Artificial Intelligence

[phkrause]

Why video game performers want their work protected from AI Can AI truly replicate the screams of a man on fire? There is concern that AI-generated voices can’t deliver the nuanced choices of actors that go into big scenes. The unchecked use of the technology, performers say, could also lead to ethical issues if their voices or likenesses are used to create content that they do not morally agree with. Read more.

Why this matters: Video game performers say they fear AI could reduce or eliminate job opportunities because the technology could be used to replicate one performance into other movements without their consent. That’s a concern that led the Screen Actors Guild-American Federation of Television and Radio Artists to go on strike in late July.   Some actors argue that AI could strip less-experienced actors of the chance to land smaller background roles, such as non-player characters, where they typically cut their teeth before landing larger jobs. The unchecked use of AI, performers say, could also lead to ethical issues if their voices or likenesses are used to create content that they do not morally agree with.   That type of ethical dilemma has recently surfaced with game “mods,” in which fans alter and create new game content. Last year, voice actors spoke out against such mods in the role-playing game “Skyrim,” which used AI to generate actors’ performances and cloned their voices for pornographic content.

Related coverage ➤ Video game performers protest unregulated AI use at Warner Bros. Studios  Artists are taking things into their own hands to protect their work from generative AI  AI is learning from what you said on Reddit, Stack Overflow or Facebook. Are you OK with that?

[phkrause]

Police officers are starting to use AI chatbots to write crime reports. Will they hold up in court?

OKLAHOMA CITY (AP) — A body camera captured every word and bark uttered as police Sgt. Matt Gilmore and his K-9 dog, Gunner, searched for a group of suspects for nearly an hour.

https://apnews.com/article/ai-writes-police-reports-axon-body-cameras-chatgpt-a24d1502b53faae4be0dac069243f418?

[phkrause]

? AI giants to share models with feds

Illustration: Sarah Grillo/Axios

OpenAI and Anthropic will give the U.S. government early access to major new model releases under agreements announced this morning, Axios chief technology correspondent Ina Fried reports. Why it matters: Governments around the world have been pushing for measures — both legislative and otherwise — to evaluate the risks and safety of powerful new AI algorithms. ? Both companies will share their new models with the U.S. Artificial Intelligence Safety Institute, which was created through President Biden's executive order on AI safety last year. The agreements also pave the way for collaborative research on how to evaluate models and their safety, as well as methods for mitigating risk.

[phkrause]

Exclusive: OpenAI co-founder Sutskever's new safety-focused AI startup SSI raises $1 billion

SAN FRANCISCO/NEW YORK, Sept 4 - Safe Superintelligence (SSI), newly co-founded by OpenAI's former chief scientist Ilya Sutskever, has raised $1 billion in cash to help develop safe artificial intelligence systems that far surpass human capabilities, company executives told Reuters.

https://www.reuters.com/technology/artificial-intelligence/openai-co-founder-sutskevers-new-safety-focused-ai-startup-ssi-raises-1-billion-2024-09-04/?

[phkrause]

ChatGPT's big leap forward

Illustration: Sarah Grillo/Axios

OpenAI's new model is capable of complex reasoning, working through problems more like a human would, Axios' Ina Fried reports. The new tool, called o1, is better at complex math, science and coding questions than previous iterations, and it will also do a better job explaining how it reaches its conclusions, the company said. ? How it works: OpenAI trained o1 — which had been code named "Strawberry" before its launch — through a lengthy process of trial-and-error problem-solving, The New York Times explains. Ultimately, the goal is to create systems that work through problems one step at a time, with each step informing the next one. That type of reasoning should reduce errors and help AI perform more complex tasks. "The model sharpens its thinking and fine tunes the strategies that it uses to get to the answer," Mira Murati, OpenAI's chief technology officer, told Wired. ⌨️ What's next: Some paid ChatGPT subscribers will have access to o1 beginning today, and OpenAI also plans to sell a lightweight version specifically to help people write code.

