U.S. government – Digitex Solutions

China’s DeepSeek has some big AI claims; not all experts are convinced

digitex — Thu, 30 Jan 2025 11:49:08 +0000

Chinese artificial intelligence firm DeepSeek rocked markets this week with claims its new AI model outperforms OpenAI’s and cost a fraction of the price to build.The assertions — specifically that DeepSeek’s large language model cost just $5.6 million to train — have sparked concerns over the eyewatering sums that tech giants are currently spending on computing infrastructure required to train and run advanced AI workloads.But not everyone is convinced by DeepSeek’s claims.CNBC asked industry experts for their views on DeepSeek, and how it actually compares to OpenAI, creator of viral chatbot ChatGPT which sparked the AI revolution.What is DeepSeek?Last week, DeepSeek released R1, its new reasoning model that rivals OpenAI’s o1. A reasoning model is a large language model that breaks prompts down into smaller pieces and considers multiple approaches before generating a response. It is designed to process complex problems in a similar way to humans.DeepSeek was founded in 2023 by Liang Wenfeng, co-founder of AI-focused quantitative hedge fund High-Flyer, to focus on large language models and reaching artificial general intelligence, or AGI.AGI as a concept loosely refers to the idea of an AI that equals or surpasses human intellect on a wide range of tasks.Much of the technology behind R1 isn’t new. What is notable, however, is that DeepSeek is the first to deploy it in a high-performing AI model with — according to the company — considerable reductions in power requirements.”The takeaway is that there are many possibilities to develop this industry. The high-end chip/capital intensive way is one technological approach,” said Xiaomeng Lu, director of Eurasia Group’s geo-technology practice.”But DeepSeek proves we are still in the nascent stage of AI development and the path established by OpenAI may not be the only route to highly capable AI.” How is it different from OpenAI?DeepSeek has two main systems that have garnered buzz from the AI community: V3, the large language model that unpins its products, and R1, its reasoning model.Both models are open-source, meaning their underlying code is free and publicly available for other developers to customize and redistribute.DeepSeek’s models are much smaller than many other large language models. V3 has a total of 671 billion parameters, or variables that the model learns during training. And while OpenAI doesn’t disclose parameters, experts estimate its latest model to have at least a trillion.In terms of performance, DeepSeek says its R1 model achieves performance comparable to OpenAI’s o1 on reasoning tasks, citing benchmarks including AIME 2024, Codeforces, GPQA Diamond, MATH-500, MMLU and SWE-bench Verified.Read more DeepSeek coverageIn a technical report, the company said its V3 model had a training cost of only $5.6 million — a fraction of the billions of dollars that notable Western AI labs such as OpenAI and Anthropic have spent to train and run their foundational AI models. It isn’t yet clear how much DeepSeek costs to run, however.If the training costs are accurate, though, it means the model was developed at a fraction of the cost of rival models by OpenAI, Anthropic, Google and others.Daniel Newman, CEO of tech insight firm The Futurum Group, said these developments suggest “a massive breakthrough,” although he shed some doubt on the exact figures.”I believe the breakthroughs of DeepSeek indicate a meaningful inflection for scaling laws and are a real necessity,” he said. “Having said that, there are still a lot of questions and uncertainties around the full picture of costs as it pertains to the development of DeepSeek.”Meanwhile, Paul Triolio, senior VP for China and technology policy lead at advisory firm DGA Group, noted it was difficult to draw a direct comparison between DeepSeek’s model cost and that of major U.S. developers.”The 5.6 million figure for DeepSeek V3 was just for one training run, and the company stressed that this did not represent the overall cost of R&D to develop the model,” he said. “The overall cost then was likely significantly higher, but still lower than the amount spent by major US AI companies.” DeepSeek wasn’t immediately available for comment when contacted by CNBC.Comparing DeepSeek, OpenAI on priceDeepSeek and OpenAI both disclose pricing for their models’ computations on their websites.DeepSeek says R1 costs 55 cents per 1 million tokens of inputs — “tokens” referring to each individual unit of text processed by the model — and $2.19 per 1 million tokens of output.In comparison, OpenAI’s pricing page for o1 shows the firm charges $15 per 1 million input tokens and $60 per 1 million output tokens. For GPT-4o mini, OpenAI’s smaller, low-cost language model, the firm charges 15 cents per 1 million input tokens.Skepticism over chipsDeepSeek’s reveal of R1 has already led to heated public debate over the veracity of its claim — not least because its models were built despite export controls from the U.S. restricting the use of advanced AI chips to China. DeepSeek claims it had its breakthrough using mature Nvidia clips, including H800 and A100 chips, which are less advanced than the chipmaker’s cutting-edge H100s, which can’t be exported to China.However, in comments to CNBC last week, Scale AI CEO Alexandr Wang, said he believed DeepSeek used the banned chips — a claim that DeepSeek denies.Nvidia has since come out and said that the GPUs that DeepSeek used were fully export-compliant.The real deal or not?Industry experts seem to broadly agree that what DeepSeek has achieved is impressive, although some have urged skepticism over some of the Chinese company’s claims.”DeepSeek is legitimately impressive, but the level of hysteria is an indictment of so many,” U.S. entrepreneur Palmer Luckey, who founded Oculus and Anduril wrote on X.”The $5M number is bogus. It is pushed by a Chinese hedge fund to slow investment in American AI startups, service their own shorts against American titans like Nvidia, and hide sanction evasion.”Seena Rejal, chief commercial officer of NetMind, a London-headquartered startup that offers access to DeepSeek’s AI models via a distributed GPU network, said he saw no reason not to believe DeepSeek.”Even if it’s off by a certain factor, it still is coming in as greatly efficient,” Rejal told CNBC in a phone interview earlier this week. “The logic of what they’ve explained is very sensible.”However, some have claimed DeepSeek’s technology might not have been built from scratch.”DeepSeek makes the same mistakes O1 makes, a strong indication the technology was ripped off,” billionaire investor Vinod Khosla said on X, without giving more details.It’s a claim that OpenAI itself has alluded to, telling CNBC in a statement Wednesday that it is reviewing reports DeepSeek may have “inappropriately” used output data from its models to develop their AI model, a method referred to as “distillation.””We take aggressive, proactive countermeasures to protect our technology and will continue working closely with the U.S. government to protect the most capable models being built here,” an OpenAI spokesperson told CNBC.Commoditization of AIHowever the scrutiny surrounding DeepSeek shakes out, AI scientists broadly agree it marks a positive step for the industry.Yann LeCun, chief AI scientist at Meta, said that DeepSeek’s success represented a victory for open-source AI models, not necessarily a win for China over the U.S. Meta is behind a popular open-source AI model called Llama.”To people who see the performance of DeepSeek and think: ‘China is surpassing the US in AI.’ You are reading this wrong. The correct reading is: ‘Open source models are surpassing proprietary ones’,” he said in a post on LinkedIn.”DeepSeek has profited from open research and open source (e.g. PyTorch and Llama from Meta). They came up with new ideas and built them on top of other people’s work. Because their work is published and open source, everyone can profit from it. That is the power of open research and open source.”WATCH: Why DeepSeek is putting America’s AI lead in jeopardy- CNBC’s Katrina Bishop and Hayden Field contributed to this report

