I got some links for you

File under US-China tech cold war.

Anthropic to stop selling services to Chinese companies:

The policy, which takes effect immediately, will potentially apply to Chinese companies from ByteDance and Tencent to Alibaba. “We are taking action to close a loophole that allows Chinese companies to access frontier AI,” said the executive, who added that the policy would also apply to US adversaries including Russia, Iran and North Korea. The executive said the policy was designed “to align with our broader commitment that transformational AI capabilities advance democratic interests in US leadership in AI”. The shift reflects rising concerns in the US about Chinese groups setting up subsidiaries abroad in an effort to conceal their attempts to obtain American technology.

File under ironic.

OpenAI builds models that supposedly are meant to kill jobs and usher in Keynes’s vision of a 15-hour workweek. Now, the same company is building a jobs portal!

That’s why we’re working with a broad range of organizations—from major employers like Walmart and John Deere, to professional services firms like Boston Consulting Group and Accenture, to job search and hiring platform Indeed, to community organizations like the Texas Association of Business and the Bay Area Council, to state governments like the Delaware governor’s office—to help everyone take advantage of the opportunities that AI has to offer. And together, we’re focused on two big initiatives:

First, we’re working to build out the OpenAI Jobs Platform. 

If you’re a business looking to hire an AI-savvy employee, or you just need help with a specific task, finding the right person can be hit-or-miss. The OpenAI Jobs Platform will have knowledgeable, experienced candidates at every level, and opportunities for anyone looking to put their skills to use. And we’ll use AI to help find the perfect matches between what companies need and what workers can offer.

Ok, wait until ChatGPT replaces all the human employees and replaces all the job listings on the portal with its gazillion models.

Well, at least AI is creating one job—chief AI officer, whatever the hell that means.

Did ChatGPT save Google?

Much has changed since the end of the liability trial, but some things have not. Google remains the dominant firm in the relevant markets, with no existing rivals gaining significant market share and no new competitors entering the field.

However, artificial intelligence technologies, particularly generative AI (GenAI), may prove to be a game-changer. Tens of millions of people now use GenAI chatbots, such as ChatGPT, Perplexity, and Claude, to find information they previously would have searched for on the internet. While these chatbots are not yet close to replacing general search engines (GSEs), the industry anticipates that developers will continue to add features that make them function more like them. -— DC District Court Judge Amit Mehta

The emergence of GenAI has changed the course of this case. No witnesses at the liability trial testified that GenAI posed a near-term threat to GSEs. In contrast, the very first witness at the remedies hearing placed GenAI front and center as a nascent competitive threat.

Google doesn’t have to divest Chrome and also can continue paying for search placement. So Mozilla won’t die.

Mehta declined to grant some of the more ambitious proposals from the Justice Department to remedy Google’s behavior and restore competition to the market. Besides letting Google keep Chrome, he’ll also let the company continue to pay distribution partners for preloading or placement of its search or AI products. But he did order Google to share some valuable search information with rivals that could help jumpstart their ability to compete, and bar the search giant from making exclusive deals to distribute its search or AI assistant products in ways that might cut off distribution for rivals.

Ok, hold on, we are back on track for Keyne’s vision. DeepSeek is building a digital butler. Soon we all will know what it feels like Bruce Wayne having to boss around Alfred.

The Hangzhou-based startup is building an AI model that’s designed to carry out multi-step actions on a person’s behalf with minimal direction from the user, said the people. The system is also meant to learn and improve based on its prior actions, the people said, speaking on condition of anonymity because the information is private.

Switzerland apparently launched its own AI model, and it’s open source. File this under who gives a shit? What it should be building is infrastructure to attract the new-age money laundering and tax-evading autonomous AI agents.

It’s not just Switzerland; even Microsoft is building its own models. Well, who isn’t? More OpenAI drama ahead? Imma grab the popcorn. Hey ChatGpt, what popcorn…

AI isn’t killing search…yet.

Rand wrote:

My takeaway is that traditional search isn’t going anywhere, even for the heavy adopters of AI. The more data we gather, the more I’m convinced the “AI vs. Search” narrative is largely made-up by media and influencers seeking attention, rather than an accurate reflection of reality.

Rand also referenced new research from Semrush who said ChatGPT is not replacing Google Search.

The report looked at millions of US devices and showed that 95%+ remain regular users of traditional search engines. That’s down by a negligible <1% over the past 2.5 years, even as AI tool use has nearly quintupled (from 8% to 38% over the same period), Rand added.

Google to power Apple Siri? Hey Siri, you suck!

File this under “we are fucked!” Hope it’s not in the wrong orifice. That’s an anal joke, in case you are wondering. Anal = ass if you are further wondering.

The Pentagon is apparently in a rush to integrate AI. Well, who isn’t?

Last year Schneider, director of the Hoover Wargaming and Crisis Simulation Initiative at Stanford University, began experimenting with war games that gave the latest generation of artificial intelligence the role of strategic decision-makers. In the games, five off-the-shelf large language models or LLMs — OpenAI’s GPT-3.5, GPT-4, and GPT-4-Base; Anthropic’s Claude 2; and Meta’s Llama-2 Chat — were confronted with fictional crisis situations that resembled Russia’s invasion of Ukraine or China’s threat to Taiwan.