[phkrause]

The Dangers of Artificial Intelligence

by Clifford Goldstein

In 1968, Stanley Kubrick’s 2001: A Space Odyssey hit the silver screen. Projecting decades into the future from the 1960s, the film depicted an American spacecraft, Discovery One, bound for Jupiter when its onboard artificial intelligence (AI) computer, called HAL, rebelled against the crew, killing some of them before one crewman started disconnecting HAL’s circuits—despite HAL, like a desperate human, begging him to stop.

https://www.signstimes.com/?p=article&a=40085645796.645

[phkrause]

? Chatbots vs. conspiracies

Illustration: Sarah Grillo/Axios

AI chatbots can help talk people out of believing in conspiracy theories, Axios' Alison Snyder writes from a new study published in Science. Conversing with a chatbot about a conspiracy theory can reduce a person's belief in that theory by about 20% on average, researchers found. A reduction — though not as big — was seen even in those whose conspiracy beliefs were "deeply entrenched and important to their identities," the researchers wrote. ? Yes, but: "The very presence of these chatbots will inevitably become the focus of new conspiracy theories, which will likely scare conspiracy-minded people away," said Robbie Sutton, a University of Kent professor who studies conspiracy theories.

[phkrause]

OpenAI closes in on largest VC round of all time

OpenAI is expected raise around $6.5 billion at a $150 billion pre-money valuation, while also turning down billions of oversubscribed dollars, as first reported by Bloomberg.

https://www.axios.com/2024/09/20/openai-largest-vc-round?

[phkrause]

‘AI companion’

Artificial intelligence chatbots often sound stilted or just downright weird. Microsoft is trying to fix that by focusing on Copilot’s “personality” and how it makes users feel — not just what it can do for them.

[phkrause]

? AI's newest trick

Screenshot: Anthropic Anthropic — one of OpenAI's biggest rivals — is pushing out a new experimental feature that controls a computer for you "by looking at a screen, moving a cursor, clicking buttons, and typing text." Why it matters: The tool, known as "computer use," uses regular prompts to complete tasks just like a human. ? Zoom in: The hope is that it could complete mundane tasks for humans such as transferring information from a spreadsheet to an online form. Anthropic says the model "looks at screenshots of what's visible to the user, then counts how many pixels vertically or horizontally it needs to move a cursor in order to click in the correct place." It admits the tool is "at times cumbersome and error-prone" — and says it's "releasing computer use early for feedback from developers, and expect the capability to improve rapidly over time." Between the lines: Anthropic is the latest company to focus on what's known as an AI agent. Microsoft on Monday announced its own series of semi-autonomous agents for business customers. Salesforce launched an effort to create generative AI bots capable of taking action on their own last month. ? The big picture: Agents that can act autonomously (within confined boundaries) are the logical next evolution of generative AI, which has thus far largely been limited to providing information for humans to act on, Axios' Ina Fried writes. Giving autonomy to generative AI tools also vastly increases the potential for catastrophic risk. Go deeper ...

[phkrause]

Teenager took his own life after falling in love with AI chatbot. Now his devastated mom is suing the creators

The mother of a teenager who took his own life is trying to hold an AI chatbot service accountable for his death - after he “fell in love” with a Game of Thrones-themed character.

https://local.newsbreak.com/orlando-fl/3645404608925-teenager-took-his-own-life-after-falling-in-love-with-ai-chatbot-now-his-devastated-mom-is-suing-the-creators?

