technology – Digitex Solutions

Google adds limits to ‘Work from Anywhere’ policy that began in Covid

digitex — Thu, 09 Oct 2025 09:36:00 +0000

Sundar Pichai, chief executive officer of Alphabet Inc., during the Bloomberg Tech conference in San Francisco, California, US, on Wednesday, June 4, 2025. David Paul Morris | Bloomberg | Getty ImagesGoogle is continuing to put restrictions on remote work, this time with a popular policy called “Work from Anywhere” that was established during the Covid pandemic.The policy has allowed employees to work from a location outside of their main office for up to four weeks per calendar year. According to internal documents viewed by CNBC, working remotely for even a single day will now count for a full week.”Whether you log 1 WFA day or 5 WFA days in a given standard work week, 1 WFA week will be deducted from your WFA weekly balance,” according to a document that was circulated over the summer, shortly before the change went into effect. Google isn’t altering its current hybrid schedule, which was also put in place during the pandemic, allowing employees to work from home two days a week. WFA days are distinct from that policy, giving staffers the flexibility to work remotely, but not at home.”WFA weeks cannot be used to work from home or nearby,” the document says.Google didn’t immediately respond to request for comment.Tech companies are increasingly forcing employees to spend more time in the office, with the peak of Covid now about five years in the past. Microsoft said last month that employees will be expected to work in an office three days a week starting next year, switching from a policy that allowed most of them to work from home 50% of the time or more with manager approval. Amazon went further, instructing corporate staffers to spend five days a week in the office.Google began offering some U.S. full-time employees voluntary buyouts at the beginning of 2025, and has notified remote workers from several units their jobs would be considered for layoffs if they didn’t return to offices to work a hybrid schedule.According to the latest changes, employees can’t work from a Google office in a separate state or country during their WFA time due to “legal and financial implications of cross border work.” If in a different location, employees may be required to work during the business hours that align with that time zone, the rules state.The WFA update doesn’t apply to all Google staffers and may exclude data center workers, and those who are required to be in physical offices. Violations of the policy will result in disciplinary action or termination, the document says.The issue came up at a recent all-hands meeting.A top-rated question that was submitted on Google’s internal system described the update as “confusing.””Why does even one day of WFA count as a whole week, and can we reconsider the restriction on using WFA weeks to work from home?” the question said.John Casey, Google’s vice president of performance and rewards, said at the meeting that WFA “was meant to meet Googlers where they were during the pandemic,” according to audio obtained by CNBC.”The policy was always intended to be taken in increments of a week and not be used as a substitute for working from home in a regular hybrid work week,” Casey said.WATCH: Google adds Gemini to Chrome for all users in push to bolster AI search

Source link

Google’s AI Overviews Are Quietly Rewiring How The World Searches

digitex — Fri, 25 Apr 2025 12:00:26 +0000

Google is riding a powerful AI wave. On its Q1 2025 earnings call, Alphabet and Google CEO Sundar Pichai revealed that AI-powered Search is now being used by over 1.5 billion people every month, marking one of the most significant milestones in the company’s recent history. This explosive adoption of AI Overviews underscores how deeply artificial intelligence is transforming Google’s core business.
Pichai noted that Search, Cloud, and YouTube were the top-performing verticals this quarter. “We’re pleased with our strong results this quarter,” he said, adding that AI was powering many of the company’s latest features across products.
The standout announcement? AI Overviews – the generative, smart summaries in Google Search – are now reaching more than 1.5 billion monthly users, with usage continuing to grow as people discover the tool’s usefulness for a wider range of queries. “Nearly a year after we launched AI Overviews in the US, we continue to see that usage growth is increasing,” said Pichai.
Gemini 2.5 and AI Stack: The New Backbone of Google Products
Pichai also emphasised the company’s comprehensive AI stack – from infrastructure to research to product integration – as a key differentiator. He highlighted the rollout of Gemini 2.5, Google’s most powerful AI model yet, calling it a breakthrough in reasoning, science, and math. “It’s widely recognised as the best model in the industry,” he stated.
The new Ironwood TPU, Google’s most powerful processor to date, brings over 10x performance improvements and double the power efficiency. Google’s strategic partnership with NVIDIA continues to pay off, with early access to the latest generation of GPUs, including the groundbreaking B200 and GB200 Blackwell chips.
“All 15 of our products with a half billion users now use Gemini models,” Pichai said, showcasing how deeply embedded AI has become in the company’s ecosystem—from Android and Pixel to Google Assistant.
Cloud and YouTube Continue to Grow, Waymo Expands Reach
On the cloud front, Google highlighted strong customer adoption of its Vertex AI Platform, which now supports over 200 models. Enterprise partners like Lowe’s, KPMG, and Verizon are tapping into Google’s AI agents and tools to automate workflows and drive efficiencies. Google Workspace, too, is delivering over 2 billion AI assists every month.
On YouTube’s 20th anniversary, Pichai noted major growth in subscriptions and TV viewership. “TV is the primary device for YouTube viewing in the US,” he said, pointing out that YouTube has been number one in US streaming watch time for two straight years. The platform now boasts 125 million Premium and Music subscribers and over 1 billion monthly podcast users.
Meanwhile, Waymo, Google’s self-driving unit, is logging over 250,000 paid passenger trips each week, up five-fold from last year. The autonomous service is expanding across cities, including a future launch in Washington, DC, in 2026.
Looking Ahead: More AI Innovations Coming
Pichai closed his remarks with excitement for upcoming showcases like Google I/O, Brandcast, and Google Marketing Live. “This quarter was super exciting,” he said, promising more innovations ahead. “And we are focused on bringing this to people and customers everywhere.”
With AI now embedded across its products and services, Google appears set on redefining its future—and ours—through machine learning, automation, and intelligent assistance at scale.