The results? Almost all of the AI models showed a preference to escalate aggressively, use firepower indiscriminately and turn crises into shooting wars — even to the point of launching nuclear weapons. “The AI is always playing Curtis LeMay,” says Schneider, referring to the notoriously nuke-happy Air Force general of the Cold War. “It’s almost like the AI understands escalation, but not de-escalation. We don’t really know why that is.”

File this under “hmmm, that’s interesting.”

Google is building a blockchain. That’s like my Dhobi guy offering management consulting services.

Google Cloud and derivatives marketplace CME Group said Tuesday (March 25) that they are piloting solutions for wholesale payments and tokenization of assets.

CME successfully completed the first phase of integration and testing of Google Cloud’s distributed ledger, Google Cloud Universal Ledger (GCUL), the companies said in a Tuesday press release.

The companies added that they will begin direct testing with market participants later this year and plan to launch new services in 2026.

The GCUL distributed ledger is designed to simplify the management of accounts and assets, facilitate transfers on a private and permissioned network, and be easily integrated and used by financial institutions in traditional finance, according to the release.

Gallup data on AI usage:

The use of AI at work is accelerating. In the past two years, the percentage of U.S. employees who say they have used AI in their role a few times a year or more has nearly doubled, from 21% to 40%. Frequent AI use (a few times a week or more) has also nearly doubled, from 11% to 19% since Gallup’s first measure in 2023. Daily use has doubled in the past 12 months alone, from 4% to 8%.

So 8% of the people are using the thing that will replace them, and the rest are busy watching Keeping up with the Kardashians?

Your replacement is getting cheaper and cheaper.

But.

Despite that drop in cost per token, what’s driving up costs for many AI applications is so-called reasoning. Many new forms of AI re-run queries to double-check their answers, fan out to the web to gather extra intel, even write their own little programs to calculate things, all before returning with an answer that can be as short as a sentence. And AI agents will carry out a lengthy series of actions based on user prompts, potentially taking minutes or even hours.

As a result, they deliver meaningfully better responses, but can spend a lot more tokens in the process. Also, when you give them a hard problem, they may just keep going until they get the answer, or fail trying.

Ok, so how long do we have until we all get fired by an “agent”? The worst part is you can’t even fucking punch it after getting fired by it.

Ok, good news: after firing our sorry asses, the agent will find us the best and most understanding and emotionally available human girlfriends and boyfriends. We all can sit on the couches and eat the last can of Pringles we can afford post-firing and get fat while the agent swipes right for us:

Since returning to Bumble in March after stepping down as chief executive in late 2023, she’s led a secret project to build a new, AI-powered matchmaking app that’s based on a surprising hypothesis. What if AI could understand us better than we understand ourselves? And what if that understanding gave it the ability to choose better matches for us than we might for ourselves?

Wolfe Herd and a team at Bumble have consulted expert psychologists and relationship counselors to program a matchmaking AI. They are looking to release a beta version of the app, which will be separate from the existing Bumble, to select users this fall.

The goal is to create “the world’s smartest and most emotionally intelligent matchmaker in existence,” Wolfe Herd said during an August interview on the back patio of a restaurant in Hotel Jerome, a luxurious, rustic hotel in Aspen, Colo. She splits time between Colorado and Texas.

Good only, no?

Ouch. These people got fired even before AI came for them:

Indian online gaming app Mobile Premier League (MPL) will sack about 60% of its local workforce as part of a major downsizing after the government banned paid games, said a company source with knowledge of the plan, in the first such reaction to a new law.

Prime Minister Narendra Modi’s government this month banned online paid games, citing financial and addiction risks especially among youth, leading to a shutdown of many gaming apps offering paid fantasy cricket, rummy and poker games

What’s left? We become massagers on Urban Clap? You want Thai massage?

Do you want to understand the agent that will fire your sorry ass and turn you into the person that says you want Thai massage?

Read this article.

Fucking Jane Street. They made $10 billion without selling trading seminars.

New York: Jane Street Group pulled in a record $10.1 billion in net trading revenue in the second quarter, beating out all of Wall Street’s biggest banks as the market-making giant reaped the benefit of trade war volatility.

Wait, so soon they’ll fill all traders, replace em with AI, make more money, come back to India, and fuck over Indian retail traders, but this time “powered by AI”?

This company makes videos in 40 milliseconds. Fun fact: it takes 30 to fire you.

To make real-time video possible, Mirage had to process video fast, blazingly so. The way to do this, Leitersdorf explains, is “writ[ing] lots of very optimized GPU code.” Most developers who work with NVIDIA GPUs use something called CUDA, a programming toolkit that makes it easier to tell the GPU what to do. But the Decart team went a level deeper. Instead of going through CUDA, they wrote code directly in PTX, the GPU’s assembly language. That’s the computer’s native tongue: harder for humans to write, but much faster for the hardware to understand. This decision let them squeeze every ounce of performance out of the hardware. It’s how Mirage can process video with only a 40-millisecond delay today—with plans to shrink that to just 16 milliseconds.