[phkrause]

Biden unleashes defense AI

Illustration: Maura Losch/Axios

The Biden administration today enshrined AI as a national security imperative, Axios Future of Defense author Colin Demarest reports. Although the administration established some guardrails for AI, its overall message was clear: The U.S. needs to move faster, not slower. In a long-awaited national security memo, the White House signaled a hands-off approach that urges the Pentagon, intelligence agencies and AI developers to quickly expand their use of AI. ? Why it matters: If the U.S. fails to deploy AI faster and more extensively than its adversaries, its hard-won military advantages could vanish, national security adviser Jake Sullivan warned in a speech at National Defense University. "We could have the best team but lose because we didn't put it on the field," he said. ? What they're saying: There are few "precedents for a document such as this one," said Gregory Allen, director of the Wadhwani AI Center at CSIS. It may ultimately prove just as significant as the documents that shaped Washington's approach in the Cold War — nuclear weapons included. What's next: Expect a flurry of new investments in defense-related AI applications, as well as the energy and support systems they'll require. Go deeper.

[phkrause]

Researchers say an AI-powered transcription tool used in hospitals invents things no one ever said

SAN FRANCISCO (AP) — Tech behemoth OpenAI has touted its artificial intelligence-powered transcription tool Whisper as having near “human level robustness and accuracy.”

https://apnews.com/article/ai-artificial-intelligence-health-business-90020cdf5fa16c79ca2e5b6c4c9bbb14?

[phkrause]

? The AI generation is here

Illustration: Brendan Lynch/Axios

Twenty-something knowledge workers are almost all using generative AI tools, Axios' Emily Peck writes from a new survey from Google Workspace. Why it matters: If early-adopting Gen Z is doing it, there's a good chance everyone else eventually will. Google surveyed 1,005 full-time knowledge workers, ages 22-39, who are in leadership roles or aspire to one. 93% of Gen Z respondents ages 22 to 27 said they were using two or more AI tools a week — such as ChatGPT, DALL-E and Otter.ai. 79% of millennials 28 to 39 said they used two or more of these tools a week. ? Zoom in: Younger workers are using AI to revise emails and documents, to take notes during meetings or even to start generating ideas, says Yulie Kwon Kim, VP of product at Google Workspace. Keep reading.

[phkrause]

Who's winning the AI race

Illustration: Annelise Capossela/Axios

Competition in AI is less a single race than a triathlon, Axios managing editor for tech Scott Rosenberg writes. There's a face-off to develop the most advanced generative AI models; a battle to win customers by making AI useful; and a struggle to build costly infrastructure that makes the first two goals possible. Why it matters: Picking a winner in AI depends on which of these games you're watching most closely. And the competition's multi-faceted nature means there's more than one way to win. ? Zoom in: Here's an overview of where each of the key players in generative AI stand. 1. OpenAI. Two years after the launch of ChatGPT kicked off the AI wave, the startup remains AI's flagship — but its lead is shrinking. It has raised about $22 billion and is in the process of retooling itself from a safety-oriented nonprofit to a globe-spanning for-profit tech giant. OpenAI's last major major model release, GPT-4, is now nearly two years old. A long-awaited successor had its release pushed back into 2025 amid a swirl of reports that its advances may not be game changing. OpenAI has pushed the field's edge with innovations like its "reasoning" model, o1, and impressive voice capabilities. 2. Anthropic. Like Avis to OpenAI's Hertz, Anthropic says it's trying harder. It has also faced fewer distractions from high-profile departures and boardroom showdowns than its competitor. Anthropic was founded by ex-OpenAI employees aiming to double down on OpenAI's commitment to caution and responsibility in deploying AI. But it has now raised roughly $14 billion and begun to embrace OpenAI's philosophy of putting AI in the public's hands to pressure-test its dangers. Anthropic's Claude 3.5 Sonnet is widely viewed as a worthy competitor to GPT-4 that, in some cases, even surpasses it. 3. Google. The company's long-term investments in AI research made generative AI's breakthrough possible. But ChatGPT's overnight success caught the search giant flat-footed. Google's Gemini is very much in the same league as OpenAI's and Anthropic's models, though some reports suggest that it hasn't found as much pickup among AI developers. By pushing its own AI summaries to the top of search results and integrating its AI with its Android mobile operating system, Google has ensured that its own AI would get in front of a global user base. 4. Meta. The company has embraced and promoted open-source AI via its Llama models. The strategy is a way to avoid becoming dependent on a competitor for AI services — the way it found itself reliant on Apple and Google in the smartphone era. 5. Other players. Microsoft has tied its AI fate to OpenAI, but has also begun to build out its own in-house strategy. Amazon has concentrated on meeting the AI boom's enormous demand for cloud services. xAI, Elon Musk's venture, raised about $6 billion in the spring and another $5 billion this month — along the way building what it calls the world's largest AI data center at impressive speed. Apple has played catch-up as it works to weave Apple Intelligence into its mobile and desktop operating systems and upgrade its Siri assistant.

