Connect with us

Published

on

Sundar Pichai, chief executive officer of Alphabet Inc., during the Google I/O Developers Conference in Mountain View, California, on Wednesday, May 10, 2023.

David Paul Morris | Bloomberg | Getty Images

Google’s new large language model, which the company announced last week, uses almost five times as much training data as its predecessor from 2022, allowing its to perform more advanced coding, math and creative writing tasks, CNBC has learned.

PaLM 2, the company’s new general-use large language model (LLM) that was unveiled at Google I/O, is trained on 3.6 trillion tokens, according to internal documentation viewed by CNBC. Tokens, which are strings of words, are an important building block for training LLMs, because they teach the model to predict the next word that will appear in a sequence.

Google’s previous version of PaLM, which stands for Pathways Language Model, was released in 2022 and trained on 780 billion tokens.

While Google has been eager to showcase the power of its artificial intelligence technology and how it can be embedded into search, emails, word processing and spreadsheets, the company has been unwilling to publish the size or other details of its training data. OpenAI, the Microsoft-backed creator of ChatGPT, has also kept secret the specifics of its latest LLM called GPT-4.

The reason for the lack of disclosure, the companies say, is the competitive nature of the business. Google and OpenAI are rushing to attract users who may want to search for information using conversational chatbots rather than traditional search engines.

But as the AI arms race heats up, the research community is demanding greater transparency.

Since unveiling PaLM 2, Google has said the new model is smaller than prior LLMs, which is significant because it means the company’s technology is becoming more efficient while accomplishing more sophisticated tasks. PaLM 2, according to internal documents, is trained on 340 billion parameters, an indication of the complexity of the model. The initial PaLM was trained on 540 billion parameters.

Google didn’t immediately provide a comment for this story.

A.I. takes center stage at Alphabet's annual Google I/O conference

Google said in a blog post about PaLM 2 that the model uses a “new technique” called “compute-optimal scaling.” That makes the the LLM “more efficient with overall better performance, including faster inference, fewer parameters to serve, and a lower serving cost.”

In announcing PaLM 2, Google confirmed CNBC’s previous reporting that the model is trained on 100 languages and performs a broad range of tasks. It’s already being used to power 25 features and products, including the company’s experimental chatbot Bard. It’s available in four sizes, from smallest to largest: Gecko, Otter, Bison and Unicorn. 

PaLM 2 is more powerful than any existing model, based on public disclosures. Facebook’s LLM called LLaMA, which it announced in February, is trained on 1.4 trillion tokens. The last time OpenAI shared ChatGPT’s training size was with GPT-3, when the company said it was trained on 300 billion tokens at the time. OpenAI released GPT-4 in March, and said it exhibits “human-level performance” on many professional tests.

LaMDA, a conversation LLM that Google introduced two years ago and touted in February alongside Bard, was trained on 1.5 trillion tokens, according to the latest documents viewed by CNBC.

As new AI applications quickly hit the mainstream, controversies surrounding the underlying technology are getting more spirited.

El Mahdi El Mhamdi, a senior Google Research scientist, resigned in February over the company’s lack of transparency. On Tuesday, OpenAI CEO Sam Altman testified at a hearing of the Senate Judiciary subcommittee on privacy and technology, and agreed with lawmakers that a new system to deal with AI is needed.

“For a very new technology we need a new framework,” Altman said. “Certainly companies like ours bear a lot of responsibility for the tools that we put out in the world.”

— CNBC’s Jordan Novet contributed to this report.

WATCH: OpenAI CEO Sam Altman calls for A.I. oversight

OpenAI CEO Sam Altman call fors A.I. oversight in testimony to congress

Continue Reading

Technology

Perplexity AI wrapping talks to raise $500 million at $14 billion valuation

Published

on

By

Perplexity AI wrapping talks to raise 0 million at  billion valuation

Dado Ruvic | Reuters

Perplexity AI is in late-stage talks to raise $500 million at a $14 billion valuation, a source familiar with the situation confirmed to CNBC Monday.

Accel, the Palo Alto-based venture capital firm, will lead the round, according to the source, who spoke anonymously because the round is not yet finalized. The Wall Street Journal first reported on the late-stage numbers.

The funding is on the lower end of Perplexity’s planned raise, which CNBC reported in March. During those early-stage talks, Perplexity was looking to raise between $500 million and $1 billion in funding at an $18 billion post-money valuation, per a source familiar.

The artificial intelligence search engine company competes against the likes of Google and Microsoft-backed OpenAI. Its valuation in December was $9 billion, triple its $3 billion valuation in June 2024.

Read more CNBC reporting on AI

Perplexity has just under $100 million in annual recurring revenue, or ARR, the source told CNBC in March.

Perplexity has been in the middle of the generative AI boom that began in late 2022 with the launch of OpenAI’s ChatGPT, and it’s betting big on its upcoming AI agent web browser, called Comet. But Perplexity faces increasing competition in the AI search market.

In March, Anthropic launched its web search product, allowing its chatbot Claude to display real-time search results to a subset of users.

Last fall, OpenAI launched a search feature within ChatGPT, its viral chatbot, that positioned it to better compete with Perplexity, as well as leading search engines such as Google and Microsoft‘s Bing.