Source link

How to Keep Your Brand Alive in the Age of ChatGPT and AI Search

digitex — Thu, 24 Apr 2025 19:57:43 +0000

Opinions expressed by Entrepreneur contributors are their own.
The digital landscape is undeniably shifting, with platforms like ChatGPT attracting over 200 million weekly active users globally. This rise of LLMs is seriously changing how consumers discover and interact with brands.While traditional search certainly has its strengths and will remain a significant channel, exploring these new avenues is crucial for businesses to broaden their reach, maintain a competitive edge and connect with customers in this evolving digital world.In this article, I will delve into the pillars of staying competitive in the age of LLMs.

1. Optimize for conversational searchLLMs prioritize natural, question-based language over short, keyword-heavy queries. Speaking business, this shift calls for long-tail keywords, clear, concise answers that AI models can easily extract and structured data to help AI interpret your site. Think about how your customers actually ask questions and make content that directly answers those inquiries – thus, you’re more likely to be favored both by clients and AI.Furthermore, long-tail keywords typically face less competition, making it easier for the average website to achieve visibility. Google itself reports that 15% of all Google searches are entirely new, highlighting the demand for fresh, niche content optimized for emerging search trends.

Centime, a financial software company, optimized its content for voice search to align with how people naturally ask questions. Their “Ultimate Guide to Cash Flow Forecasting” was rewritten to directly answer conversational queries. This move proved to work well: traffic, particularly from voice searches, rose 40%, improving both SEO performance and user retention.Related: AI Is Most Likely to Replace These 3 Professions: AI Experts2. Stay relevantWhen speaking of consumer engagement in the age of LLMs, relevance is everything. Generic, mass-produced content won’t cut through the AI-driven search engines. That’s when useful, clear and deeply aligned with your audience’s needs, content becomes crucial. We speak a lot about the rising demand for personalization. The same story is happening to tailored answers when it comes to AI-generated results.Caleres, parent company of fashion brands like Sam Edelman and Veronica Beard, relaunched 13 e-commerce sites using AI-powered product search and discovery by Coveo. Such an AI-trust yielded a 23% increase in conversion rates and a 5.5% boost in revenue per visitor.

3. Build strong brand authorityIn the world of LLMs, brand authority is defined by your presence across high-quality, editorially overseen platforms. AI search engines also factor in online mentions, citations and overall brand authority when generating responses, so focus on strategic partnerships and collaborations to strengthen your backlink profile.Engage actively on social platforms like LinkedIn and Instagram to prove credibility. Invest actively in content marketing to become the go-to resource for valuable information within your industry.One off-trail solution is building a Wikipedia page. As one of the most cited sources in AI training data, Wikipedia plays a huge role in verifying your brand’s authority. A properly sourced page improves your chances of being referenced in AI-generated answers.A recent study tracking major toy brands like Fisher-Price and Melissa & Doug found that mentions on trusted media like Tinybeans significantly boosted their visibility in AI search engines. This highlights the importance of strategic digital PR and building relationships with influential publications to improve AI search rankings.

Related: AI Won’t Replace You — But an AI-Equipped Professional Will4. Keep monitoring and improvingLLMs evolve fast, just like user behavior. ChatGPT alone processes over a billion queries a day. To stay competitive, brands must track performance, keep abreast of new developments in AI search algorithms and pivot quickly, considering a multi-faceted approach combining native platform tools with specialized third-party solutions.Google Search Console and Bing Webmaster can provide insights into how your brand appears in AI-driven search features. Third-party monitoring tools like BrightEdge, Semrush and Brandwatch offer features for AI content optimization, SEO insights and performance tracking across AI-powered tools.Regularly run test customer-centric queries. Analyze how competitors are being mentioned in AI results and identify opportunities for your business. Manual checks provide the human insight that automated tools often miss.

The rise of AI search doesn’t replace traditional SEO but expands it. To stay competitive, brands must build authority and adapt across all search touchpoints, from LLMs to Google to social platforms.