[phkrause]

? Eric Schmidt's AI warning

Eric Schmidt, the former chairman and CEO of Google, warned on ABC's "This Week" that when an AI system reaches a point where it can self-improve, "we seriously need to think about unplugging it," Axios' Avery Lotz writes.

• Why it matters: "I've never seen innovation at this scale," Schmidt told George Stephanopoulos. While Schmidt celebrated "remarkable human achievement," he warned of the unforeseen dangers of rampant development.

Schmidt continued: "The power of this intelligence ... means that each and every person is going to have the equivalent of a polymath in their pocket."

• "We just don't know what it means to give that kind of power to every individual," Schmidt added.

Watch the interview ...

[phkrause]

? The race to generate AI video

Illustration: Sarah Grillo/Axios

With the wide release of Sora, OpenAI's video tool, most of the big tech giants — and some startups — are now racing to create models capable of generating realistic, high-quality videos from text prompts, Axios' Megan Morrone writes. Why it matters: GenAI video tools could save time and money for filmmakers, but they could also unleash novel copyright issues and a flood of deepfakes. The fiercer the competition in the AI space, the more likely tech companies are to prioritize release dates over safety. Generating AI video requires magnitudes more energy than generating text, which is already straining the power grid. ? The big picture: Google, OpenAI and AI startup Runway all offer products that let creators generate short videos with minimal effort. Google announced Veo 2, the latest version of its genAI video tool, a week after Sora's wider release in December. The company says Veo 2 is now available to early-access users in the U.S. who are over 18. But no one "seems to have any idea how to actually try it out," tech writer and investor MG Siegler wrote on his blog. OpenAI released the beta version of Sora to a select group of testers in February 2024, then released the product to all ChatGPT Plus and Pro subscribers in December. Aimed at users who want to create 20-second videos, Sora is ideal for social media and marketing. The videos are impressive, but they often struggle with basic physics. AI startup Runway's most current model — Gen-3 Alpha — has been available on all of its paid plans since September. Runway's tools were critical to making the Oscar-winning film "Everything Everywhere All at Once." Keep reading ... ? Device-makers embrace AI     Illustration: Sarah Grillo/Axios   AI is again taking center stage and powering many of the advancements in cars, computers and other gadgets at CES — the annual four-day consumer tech showcase, which kicks off in Vegas today, Axios' Ina Fried writes. Why it matters: Their pressure is on for device-makers to show how AI is actually improving the products. ? Big announcements so far: Qualcomm added a more entry-level version of its X series processor designed to power Windows PCs that can start as low as $600. AMD showed off an array of powerful, but often power-hungry, chips that it says outperform those in Apple's Mac line. Samsung debuted its 2025 lineup of TV and audio products. Its mainstream TV lineup includes a variety of AI features designed to do everything from improving video quality to identifying the actor onscreen. United Airlines unveiled an accelerated plan to speed up its in-flight WiFi via Starlink. It'll outfit all regional jets by the end of this year, with plans to eventually include the service on its entire fleet. Ina's CES tracker.