Source link

How I got rid of the A.I. tool from Workspace.

digitex — Wed, 29 Jan 2025 19:16:03 +0000

Sign up for the Slatest to get the most insightful analysis, criticism, and advice out there, delivered to your inbox daily.

Last Tuesday, I returned from a blissfully computerless vacation abroad and logged back into work, only to find an uninvited visitor in my inbox: Gemini, Google’s A.I. assistant, had been deployed to my workspace without any warning. When I logged into Gmail for the first time in two weeks, a side panel popped out, introducing itself like the ghost of Clippy past. “Hi Shasha, how can I help you today?” it asked, along with offers of a few options like summarizing conversations, drafting an email, and showing me unread emails—the latter of which would have been immediately possible had that pop-up not been in my way to begin with.

A.I. is controversial, and for good reason. Generative tools such as the large language model ChatGPT, or image generators like Midjourney, simultaneously threaten livelihoods and spread harmful misinformation. That’s not to mention the enormous environmental impact these resource-draining systems have. There is no doubt A.I. is here to stay, but the question is in what capacity and for whose benefit. And while there is no blanket answer, ethically speaking, it should always involve personal consent, especially when that consent is being presumed by a billion-dollar company.

So I headed to my user settings to turn Gemini off. But if you also use Google for work and have tried to do the same, you’ll have also noticed it’s not possible. Now, this is where most people’s struggle would end, at a loss for choice, and with maybe an annoyed email to their IT guy.

But I happen to be Slate’s annoyed IT guy—so I logged into my admin panel, located the brand-new “Generative AI” tab, and navigated to the object of my concern: “Gemini for Google Workspace” … only to find there were no settings to manage whatsoever, nor any option to turn it off.

Shasha Léonard

This is particularly baffling, given that most—if not all—of Workspace’s features are customizable through the admin panel. That’s literally what we pay Google for.

For example, if I want to manage Drive external/internal viewing settings for certain users, I can do that. If I want to limit a user’s sharing abilities in Calendar, I can do that. If, for whatever reason, I want to turn those features off entirely despite being major components of a user’s workflow, I can do that. Why would Gemini be any different?