Source link

Beijing embraces DeepSeek to lead AI adoption as it looks for new growth drivers

digitex — Fri, 21 Feb 2025 10:50:32 +0000

HONG KONG, CHINA – JANUARY 28: In this photo illustration, the DeepSeek apps is seen on a phone in front of a flag of China on January 28, 2025 in Hong Kong, China. Anthony Kwan | Getty Images News | Getty ImagesDeepSeek’s sudden splash in the large language model space has given China a powerful tool to catalyze artificial-intelligence adoption in the country and boost economic growth.While Goldman Sachs pegs a 20-basis-point to 30-basis-point boost to China’s GDP over the long term — by 2030 — its expects the country’s economy to start reflecting the positive impact of AI adoption from next year itself as AI-driven automation improves productivity.”The recent emergence of DeepSeek … suggests faster AI development and adoption in China than we previously anticipated,” economists at the Wall Street bank said.The enthusiasm around DeepSeek is also being reflected in the sharp rally in China stocks, with the MSCI China index soaring over 21% from its January low, according to LSEG data.The startup’s rise is triggering a reassessment of China’s “investability” after an extended period of limited attention, Morgan Stanley said in a note this week.”DeepSeek demonstrates that China is at or near the cutting edge of AI development, which boosts the prestige of China’s economy and tech ecosystem, making them more attractive for global investors,” said Gabriel Wildau, managing director at Teneo.The company’s launch of a cheaper and more efficient AI model came as a timely confidence boost as the Chinese leadership faces a prolonged economic gloom, partly owed to the slump in its property market, while the specter of a fierce trade war with the U.S. looms large.DeepSeek’s R-1 reasoning model has been lauded as being able to match, or even outperform, leading global AI offerings amid claims of running on cheaper and less sophisticated chips. The open-source model also can be repurposed by developers outside the company to significantly boost efficiency at a lower operating costs.The startup has shaken China’s AI ecosystem as well, with state-owned entities as well as large tech players, including competitors, leveraging its open-sourced architecture.”The scale and speed of [AI] adoption [in China] is amazingly fast right now, and it’s not slowing down,” said Wei Sun, principal analyst of artificial intelligence at Counterpoint Research.Beijing’s stamp of approval In a well-choreographed meeting earlier this week, Chinese President Xi Jinping warmly greeted DeepSeek founder Liang Wenfeng and granted him a coveted front-row seat next to leaders of the country’s biggest private enterprises.That showed Beijing is eager to support the company, said Huiyao Wang, founder and president of Center for China and Globalization, a Beijing-based think tank.”DeepSeek represents exactly what Beijing is keen to see by ‘new-quality productive force’ that will push China forward,” Wang added, referring to a strategy coined by Xi last year that bets on technological breakthroughs to fuel growth and productivity gains across the economy.Chinese leadership last year vowed “a leap forward” by spurring new growth drivers based on innovation in advanced sectors, such as AI and semiconductors, as U.S. export controls on advanced equipment and the most advanced semiconductors thwarted its ability to make major tech breakthroughs.With Beijing signaling support for the startup, a growing number of local governments, from Hohhot in northern China to the southern city of Guangzhou and Shenzhen, are launching DeepSeek-powered “public servants” to automate governance, handling requests from administrative paper work to general public services.At least three state-owned telecommunications operators have also adopted the cutting-edge model in recent weeks.Private businesses have tapped the new model to see how it can improve productivity. Automakers, financial services companies, smartphone makers and cloud computing operators including Alibaba, Huawei and Tencent have rushed in recent weeks to integrate with DeepSeek.”With DeepSeek becoming a global household name in a matter of weeks, Beijing is [using it as an opportunity] to showcase China’s tech champions and demonstrate Chinese tech resilience and innovation in the face of US-led controls,” said Reva Goujon, director at Rhodium Group.Labor worriesEconomists, however, warned that the pace of AI adoption should be “managed carefully” in China, which is already facing a weak labor market and high unemployment rate.The “job destruction” effects by AI, while raising labor productivity, could exacerbate deflation and further weaken the economy, Goldman Sachs said.The youth unemployment rate in China has remained above 15%, with over 10 million fresh graduates piling into the job market every year. Job losses have been reported in recent years in the real estate sector, among civil servants, and the financial sector.Compared with the U.S. though, the Chinese labor market is less prone to AI automation risks due to a higher share of less-exposed, physically intensive jobs,” Goldman Sachs pointed out. Agriculture, manufacturing and construction make up 50% of all jobs in China, comparing to only 19% of total employment in the U.S.Sectors that are more prone to adopt AI-driven task automation, such as finance, insurance and services, constitute 14% of jobs stateside, but less than 3% in China, according to the bank’s estimates.A Pew study in 2023 found that 19% of U.S. workers are in jobs with high exposure to AI. That study used the term “exposure” as it’s unclear whether AI’s impact will be positive or negative.While AI application may cause the number of displaced workers to rise in the near term, these workers will eventually find jobs in other sectors where labor has a competitive advantage, helping employment to grow again, Goldman said.— CNBC’s Dylan Butts, Evelyn Cheng contributed to this report.

