topics:
Continue Reading

Technology

Elon Musk says he won’t vote for Biden over Trump, calls Haley ‘pro-censorship’

Published

7 hours ago

on

November 29, 2023

By

Elon Musk says he won't vote for Biden over Trump, calls Haley 'pro-censorship'

Elon Musk, chief executive officer of Tesla Inc., during a fireside discussion on artificial intelligence risks with Rishi Sunak, UK prime minister, not pictured, in London, UK, on Thursday, Nov. 2, 2023. Sunak convened this week’s AI summit in an effort to position the UK at the forefront of global efforts to stave off the risks presented by the rapidly-advancing technology, which in the prime minister’s own words, could extend as far as human extinction. Photographer: Tolga Akmen/EPA/Bloomberg via Getty Images

Bloomberg | Bloomberg | Getty Images

Elon Musk said Wednesday that he won’t vote for President Joe Biden in the 2024 election, even if former president Donald Trump is the Republican nominee.

“I would not vote for Biden,” Musk said during a wide-ranging interview with Andrew Ross Sorkin at the DealBook Summit in New York. “I’m not saying I’d vote for Trump.”

When asked what he’d do if those were the two nominees, Musk said, “This is definitely a difficult choice here.”

Musk, who says he supported Barack Obama’s candidacy, has moved rightward in his politics in recent years, writing in a tweet last year that “today’s Democratic Party has been hijacked by extremists.”

While he hasn’t endorsed a specific candidate for the 2024 election, Musk said last year that Florida Governor Ron DeSantis was his preferred choice at the time. He also hosted DeSantis’s campaign launch on X, formerly Twitter. More recently, Musk has said that Vivek Ramaswamy is “looking like a strong candidate.”

Musk told Sorkin on Wednesday that he disagrees with Ramaswamy on climate issues, but he shares some of the candidate’s views on government overreach and censorship. DeSantis’s name did not come up in the interview.

When asked if he could support Nikki Haley among the Republicans, Musk said no and described the former South Carolina governor as a “pro-censorship candidate.”

In terms of which party is more favorable towards freedom of speech, Musk said that “on balance, the Democrats appear to be more pro-censorship than Republicans,” which he characterized as a change from the past.

“We certainly get more complaints from the left than the right,” Musk said.

WATCH: Elon Musk shoots himself in the foot’ by reposting ‘wacky, fringe theories.’

Elon Musk 'shoots himself in the foot' by reposting 'wacky, fringe theories', says Walter Isaacson

Continue Reading

Technology

Elon Musk claims advertisers are trying to ‘blackmail’ him, says ‘Go f— yourself’

Published

9 hours ago

on

November 29, 2023

By

Elon Musk claims advertisers are trying to 'blackmail' him, says 'Go f--- yourself'

Tesla and SpaceX’s CEO Elon Musk reacts during an in-conversation event with British Prime Minister Rishi Sunak in London, Britain, Thursday, Nov. 2, 2023. 

Kirsty Wigglesworth | Reuters

Speaking at the 2023 DealBook Summit in New York on Wednesday, Elon Musk, the owner of social media site X (formerly Twitter), scoffed at advertisers threatening to leave the platform because of antisemitic posts he amplified there.

“If somebody’s gonna try to blackmail me with advertising? Blackmail me with money? Go f—yourself.” He added, “Don’t advertise.”

He also implied that fans of his, and of X, would boycott those advertisers in kind. He specifically took aim at Disney.

“The whole world will know that those advertisers killed the company and we will document it in great detail,” Musk threatened.

He also told interviewer Andrew Ross Sorkin, “I have no problem being hated. Hate away.”

In recent weeks, Musk has promoted and sometimes verbally endorsed what the White House called “antisemitic and racist hate” on X, formerly Twitter, the social media platform he owns and runs as CTO.

He called those tweets, “one of the most foolish if not the most foolish thing I’ve ever done on the platform.”

“I’m sorry for that tweet or post,” he said. He added, “I tried my best to clarify, six ways to Sunday, but you know at least I think over time it will be obvious that in fact, far from being antisemitic, I am in fact philosemitic.”

His inflammatory posts on the social media platform led large advertisers, including Disney, Apple, and many others, to suspend campaigns there, and drove some famous users away from the platform, including Paris Mayor Anne Hidalgo.

Musk, who is also the CEO of Tesla and SpaceX, has denied that he is antisemitic, and said that on X, “Clear calls for extreme violence are against our terms of service and will result in suspension.”

He also traveled to Israel this week, where he met and spoke with Prime Minister Benjamin Netanyahu. When Netanyahu said he wanted to “deradicalize” and “rebuild” Gaza, Musk offered to help. Musk told Sorkin on stage that his visit to Israel was planned before his tweets, and were not part of an “apology tour.” Previously, Musk had said he wanted to bring SpaceX satellite communications service to Israel and humanitarian organizations in Gaza.

Musk’s personal account on X currently displays a follower count of more than 164 million — though tech blog Mashable reported in August that a majority of Musk’s listed followers appeared to be inauthentic or inactive accounts.