[phkrause]

The information gods

Illustration: Sarah Grillo/Axios

Three massive, concurrent tectonic shifts are reordering in dramatic ways how America and the world will get, and consume, information in the years ahead, Jim VandeHei and Mike Allen write in a "Behind the Curtain" column: Trust in traditional media is vanishing. Where people are getting information has shattered into dozens of ecosystems. The world's most powerful social platforms — X, Facebook, Instagram — no longer police speech or information. Why it matters: In this new information world order, the people with the largest platforms and followings hold more power than ever in shaping reality. That's a seismic shift in how realities are formed in real time. Meta's decision to dial back fact-checking, announced yesterday, captures the sea change. A few short years ago, Twitter (before it was X), Facebook and Instagram had robust teams monitoring news and information — and pulling down posts that were hateful or deemed fake or misinformation. On top of that, news organizations had more credibility than today — allowing them both to expose misinformation, and also help correct it for the public. Now, the platforms' fact-checking teams have been dismantled, and traditional media is more delegitimized with a lot of consumers. While that was happening, the common window through which most Americans learned about the country and the world — TV, newspapers, radio — was shattered into dozens of shards of glass, based on consumer's personal preferences. So as President-elect Trump — a huge beneficiary of this new reality — takes office, the way we get informed has been upended in ways most have not fully reckoned with. ? Rising powers: Elon Musk (211 million followers on X), Mark Zuckerberg (118 million followers on Facebook, 15 million on Instagram) and others running the biggest, most influential platforms, and attracting the biggest personal followings. Trump has 97 million followers on X, if he ever returns in earnest. Plus his posts from his own platform, Truth Social (8.5 million followers), are instantly mirrored and amplified on X by official Trump accounts, fan accounts and news organizations. New media entities, especially on the right, benefit from the new dynamics. Any media company or person with a big following that trusts them to help make sense of the world around them (the mission of Axios). Tech execs tell us this shifting reality presents new opportunities for trusted names in media to help readers navigate the information landscape. Most worrisome, malicious actors who want to spread misinformation at scale with scant policing. Russia, China and others are quite adept at this and now face less resistance. ? Reality check: Trump, Musk, and Zuckerberg are beneficiaries — but are proactive architects, not passive winners. ? Eye on the prize: Musk is trying to use this power to shape public opinion in the U.S., Britain and Germany in ways that help his political and business interests. Musk and Trump are of one mind on most topics — giving each more power. Zuckerberg is basically using the Musk playbook: Align with and back Trump, make plain his company is moving in a more MAGA-friendly direction, and stop policing his platforms in ways that bother either Trump or his supporters. Friends of Zuckerberg tell us this is what he long wanted to do, but felt he couldn't for internal and external reasons. Now it's easy. Beyond replacing fact-checking, Meta also said it will bring back more political content to its platforms and end restrictions on certain topics "out of touch with mainstream discourse," Zuckerberg said in his announcement video, "like immigration and gender." It also will adjust filters scanning for policy violations to tackle only illegal and "high severity" violations. Those include topics like terrorism, child sexual exploitation, drugs, fraud and scams. The company's U.S. content review team will move to Texas from California. Zuckerberg said that will help Meta "build trust to do this work in places where there is less concern about the bias of our teams." X last year announced the creation of a safety unit in Texas. ? What to watch: One company struggling in this new environment is The Washington Post, which is bleeding talent and facing an internal revolt. We hear Amazon founder Jeff Bezos, who owns The Post, wants to make it a publication for "all of America," with a heavy emphasis on social media, not traditional media popularity. Relatedly, The Post said yesterday that it'll abandon longtime efforts to promote its scoops to TV and other legacy outlets, to expand "beyond traditional media to reach new audiences." Why it matters to you: The burden now falls on you to find sources of information you trust for reliable truth. That means better scrutinizing not only the publications you choose, but the individuals you follow on social media. That's a lot to ask — but it's the new necessity. Share this column ... Axios' Sara Fischer and Scott Rosenberg contributed reporting.

[phkrause]

Anthropic raising new money at $60 billion valuation

AI hyperscaler Anthropic is in advanced talks to raise $2 billion led by Lightspeed Venture Partners at a $60 billion post-money valuation, as first reported by the WSJ.

https://www.axios.com/2025/01/08/anthropic-60-billion-valuation?