Source link

Apple unveils cheaper iPhone 16e powerful enough to run AI

digitex — Wed, 19 Feb 2025 16:39:40 +0000

Tim Cook, chief executive officer of Apple Inc., during the first day of in-store sales of Apple’s latest products at Apple’s Fifth Avenue store in New York, US, on Friday, Sept. 20, 2024.Victor J. Blue | Bloomberg | Getty ImagesApple announced a new iPhone model on Wednesday that is priced lower than its main iPhone models, which usually come out in September.The new iPhone is called iPhone 16e, and it will retail for $599 when it goes on sale later this month.The new iPhone 16e doesn’t have a home button and fingerprint sensor, instead, it uses Apple’s FaceID scanner and modern design including a sensor notch at the top of the screen. It uses Apple’s A18 chip, which is also used in the main iPhone 16 models. The updated processor means that the iPhone 16e can run all the same apps and games that more expensive iPhones can run. It comes in black and white.The phone also includes Apple’s first cellular modem, which it calls C1. iPhones have used Qualcomm modems for the past few years. It also has a single camera lens, versus as many as three on the most advanced iPhones.Apple is releasing a new low-cost iPhone as sales have been mixed in recent quarters and Apple seeks growth for its most important product category. For the December quarter, Apple’s overall iPhone sales were down 1% on an annual basis. Apple still sold over $69 billion of phones in the period.Low-end iPhones are important to Apple as it gets new customers into their ecosystem, and the new device supports Apple Intelligence, making it Apple’s least expensive new phone that can access features like image generation and notification summaries.Apple’s current models are the iPhone 16, which starts at $799, and the iPhone 16 Pro, which starts at $999. Before Tuesday’s launch, only the iPhone 16 and iPhone 15 Pro models were the only phones with access to Apple Intelligence.Apple has released a less-expensive iPhone model to round out the bottom of its lineup since 2016. The iPhone SE, as it was called then, got further updates in 2020 and 2022. Generally, the iPhone SE reuses an older design than the newest iPhones, but Apple updates its components such as its processor so it can receive ongoing software updates.After the old iPhone SE sells out, there won’t be any new iPhones with a fingerprint scanner. It was also less expensive than the iPhone 16e at $429.The launch also marks a new approach to Apple’s famous product launches, which garner media attention around the world. Previously, Apple would reveal new products live and on-stage, at a presentation on its campus in California. In 2020, Apple stopped inviting people to live launches and started screening marketing videos on its website and YouTube instead.In recent years, Apple has started to quietly release new products through press releases, such as the new Macs it announced last September. It may be testing how much buzz it can get for one of these quieter launches. Last week, ahead of Wednesday’s launch, Apple CEO Tim Cook posted a message on social media: “Get ready to meet the newest member of the family.”Zoom In IconArrows pointing outwardsDon’t miss these insights from CNBC PRO

Source link

Elon Musk’s xAI launches Grok 3 model amid tight AI competition

digitex — Tue, 18 Feb 2025 11:18:29 +0000

Muhammed Selim Korkutata | Anadolu | Getty ImagesElon Musk’s xAI on Tuesday unveiled its latest artificial intelligence model, Grok 3, claiming it can outperform offerings from OpenAI and China’s DeepSeek based on early testing, which included standardized tests on math, science and coding. “We’re very excited to present Grok 3, which is, we think, an order of magnitude more capable than Grok 2 in a very short period of time,” Musk said at a demonstration of Grok 3 that was streamed on his social media platform X. The team also said it was launching a new product called “Deep Search,” which would act as a “next generation search engine.” Grok 3 will be rolled out for premium X subscribers, starting Tuesday stateside, and will also be accessible through a separate subscription for the model’s web and app versions, the xAI team said.Speaking at The World Governments Summit in Dubai last week Musk had dubbed the model “scary smart,” with powerful reasoning capabilities, claiming it outperformed all other existing models in xAI’s internal tests. “This might be the last time that an AI is better than Grok,” Musk said at the time, adding that it was trained on “a lot of synthetic data,” and was capable of reflecting upon its mistakes to achieve logical consistency. The xAI team claimed that an early iteration of Grok 3 had been given better ratings than existing competitors on Chatbot Arena, a crowdsourced website that pits different AI models against each other in blind tests.Toward the end of the product demo, Musk said that the company will keep improving the model.”We should emphasize that this is kind of a beta, meaning that you should expect some imperfections at first, but we will improve it rapidly, almost every day,” he said, adding that the voice assistance for the model would be released at a later time.Intense competition Musk, who has been quite vocal about the potential dangers of artificial intelligence, started xAI in 2023 entering the generative AI market that includes OpenAI’s ChatGPT. In September last year, OpenAI launched its most advanced model, the o1, which came with reasoning abilities and was able to solve relatively complex science, coding and math tasks. Musk, along with Sam Altman, helped create OpenAI as a nonprofit in 2015.However, in recent years Musk and OpenAI’s leadership have been feuding. Musk recently led an investor group that submitted a proposal to buy the AI startup’s nonprofit parent for $97.4 billion — an offer OpenAI declined. Last month, Chinese start-up DeepSeek shocked the AI market when it released a technical paper that claimed one of its open source models was able to rival the performance of OpenAI’s o1 model despite using a cheaper, less energy-intensive process.It accomplished the feat in the face of the U.S. restricting leading AI chipmaker Nvidia from selling its cutting-edge GPUs — used for training AI models — to China. XAI has a “Colossus supercomputer,” for training AI, which it said last year was utilizing a cluster of 100,000 advanced Nvidia GPUs for AI training. On Tuesday, the company revealed that it doubled the size of its GPU cluster for the training of Grok 3.While many AI and tech experts have told CNBC that DeepSeek has intensified AI competition, showing what can be done with less advanced technology, others are more skeptical about its impact.