Google has released AI Overviews within its search product as well, though it sparked controversy over high-profile errors soon after its release.

Continue Reading

Technology

Trump says he talked to Apple CEO Tim Cook after China tariff rollback

Published

on

By

Trump says he talked to Apple CEO Tim Cook after China tariff rollback

Apple CEO Tim Cook, center, watches during the inauguration ceremonies for President Donald Trump, right, and Vice President JD Vance, left, in the rotunda of the U.S. Capitol in Washington, Jan. 20, 2025.

Shawn Thew | Afp | Getty Images

President Donald Trump said Monday that he talked to Apple CEO Tim Cook after the U.S. and China agreed to suspend most tariffs for 90 days.

Wall Street and Apple investors cheered the pause on Chinese tariffs. Apple stock was up 6% in trading on Monday, versus 3% for the Nasdaq.

“I spoke to Tim Cook this morning, and he’s going to, I think, even up his numbers,” Trump said in the Oval Office. “$500 billion, he’s going to be building a lot of plants in the United States for Apple. And we look forward to that.”

Apple previously said in February it would spend $500 billion to expand many of its operations in the U.S., including assembling AI servers in Houston.

Any cooling of a U.S.-China trade war is expected to boost Apple, which does the majority of its device production in the country, and also counts the region as its third-largest by sales.

Read more CNBC tech news

Still, it’s not clear how much Monday’s announcement immediately helped Apple.

In April, most of Apple’s most important products, such as smartphones and computers, received exemptions on some of the highest 145% tariffs, but there are still 30% tariffs on Chinese imports even after Sunday’s deal. Apple still faces 10% tariffs in some of its secondary production locations, such as India and Vietnam.

The Trump administration wants Apple to bring device production, including iPhone manufacturing, to the United States, a move that many experts believe would be unlikely and expensive.

Earlier this month, Cook told investors about the company’s tariff strategy on an earnings call. He said that Apple is currently sourcing American-bound products from production locations in Vietnam and India, but didn’t want to speculate beyond June, calling the situation “difficult to predict.”

An Apple spokesperson declined to comment.

Continue Reading

Technology

U.S.-China breakthrough send tech and chip stocks soaring

Published

on

By

U.S.-China breakthrough send tech and chip stocks soaring

HANGZHOU, CHINA – JUNE 3, 2024 – The NVIDIA logo and the Apple logo are pictured in Hangzhou city, Zhejiang province, China, June 6, 2024. On June 5, Eastern time, Nvidia’s stock market value exceeded $3 trillion, officially surpassing Apple’s market value and becoming the world’s second largest technology giant by market value. It is worth noting that in just over 3 months, Nvidia’s market value soared from $2 trillion to $3 trillion. (Photo credit should read CFOTO/Future Publishing via Getty Images)

Cfoto | Future Publishing | Getty Images

Global technology and chip stocks rallied on Monday after the U.S. and China agreed to pause most tariffs on each other’s goods.

Technology stocks — such as semiconductor firms and smartphone makers — have been hit hard as trade tensions between the world’s two largest economies threatened to disrupt supply chains and hurt some of the biggest U.S. businesses.

But investors breathed a sigh of relief after talks between the U.S. and China over the weekend yielded a temporary pause in “reciprocal” tariffs.

In the U.S., Nvidia, which still faces a number of restrictions on the chips it is allowed to ship to China, was around 4% higher in premarket trade, while AMD was up 5%. Broadcom was also around 5% higher, along with Qualcomm.

Other companies in the semiconductor supply chain also jumped. Marvell, which last week postponed a previously scheduled investor day due to macroeconomic uncertainty, surged 7.5% in premarket trade.

Taiwan Semiconductor Manufacturing Co., the world’s largest chipmaker, saw its U.S.-listed shares jump around 4% in the premarket. TSMC’s Taiwan-listed stock closed before the tariff announcement.

In Europe, ASML, a supplier of critical machinery required to manufacture the most advanced chips, rallied 4.5% in early trade. Infineon was also sharply higher.

Semiconductors and some electronics received an exemption from President Donald Trump’s reciprocal tariffs last month, but the U.S. signaled the reprieve was temporary and that these products could still be in line for special duties.

Investors have been concerned about the impact on major tech stocks, especially those with exposure to China such as Apple and Amazon, whose shares have been under pressure this year.

Apple, which still makes 90% of its iPhones in China, said during its earnings report this month that it expects tariffs will add $900 million to its costs for the current quarter. Apple shares were more than 7% higher.

Amazon was up more than 8% in premarket trade Monday. Many sellers on Amazon rely on Chinese products.

U.S.-listed Chinese tech stocks also surged. Chinese e-commerce giants Alibaba and JD.com were higher, alongside internet firm Baidu.

“With US/China clearly on an accelerated path for a broader deal we believe new highs for the market and tech stocks are now on the table in 2025 as investors will likely focus on the next steps in these trade discussions which will happen over the coming months,” Daniel Ives, global head of technology research at Wedbush Securities, said in a note on Monday.

“This morning is a huge win for the bulls and a best case scenario post this weekend in our view.”

Continue Reading

Trending