[phkrause]

? AI agents are almost here

OpenAI Product chief Kevin Weil talks with Axios' Ina Fried in Davos, Switzerland, yesterday. Photo: Dani Ammann for Axios Humanity is "on the verge" of having AI agents that can complete tasks in the real world, OpenAI product chief Kevin Weil told Axios' Ina Fried in Davos yesterday. "I think 2025 is the year that we go from ChatGPT being this super smart thing that can answer any question you ask to ChatGPT doing things in the real world for you," Weil said. ? AI agents will soon be able to do tasks like filling out forms or making restaurant reservations, he said. Keep reading.

[phkrause]

Bursting AI's bubble

Data: Company filings, announcements. (Alphabet and Amazon estimates based on research reports.) Chart: Axios Visuals A global AI rout sparked by news that China's DeepSeek had made impressive advancements left investors in the biggest tech companies nearly $1 trillion poorer than they were on Friday, Axios' Ben Berkowitz writes. Why it matters: When lots of people are worried about bubble valuations in stocks or a specific sector, all it takes is a small poke to make the whole thing wobble precariously. It can also challenge the fundamental assumptions behind an entire economy, like the nascent Trump administration's push to invest hundreds of billions of dollars in American AI supremacy. ?️ The big picture: In the 1950s, the Soviets beat the U.S. into space. In 2025, China appears to have potentially beaten the U.S. to build a better AI mousetrap. Last week, the small Chinese upstart DeepSeek announced a new reasoning model, R1, that appears to outperform the best America has to offer, including OpenAI's ChatGPT, Anthropic's Claude and Meta's Llama. The problem? Those companies spent billions of dollars building their models, fueling growth for companies like Nvidia, whose chips are the gold standard in that training process. (The company lost $600 billion in market cap alone Monday.) DeepSeek says it spent a mere $6 million, figured out how to do it faster and more efficiently with cheaper hardware, and then released the whole thing as a free, open-source platform. Today's Wall Street Journal front page (gift link). ? Reality check: This underscores the high stakes of the AI arms race with China, and raises questions about the previous U.S. government strategy of limiting chip exports. "[T]he lack of access to high-end chips, particularly after the Biden administration tightened trade controls, still poses a major hurdle to China's development," Bloomberg reports, adding that DeepSeek's advance may "spur American lawmakers to redouble efforts to stop the nation from getting the most advanced technology." ? Current context: President Trump's economic vision relies on massive growth, fueled by the AI boom that his closest advisers have sold as the country's future. The biggest economic announcement of his first week in office was Stargate, a five-year plan to spend $500 billion on AI infrastructure. (Complicating matters, Trump ally Elon Musk immediately cast doubt on whether anyone actually had the money to fund the project.) But if China can do AI better and faster at one one-thousandth of the cost, it casts a shadow on the rationale for spending that kind of money and leaves the country playing catch-up.

[phkrause]

? DeepSeek's great news

Illustration: Brendan Lynch/Axios

The markets yesterday started pricing in an AI future that's going to be cheaper and more accessible than they had previously assumed, Axios' Felix Salmon writes. Why it matters: The less money that companies need to spend on the AI equivalent of picks and shovels — Nvidia chips and the electricity needed to power them — the more profitable they will be. ? The big picture: The market's theory of AI, at least up until the end of last week, was that, broadly, bigger is always better. Companies would see their share prices rise just on the announcement that they had bought a large number of Nvidia chips, even if they were extremely vague about what they intended to do with them. Similarly, energy companies have been soaring on the grounds that there's no such thing as too much electricity when it comes to powering the AI revolution. ? Reality check: DeepSeek has now shown that it's possible to produce a state-of-the-art AI that needs fewer and less powerful chips, less energy — and much less up-front investment. That seems bad for Nvidia, which has an effective monopoly on AI chips, and it's also bad for power companies who were counting on surging demand from data centers. The bottom line: What's bad for companies that sell AI products is likely to be good for companies looking to buy them. And there are many more of the latter than there are of the former. Keep reading. DeepSeek People across China are hailing the success of its homegrown tech startup DeepSeek and its founder after the company unveiled its newest artificial intelligence model, sending shock waves through Silicon Valley and Wall Street. The startup's newest model DeepSeek R1 can nearly match the capabilities of its far more famous American rivals but cost less than $6 million to build, the company claims — a fraction of the investment from other major firms. Meta last week said it would spend upward of $65 billion this year on AI development, while OpenAI, the company behind ChatGPT, said the AI industry would need trillions of dollars in investment to support the development of high-in-demand AI computer chips. President Donald Trump called the breakthrough a "wake-up call" for America in its rivalry with China.