Source link

EU kicks off landmark AI Act enforcement as first restrictions apply

digitex — Mon, 03 Feb 2025 08:23:45 +0000

The European Union is so far the only jurisdiction globally to drive forward comprehensive rules for artificial intelligence with its AI Act.Jaque Silva | Nurphoto | Getty ImagesThe European Union formally kicked off enforcement of its landmark artificial intelligence law Sunday, paving the way for tough restrictions and potential large fines for violations.The EU AI Act, a first-of-its-kind regulatory framework for the technology, formally entered into force in August 2024.On Sunday, the deadline for prohibitions on certain artificial intelligence systems and requirements to ensure sufficient technology literacy among staff officially lapsed.That means companies must now comply with the restrictions and can face penalties if they fail to do so.The AI Act bans certain applications of AI which it deems as posing “unacceptable risk” to citizens.Those include social scoring systems, real-time facial recognition and other forms of biometric identification that categorize people by race, sex life, sexual orientation and other attributes, and “manipulative” AI tools.Companies face fines of as much as 35 million euros ($35.8 million) or 7% of their global annual revenues — whichever amount is higher — for breaches of the EU AI Act.The size of the penalties will depend on the infringement and size of the company fined.That’s higher than the fines possible under the GDPR, Europe’s strict digital privacy law. Companies face fines of up to 20 million euros or 4% of annual global turnover for GDPR breaches.’Not perfect’ but ‘very much needed’It’s worth stressing that the AI Act still isn’t in full force — this is just the first step in a series of many upcoming developments.Tasos Stampelos, head of EU public policy and government relations at Mozilla, told CNBC previously that while it’s “not perfect,” the EU’s AI Act is “very much needed.””It’s quite important to recognize that the AI Act is predominantly a product safety legislation,” Stampelos said in a CNBC-moderated panel in November.”With product safety rules, the moment you have it in place, it’s not a done deal. There are a lot of things coming and following after the adoption of an act,” he said.”Right now, compliance will depend on how standards, guidelines, secondary legislation or derivative instruments that follow the AI Act, that will actually stipulate what compliance looks like,” Stampelos added.In December, the EU AI Office, a newly created body regulating the use of models in accordance with the AI Act, published a second-draft code of practice for general-purpose AI (GPAI) models, which refers to systems like OpenAI’s GPT family of large language models, or LLMs.The second draft contained exemptions for providers of certain open-source AI models while including the requirement for developers of “systemic” GPAI models to undergo rigorous risk assessments.Setting the global standard?Several technology executives and investors are unhappy with some of the more burdensome aspects of the AI Act and worry it might strangle innovation.In June 2024, Prince Constantijn of the Netherlands told CNBC in an interview that he’s “really concerned” about Europe’s focus on regulating AI.”Our ambition seems to be limited to being good regulators,” Constantijn said. “It’s good to have guardrails. We want to bring clarity to the market, predictability and all that. But it’s very hard to do that in such a fast-moving space.”Still, some think that having clear rules for AI could give Europe leadership advantage.”While the U.S. and China compete to build the biggest AI models, Europe is showing leadership in building the most trustworthy ones,” Diyan Bogdanov, director of engineering intelligence and growth at Bulgarian fintech firm Payhawk, said via email.”The EU AI Act’s requirements around bias detection, regular risk assessments, and human oversight aren’t limiting innovation — they’re defining what good looks like,” he added.