Unions, China, and OpenAI

Earlier on Wednesday, the UAW launched campaigns aimed at Tesla and 12 other automakers in the U.S. Sorkin asked Musk what that means for his EV business.

Musk espoused negative general views about unions and said they create a “lords and peasants” atmosphere at companies, and “naturally try to create negativity,” pitting workers against management.

He said, “Many people at Tesla have come up, gone from workign on the line to being in senior management and there is no lords and peasants — everyone eats at the same table.”

He also added, “If Tesla gets unionized, it will be because we deserve it and we failed in some way.”

At one point, Sorkin asked, “Do you feel like anybody has leverage over you?”

Musk replied, “If we make bad products that people don’t want to use, the users will vote with their resources and use something else. My companies are overseen by regulators. SpaceX, Starlink, Tesla – are overseen by cumulatively by…a few hundred regulators because we’re in 55 countries.”

Later, he noted that he complies with nearly all the regulations levied upon his companies, but “once in awhile” he disagrees with a regulation and would object to it and disobey. “I’m incredibly rule-following,” he claimed.

Sorkin asked, “How do you think about the leverage that the Chinese have over you?” alluding to Tesla’s factory there and the company’s reliance on Chinese consumers for a percentage of its sales. Sorkin added, “Is it hypocritical for you to be doing business in China, or other countries, as it relates to X and other things that don’t follow this free speech path that you have espoused?”

The CEO replied, “The best that the platform can do is adhere to the laws of any given country. Do you think there’s something more we can do than that?”

He later added that he believes the Chinese electric car companies are extremely competitive, and said that many people believe the top ten EV companies in the world will be Tesla and nine Chinese makers.

On OpenAI and its recent boardroom struggles, Musk said he had talked to a lot of people but had not found out what precisely led to the recent firing and then re-hiring of CEO Sam Altman. He also said he has “mixed feelings” about Altman personally, hinting that he feels like the OpenAI CEO has too much power. “The ring of power can corrupt.”

When it was founded, OpenAI’s original board included both Altman and Musk, but Musk left in 2018 after poaching a star engineer from the company to run Autopilot software engineering at Tesla.

Musk also said that he’s worried about the danger of AI harming humanity, and that he was “having trouble sleeping at night” because of it.

This is breaking news. Please check back for updates.

Continue Reading

Technology

FTC Chair Lina Khan defends her track record when it comes to blocking mergers and doesn’t subscribe to Amazon Prime

Published

9 hours ago

on

November 29, 2023

By

FTC Chair Lina Khan defends her track record when it comes to blocking mergers and doesn't subscribe to Amazon Prime

Federal Trade Commission Chair Lina Khan speaks during The New York Times annual DealBook Summit in New York City on Nov. 29, 2023.

Michael M. Santiago | Getty Images

Federal Trade Commission Chair Lina Khan defended her track record in court when it comes to blocking mergers, saying she believes the agency should take big swings and she’s “quite pleased” with the work it has done so far under her tenure, which started in June 2021.

Speaking at The New York Times DealBook Summit on Wednesday, Khan said that whenever the FTC brings a case, “you want to win it,” but that whenever there’s a loss, the agency will “try to figure out what went wrong.”

The FTC has had some high-profile losses during Khan’s tenure, including a failed attempt to block Facebook parent Meta from buying virtual reality company Within Unlimited. It also lost a fight to stop Microsoft‘s $69 billion acquisition of gaming giant Activision Blizzard, though the agency is still appealing the court ruling.

Under her leadership, Khan said the FTC has brought 11 cases against mergers, and in five instances, the companies abandoned their plans after the agency filed suit. There were 14 deals that were dropped during the FTC’s investigation, she added.

“Big picture, of course the two cases that we lost we would’ve wanted to win, but we’re quite pleased overall with our efforts,” Khan said on stage.

Khan is in the middle of what could be a career-defining antitrust case. In September, the FTC and 17 states sued Amazon, accusing the retail giant of wielding its “monopoly power” to artificially rise prices, degrade quality for shoppers and stifle competition. The lawsuit was long anticipated, as Khan rose to prominence for her 2017 Yale Law Journal article, “Amazon’s Antitrust Paradox.”

Khan argued in the piece that the prevailing antitrust framework at the time, which focused primarily on monopolies’ harm to consumers, failed to capture the ways tech giants such as Amazon are able to dominate in the digital world even while offering lower prices and more selection to consumers.

The agency has also taken aim at Amazon’s Prime service, alleging it tricked users into signing up for the program and intentionally complicated the cancellation process. Amazon has disputed both of the FTC’s lawsuits, calling them “wrong on the facts and the law.”

In the interview Wednesday, Khan said she doesn’t subscribe to Prime, which costs $139 a year and includes perks such as free shipping, access to streaming content and discounts on Whole Foods groceries.

Asked why she hasn’t subscribed to Prime, Khan replied, “I just haven’t.”

Don’t miss these stories from CNBC PRO:

Continue Reading

Trending