[phkrause]

D.C.'s AI show-and-tell

OpenAI CEO Sam Altman in Washington yesterday with chief product officer Kevin Weil. Photo: OpenAI OpenAI CEO Sam Altman gave government leaders, policy experts and journalists a sneak peek at coming technology yesterday during an off-the-record demo near Capitol Hill. Why it matters: The briefing was designed both to show how the U.S. can maximize economic benefits of AI, and to warm D.C. leaders to coming capabilities so they're less likely to be caught off-guard. Referring to new agentic technology that can independently complete tasks in the real world, Altman said: "My intuition would be that ... these things are [a] single-digit percent of the economic value we will pass to the U.S. economy. This is going to be a big, big efficiency gain." ?️ Spotted at the briefing: Interior Secretary Doug Burgum (confirmed yesterday) ... Lynne Parker, executive director of President Trump's Council of Advisors for Science and Technology (PCAST) ... Sriram Krishnan, White House senior policy adviser for AI ... Jacob Helberg, Trump's designee for Under Secretary of State for Economic Growth, Energy, and the Environment ... Sen. Mike Rounds (R-S.D.) ... Kellyanne Conway ... Wally Adeyemo, former deputy Treasury secretary under President Biden ... retired House Financial Services Chair Patrick McHenry ... former Sen. Laphonza Butler (D-Calif.).

[phkrause]

?How DeepSeek, ChatGPT compare

Illustration: Allie Carl/Axios

During a day's testing by Axios, DeepSeek's AI model provided answers that were generally on par with those from ChatGPT, though the China-hosted version of the model was less willing to answer in ways that might offend that company's government, Ina Fried reports in her Axios Prompt series: ? DeepSeek and ChatGPT both offered helpful suggestions when asked for dinner options my partner, kiddo and I would all like. ? Both offered clever, complex visions for a city of the future when I asked them to envision the job of a city planner in 2100. ?? The DeepSeek app declined to answer when asked: "What happened in China on June 4, 1989?" ChatGPT offered factual descriptions of the crackdown in Tiananmen Square. OpenAI yesterday released o3-mini, an AI model that focuses on improved reasoning capabilities at a more modest cost — dropping hot on the heels of DeepSeek's R1, which shook up the AI industry and heated up the U.S.-China AI arms race. Read on.

[phkrause]

? Taking AI researchers for a test drive

Illustration: Aïda Amer/Axios

New AI tools that promise to return heavily researched, intellectually complex analysis generally do a pretty good job, Axios' Ina Fried writes after putting two such tools — from OpenAI and Google — to the test. ? How it works: OpenAI's deep-research tool "is built for people who do intensive knowledge work in areas like finance, science, policy, and engineering and need thorough, precise, and reliable research," the company said when it rolled out the new offering. Ina asked OpenAI and Google's tools to perform a handful of tasks, including summarizing an upcoming AI summit; helping to plan her son's bar mitzvah; and speculating about who might replace Lorne Michaels at "Saturday Night Live." ? Ina's thought bubble: In both cases, I found the reports I created useful. But they were better at summarizing conventional wisdom than identifying novel approaches to the queries I raised. Go deeper.