Source link

DeepSeek was downloaded by millions. Deleting it might come next

digitex — Sun, 02 Feb 2025 18:42:16 +0000

This week’s news that the DeepSeek Chatbot app, developed by China, was downloaded from the Apple app store significantly more times than the US-developed ChatGPT from Open AI, wiped billions off the global tech market.Leon Neal | Getty Images News | Getty ImagesPlenty of Americans are discovering the AI search powers of DeepSeek, the breakthrough Chinese generative AI app that surged to No. 1 downloaded status on Apple’s App Store last week. But in an era of U.S.-China technology rivalry and mistrust, and entities from NASA to the U.S. Navy and Taiwanese government prohibiting use of DeepSeek within days, is it wise of millions of Americans to let the app start playing around with their personal search inquiries?The sudden rise of DeepSeek — created on a rapid timeline and on a budget reportedly much lower than previously thought possible — caught AI experts off guard, though skepticism over the claims remain and some estimates suggest the Chinese company understated costs by hundreds of millions of dollars. Privacy advocates were caught off guard, too, and their concerns aren’t predicated on AI development costs, and they already warning that Americans are putting themselves and their privacy at risk.The amount of data and information that bad actors in China could harvest from DeepSeek is 20 times worse than what could be collected from a Google search, says Dewardric McNeal, managing director and senior policy analyst at risk management firm Longview Global, which advises companies on China strategy.”It is a rich trove of intelligence,” said McNeal, who has studied the details of Chinese government data sharing requirements for domestic firms.There are obvious risks, he said, such as personal banking or health information that can be stolen, and prominent cybersecurity firms are already reporting vulnerabilities in DeepSeek. DeepSeek itself reported being hit with a major cyberattack last week.But McNeal is just as worried about the “bigger picture” competition between nations.”I want us to speak broader than just the narrow data; we often don’t speak about the degree to which this information paints a mental map through understanding queries,” McNeal said.For example, Chinese intelligence could use the broader patterns of queries in DeepSeek to learn about various American industries and to sow division among the public.”The world won’t end tomorrow because I logged into DeepSeek,” McNeal said, but he added that does not mean there isn’t considerable risk involved. The AI’s open-source approach, for one, could give China access to US-based supply chains at an industry level, allowing them to learn what companies are doing and better compete against them. “National security professionals are thinking about it in those terms,” McNeal said.Matt Pearl, a special advisor to the deputy national security advisor at the National Security Council in the Biden administration and now the Strategic Technologies Program director at the Center for Strategic and International Studies, said DeepSeek’s privacy policy implies that people have control over what is collected, but it should induce alarm.”DeepSeek’s privacy policy is not worth the paper it is written on,” Pearl said. DeepSeek is subjected to PRC laws and anything entered into the app is fair game. Through keystroke patterns, a DeepSeek user can be tracked across all devices, information gathered from advertisers, and DeepSeek could also seek to leverage cameras and microphones, according to Pearl.”If they can do it technically in the app and the PRC determines it is something they want to do, then it poses a danger,” Pearl said.But the threat that Pearl said most keeps him up at night is related to cybersecurity and the potential for a mass malware injection. “It is hard to emphasize all the different potential ways in which it could be used. And, in theory, it could be done in a single update to the app,” he said.Officials at High Flyer, the Chinese-backed hedge fund which created DeepSeek, did not respond to a request for comment.ChatGPT still far ahead of DeepSeekDespite the outsized impact on the markets and leading AI firms including Nvidia, DeepSeek still has a long way to go to catch up to rival ChatGPT, which is continuing to raise a formidable war chest — a few days after the DeepSeek headlines dominated the tech and markets news cycle, OpenAI was reportedly in talks for a $40 billion funding round.DeepSeek remains far behind ChatGPT in consumer activity, according to online analytics platform Semrush, with the OpenAI app maintaining an average daily visit count in the tens of millions. But ChatGPT has experienced a recent dip in traffic — it had 22.1 million visitors on October 1, 2024, but that had declined to 14.9 million by January 19, according to Semrush. At the same time, even before it became a major national news story, DeepSeek’s online footprint was growing — from 2.3K average U.S. daily visits on October 1, 2024, to 71.2K by January 19 (a week before it caused the stock market to tank).Joe Jones, director of research and insights for The International Association of Privacy Professionals, a policy-neutral nonprofit that promotes privacy and AI governance, says that disruptors like DeepSeek can make the organization’s job more difficult.”It is challenging for people to do that work when you have proliferating laws that are complex, diverse, and often in tension, and technologies like DeepSeek that come at you from left field, upend status quos and make you rethink good governance,” Jones said. The fact that the debate is playing out across borders makes it more contentious. “This has gotten a whole lot more complex in this turbocharged geopolitical environment,” Jones added.Liang Wenfeng, founder of startup DeepSeek, delivers the keynote speech during the 10th China Private Equity Golden Bull Awards on August 30, 2019 in Shanghai, China.Vcg | Visual China Group | Getty ImagesThe challenges will not be solved any time soon, according to Brendan Englot, director of the Stevens Institute for Artificial Intelligence at the Stevens Institute of Technology. “AI is now a global international competition, and we will see breakthroughs all over the world,” Englot said.The open-source technology approach leveraged by DeepSeek — an approach promoted by Meta — means more DeepSeeks are coming. “There will be many more disruptions just like this one shortly,” Englot said.For consumers, the tradeoffs will need to be navigated with an understanding of the data implications. The tools are being designed to improve lives by increasing efficiency and creativity. “So it is tempting to share our data, but you have to assume it is fair game to be used once you do,” Englot said.DeepSeek’s success suggests that export controls on advanced chips intended to slow Chinese AI efforts might need to be even stricter, but are also no silver bullet. “To be clear, they’re not a way to duck the competition between the US and China,” wrote Dario Amodei, CEO of gen AI startup Anthropic, in a blog post this week. “In the end, AI companies in the US and other democracies must have better models than those in China if we want to prevail. But we shouldn’t hand the Chinese Communist Party technological advantages when we don’t have to.”Pearl says it may ultimately fall to the U.S. government to regulate or legislate.”The government has the ability under the same law that they had to ban TikTok; they can ban DeepSeek. The law that was passed doesn’t just apply to TikTok,” Pearl said, citing provisions of the law related to a company controlled by a foreign adversary, or deemed by the U.S. president to be a threat to national security. Pearl thinks that despite President Trump’s backing away from a TikTok ban, US-based tech companies will likely lobby harder when it comes to AI.”I think you will see U.S. tech companies lobbying the admin hard on this and saying DeepSeek will corner the AI market, and it is critical to have U.S. companies at the forefront,” Pearl said. “Trump will hear from them at the highest levels. China has made champions by keeping the USA out of their markets, why should we allow them to dominate our markets?”