Second-guessing myself, I turned off the Gemini App instead and hoped for the desired changes to take effect (which Google warns you can take some time). But a few days passed with Gemini still asking me if I needed help finishing my sentences in Gmail, so like any other irritated and long-standing denizen of the internet, I headed over to Reddit to see what other sysadmins had to say about this. What I found unsettled me.

In one post, Reddit user heretocomplain123 described how they had also encountered Gemini for Google Workspace, but couldn’t immediately turn it off. Eventually, they were able to find the solution after contacting a Google rep. “I chatted with customer support, and at my request they ADDED the settings, after which I was able to toggle off all Gemini for Workspace settings (gmail, etc.),” the user wrote.

You had to go ask Google for the ability to turn it off.

Thanks to the screenshots OP attached, many of us were able to follow suit, some waiting a few hours in the customer support queue, some waiting several days for Gemini to finally go away. I waited 10 minutes. Here are screenshots from my own chat with Google Workspace Support.

Shasha Léonard

According to what I just experienced, and what many other paid users are experiencing, Google has made opting in to generative A.I. the default. You have to go the extra mile and wait, sometimes hours, in the support queue to even have the option to opt out. Just as Meta’s search bar doubles as an A.I. search, or how the Transportation Security Administration rolled out default biometric screening at airports, these are forms of a manipulative design strategy to all but force you to engage with and train A.I. (It is also an excellent example of enshittification.)

While the ethics of consent are the most immediate here, it’s also critical to acknowledge generative A.I.’s well-documented environmental impact, as well as its growing role in modern warfare, like Meta opening its A.I. model to the U.S. military, further harming the environment. The push to normalize the technology’s usage, often without user input, raises urgent questions about accountability and sustainability.

Google Support chat assured myself and others in the subreddit that the option to opt out of Gemini for Google Workspace would be rolled out for everyone … eventually. But the issue at hand is that Gemini for Google Workspace was rolled out as a finished product without the ability to opt out or manage its settings. The issue isn’t just about Gemini’s intrusiveness—it’s about the erosion of user autonomy. By making A.I. opt-out rather than opt-in, Google is setting a dangerous precedent: the presumption of consent by default.

Utpal Dholakia describes this default in an article for Psychology Today as a technique used by anyone “interested in influencing others’ behavior,” as well as representing an asymmetrical power dynamic. “The manipulator is powerful, and the manipulated is weak,” he wrote. Governments and businesses use behavioral nudges to influence citizens and employees—not the other way around. “If I tried to nudge the U.S. government to lower my tax bill, an outcome that I might, no doubt, consider to be virtuous and acceptable, I would very quickly find myself in prison.”

Default opt-ins create a massive power imbalance—one that could “produce tremendous and irreversible harm to a great many people,” he adds.

In a recent interview, Sam Altman, CEO of OpenAI, one of the largest and most influential A.I. companies in the U.S., stated that “the entire structure of society will be up for debate and reconfiguration” as A.I. continues to evolve. If this is the case, then this social contract for how we use—or are made to use—A.I. is actively being negotiated in instances like Google pushing Gemini on our workspaces.

It is unethical that I had to reach out to Google Support to ask for the option to turn off Gemini for my workspace, because all users should have the option to opt in or out, if they so choose. Using our right to say no where we can, as annoying or futile as it may seem, is crucial, as it shapes the future of an already skewed power dynamic where saying no might not be possible anymore.

Staff
The A.I. Will See You Now
Read More

Some might argue that this approach is necessary to ensure widespread adoption and improve the tool through user feedback. Because A.I. models rely on data to evolve, making Gemini a default feature guarantees engagement. While this reasoning may seem valid, it ignores the ethics of consent. Presuming consent by default undermines trust and sets a dangerous precedent for how tech companies interact with users.

Your Infant Is Sick. If RFK Jr. Is in Charge, Your Emergency Department Visit May Look Very Different.

There Seems to Be One Principle Governing RFK Jr.’s View of Health

They Used to Love Elon Musk. Now They’re His Biggest Enemies Online.

The First Big Trump Scam Is Already Blowing Up in Everyone’s Faces

While data on A.I. usage suggests massive popular interest, it’s hard to take those numbers at face value in cases where users are automatically opted in to the experience. Just because we all woke up with U2’s Songs of Innocence on our iPhones doesn’t make it the most popular album of all time. This unreliable and inflated data then fuels a vicious cycle: Big tech companies double down on A.I., deploy products like Gemini for Workspace that users can’t opt out of, and then expect immediate returns for their shareholders—while waiting for users to embrace it enthusiastically.

Gemini for Workspace is now disabled for everyone at Slate until Google offers users the option to opt in themselves. As A.I. becomes more commonplace in our digital lives, especially when there is so little legislation to prevent its misuse, we must demand accountability and the right to say no. Opting out should never be a premium feature—it’s a basic right.

Note: No A.I. assistance was used in the writing of this article.

Source link