Source link

DeepSeek’s hardware spend could be as high as $500 million: Report

digitex — Fri, 31 Jan 2025 23:45:58 +0000

Faisal Bashir | Lightrocket | Getty ImagesChina’s DeepSeek became the biggest topic in tech this week, with many in the industry and on Wall Street focused on a single number: $6 million.In DeepSeek’s paper about its newest artificial intelligence model, the company said that its total training costs amounted to $5.576 million, based on the rental price of Nvidia’s graphics processing units. DeepSeek included a clear caveat, saying that the number included only the model’s “official training” and excluded the costs tied to “prior research and ablation experiments on architectures, algorithms, or data.”Early in the week, DeepSeek’s AI Assistant took the coveted spot for most-downloaded free app in the U.S. on Apple’s App Store, dethroning OpenAI’s ChatGPT. Global tech stocks sold off, with chipmakers Nvidia and Broadcom losing a combined $800 billion in market cap on Monday.A new report from SemiAnalysis, a semiconductor research and consulting firm, added more context to DeepSeek’s expenses. The firm estimated that DeepSeek’s hardware spend is “well higher than $500M over the company history,” adding that R&D costs and total cost of ownership are significant. Generating “synthetic data” for the model to train on would require “considerable amount of compute,” SemiAnalysis wrote.The report said the Claude 3.5 Sonnet from Anthropic cost “$10s of millions to train,” but noted that Anthropic raised billions for dollars from Amazon and Google, an indication of how much more money is required to run the models and the company.”It’s because they have to experiment, come up with new architectures, gather and clean data, pay employees, and much more,” SemiAnalysis said.DeepSeek’s own paper does not include an estimation of its compute costs. The company didn’t immediately respond to a request for comment.”To be clear DeepSeek is unique in that they achieved this level of cost and capabilities first,” SemiAnalysts wrote. The firm added that DeepSeek’s R1 “is a very good model” and that “catching up to the reasoning edge this quickly is objectively impressive.”Experts and analysts this week touted the quality of DeepSeek’s model, and noted how impressive it is considering the U.S. curbed chip exports to China three times in three years. That led to concerns that the U.S. is falling behind its chief adversary in a market that’s predicted to top $1 trillion in revenue within a decade.Bernstein analysts wrote in a note Monday that “according to the many (occasionally hysterical) hot takes we saw [over the weekend,] the implications range anywhere from ‘That’s really interesting’ to ‘This is the death-knell of the AI infrastructure complex as we know it.'”DeepSeek was founded in 2023 by Liang Wenfeng, co-founder of High-Flyer, a quantitative hedge fund focused on AI. The AI startup reportedly grew out of the hedge fund’s AI research unit in April 2023 to focus on large language models and reaching artificial general intelligence, or AGI — a branch of AI that equals or surpasses human intellect on a wide range of tasks, and that OpenAI and others are pursuing.DeepSeek is still wholly owned by and funded by High-Flyer, according to analysts at Jefferies.The buzz around DeepSeek began picking up steam earlier this month, when the startup released R1, its reasoning model that rivals OpenAI’s o1. It’s open-source, meaning that any AI developer can use it.Like other Chinese chatbots, DeepSeek’s has limitations on certain topics: When asked about some of Chinese leader Xi Jinping’s policies, for instance, DeepSeek reportedly steers the user away from similar lines of questioning.OpenAI CEO Sam Altman has praised the model publicly, but the company has also said it believes there’s evidence that DeepSeek improperly harvested OpenAI data to build its product.At an event in Washington, D.C., on Thursday hosted by OpenAI, Altman said DeepSeek is “clearly a great model.””This is a reminder of the level of competition and the need for democratic Al to win,” he said. He said it also points to the “level of interest in reasoning, the level of interest in open source.”WATCH: Nvidia CEO Jensen Huang and President Trump meet on AI policy

Source link

DeepSeek, next-generation AI agents may erode value of large models

digitex — Fri, 31 Jan 2025 07:09:50 +0000

Jaap Arriens | NurPhoto via Getty ImagesLarge language models like those developed by Microsoft-backed firm OpenAI are set to become commoditized this year amid rapid advances toward next-generation artificial intelligence agents and more nimble, open-source rivals, according to top tech executives.Last week, Chinese AI firm DeepSeek released R1, an open-source reasoning model that claims to rival OpenAI’s o1 model on both cost and performance. Open-source refers to software whose source code is made freely available on the open web for possible modification and redistribution.This week, growing awareness of DeepSeek’s new model led to a severe slump in shares of Nvidia and other tech giants, as investors feared a possible retrenchment in spending on the powerful graphics processing units required to train and run advanced AI workloads.Nvidia lost close to $600 billion in market capitalization on Monday — the biggest single-day drop for any company in U.S. history.Meanwhile, executives and scientists at leading AI labs are all talking up a shift from large language models to so-called “AI agents” that can carry out actions on your behalf.LLMs are the foundational technology behind today’s generative AI apps. However, experts believe a push toward agentic AI systems — which incorporate LLM technology — this year will erode the value of these models.Commoditization of LLMsLLMs are expected to become more of a commodity in the near future as the tech becomes increasingly advanced and the costs involved in training and running them continue to drop.Thomas Wolf, co-founder and chief science officer of Hugging Face, suggested LLMs will become more integrated into intelligent systems linked to the company’s own databases.”I think people are moving from this craziness around the model, understanding that, thanks to open-source … a lot of these models are going to be free and freely available,” Wolf told CNBC. Hugging Face is a popular code repository for open-source AI projects.”In the internet revolution, we’re moving from building websites as the main business to actually building internet-native companies — so, the Airbnb of AI, the Stripe of AI,” he added. “They are not about the model. They are about the system and how you make the model useful for tasks.”The arrival of DeepSeek’s R1 model last week is viewed by some tech CEOs as a further indicator LLMs are becoming increasingly commoditized.R1 incorporates some key aspects that differentiate it from other models on the market. For example, the model incorporates a “mixed precision” framework that uses a combination of full-precision 32-bit floating point (FP32) numbers and low-precision 8-bit floating point (FP8) numbers.The latter is faster to process but can be less accurate. However, rather than rely on one or the other, DeepSeek uses FP8 for most calculations and switches to FP32 for certain tasks where a higher degree of accuracy is required.”As AI gets more efficient and accessible, we will see its use skyrocket, turning it into a commodity we just can’t get enough of,” Microsoft CEO Satya Nadella wrote on social media platform X, in an apparent reference to DeepSeek.Meanwhile, Matt Calkins, CEO of U.S. software firm Appian, told CNBC that DeepSeek’s success simply shows that AI models are going to become more of a commodity in the future.”In my opinion, we’re going to see a commoditization of AI. Many companies will achieve competitive AI, and a lack of differentiation will be bad for high-spending first-movers,” Calkins said via email.What are AI agents?In the last two years, much of the focus has been on LLM-powered products, from generative AI chatbots like OpenAI’s ChatGPT to tools that can produce images and video content.LLM-based generative apps like ChatGPT rely on a certain amount of handholding from users to continuously prompt them for outputs, whereas in contrast, AI agents are designed to be more action-oriented.If you wanted to book a doctor’s appointment, for example, you could ask ChatGPT for information on your nearest GP practices — similar to how you’d query the web using Google’s search engine.In future, tech execs predict the next generation AI will involve agents that can find out where your nearest GP is, provide available times for your doctor, and book you an appointment — all directly within a single platform. This is expected to transform the way that we interact with the web.”Over the decades since the web was invented, the standard interface of webpages and menus hasn’t really changed very much,” Bobby Blumofe, chief technology officer of Akamai, told CNBC previously.”We’re sort of living in this tyranny of being tethered to a screen and being tethered to web pages and menus you have to remember how to navigate.”The shift from models to agentsLast week, Arthur Mensch, the CEO of Mistral said that he thinks the “focus should shift to systems” that integrate both language models and contextual business data.”That’s that’s where this is shifting,” Mensch told CNBC’s Arjun Kharpal at the World Economic Forum’s annual gathering in Davos, Switzerland.”It also means that the industry that is adopting it, is going to distill its expertise into those systems — and that’s where that’s going to be necessary to actually drive the value of AI into those industries.”Mistral, which is backed by Microsoft, has swiftly become one of Europe’s buzziest AI firms. Founded only two years ago, it was last valued by investors at around $6 billion in June 2024.The trend of agents has already gained considerable traction. Last week, OpenAI launched its own agent, called Operator. OpenAI describes it as “an agent that can go to the web to perform tasks for you,” adding it’s trained to interact with “the buttons, menus, and text fields that people use daily” on the web.Microsoft, which has invested billions of dollars in OpenAI, rolled out AI agents itself in November.Anthropic, an Amazon-backed AI startup founded by ex-OpenAI research executives, last year introduced “computer use,” which is a capability that allows its AI agents to use computers like humans to complete complex tasks.Dario Amodei, CEO of Anthropic, told CNBC’s Andrew Ross Sorkin in Davos last week that Computer Use should be thought of as a precursor to a “broader vision” that integrates these agents with commonly used applications.”We are planning this year in 2025 to build something that we call the ‘virtual collaborator,'” Amodei said, describing it as “an agent that operates on your computer.””You can give it tasks. You can say, write this feature for this app, and it’ll write some code, it’ll compile the code, it’ll check the code, it’ll talk to its coworkers on Slack, or on Google Docs, or on some other platform,” Anthropic’s chief explained.Anthropic plans to gradually build out pieces of this vision over the next year — while still ensuring it’s “safe and controllable,” Amodei added.Paul O’Sullivan, Salesforce’s chief technology officer for U.K. and Ireland, predicts verbal communication will become more important in the new world of AI agents. Salesforce was arguably early to the agentic AI game, having launched its own Agentforce system in September before Microsoft, OpenAI or Anthropic.”We will converse more,” he told CNBC in a recent interview. “The screens will get smaller, our eyes will get worse. But they’ll probably be replaced with mobile devices that are more powerful than laptops anyway.”

Source link