Author name: Tim Belzer

with-new-in-house-models,-microsoft-lays-the-groundwork-for-independence-from-openai

With new in-house models, Microsoft lays the groundwork for independence from OpenAI

AI, copilot, llm, MAI-1-preview, MAI-Voice-1, microsoft, Microsoft Copilot, openai / /

Since it’s hard to predict where this is all going, it’s likely to Microsoft’s long-term advantage to develop its own models.

It’s also possible Microsoft has introduced these models to address use cases or queries that OpenAI isn’t focused on. We’re seeing a gradual shift in the AI landscape toward models that are more specialized for certain tasks, rather than general, all-purpose models that are meant to be all things to all people.

These new models follow that somewhat, as Microsoft AI lead Mustafa Suleyman said in a podcast with The Verge that the goal here is “to create something that works extremely well for the consumer… my focus is on building models that really work for the consumer companion.”

As such, it makes sense that we’re going to see these models rolling out in Copilot, which is Microsoft’s consumer-oriented AI chatbot product. Of MAI-1-preview, the Microsoft AI blog post specifies, “this model is designed to provide powerful capabilities to consumers seeking to benefit from models that specialize in following instructions and providing helpful responses to everyday queries.”

So, yes, MAI-1-preview has a target audience in mind, but it’s still a general-purpose model since Copilot is a general-purpose tool.

MAI-Voice-1 is already being used in Microsoft’s Copilot Daily and Podcasts features. There’s also a Copilot Labs interface that you can visit right now to play around with it, giving it prompts or scripts and customizing what kind of voice or delivery you want to hear.

MA1-1-preview is in public testing on LMArena and will be rolled out to “certain text use cases within Copilot over the coming weeks.”

With new in-house models, Microsoft lays the groundwork for independence from OpenAI Read More »

windows-11-25h2-update-hits-its-last-stop-before-release-to-the-general-public

Windows 11 25H2 update hits its last stop before release to the general public

Tech, Windows 11, windows 11 24h2, windows 11 25h2 / /

Microsoft’s fifth major iteration of Windows 11 is nearing its release to the general public—the Windows Insider team announced today that Windows 11 25H2 was being put into its Release Preview Channel, the final stop for most updates before they become available to everyone. That’s around two months after the first Windows builds with the 25H2 label were released to the other preview channels.

Putting a new yearly Windows update in the Release Preview channel is analogous to the “release to manufacturing” (RTM) phase of years past, back when updates shipped on physical media that needed to be manufactured. Build numbers for this version of Windows start with 26200, rather than 24H2’s 26100.

The 25H2 update doesn’t do a lot in and of itself, other than reset the clock for Microsoft’s security updates (each yearly release gets two years of security patches). Microsoft says that last year’s 24H2 update and this year’s 25H2 update “use a shared servicing branch,” which mostly means that there aren’t big under-the-hood differences between the two. Installing the 25H2 update on a PC may enable some features on your 24H2 PC that had already been installed but had been disabled by default.

Microsoft says that installing the 25H2 update removes PowerShell 2.0 and the Windows Management Instrumentation Command-line tool (both previously deprecated), and that it allows IT administrators to automatically remove some preinstalled Windows apps from the Microsoft Store via Group Policy. But Microsoft hasn’t said much about major, user-facing new features that are unique to the 25H2 update. The 23H2 update from two years ago was a similarly quiet add-on for Windows 11 22H2.

Windows 11 25H2 update hits its last stop before release to the general public Read More »

700-piece-lego-g3-imac-design-faces-long-shot-odds-to-get-made,-but-i-still-want-one

700-piece Lego G3 iMac design faces long-shot odds to get made, but I still want one

Apple, iMac, lego, Tech / /

I don’t usually get too excited about user-submitted designs on the Lego Ideas website, especially when those ideas would require negotiating a license with another company—user-generated designs need to reach 10,000 supporters before Lego considers them for production, two pretty high bars to clear even without factoring in some other brand’s conditions and requests.

But I’m both intrigued and impressed by this Lego version of Apple’s old Bondi Blue G3 iMac that has been making the rounds today. Submitted by a user named terauma, the 700-plus-piece set comes complete with keyboard, hockey-puck mouse, a classic Mac OS boot screen, and cathode ray tubes and circuit boards visible through the set’s transparent blue casing (like the original iMac, it may cause controversy by excluding a floppy disk drive). The design has already reached 5,000 supporters, and it has 320 days left to reach the 10,000-supporter benchmark required to be reviewed by Lego.

With its personality-forward aesthetics and Jony Ive-led design, the original iMac was the first step down the path that led to blockbuster products like the iPod and iPhone. It was the company’s first all-new Mac design after CEO Steve Jobs returned to the company in the late ’90s, and while it lacked some features included in contemporary PCs, its tightly integrated design and ease of setup helped it stand out against the beige desktop PCs of the day. Today’s colorful Apple Silicon iMacs are clearly inspired by the original design.

700-piece Lego G3 iMac design faces long-shot odds to get made, but I still want one Read More »

google-warns-that-mass-data-theft-hitting-salesloft-ai-agent-has-grown-bigger

Google warns that mass data theft hitting Salesloft AI agent has grown bigger

AI, Biz & IT, chat agents, credential theft, Salesforce, salesloft drift, Security / /

Google is advising users of the Salesloft Drift AI chat agent to consider all security tokens connected to the platform compromised following the discovery that unknown attackers used some of the credentials to access email from Google Workspace accounts.

In response, Google has revoked the tokens that were used in the breaches and disabled integration between the Salesloft Drift agent and all Workspace accounts as it investigates further. The company has also notified all affected account holders of the compromise.

Scope expanded

The discovery, reported Thursday in an advisory update, indicates that a Salesloft Drift breach it reported on Tuesday is broader than previously known. Prior to the update, members of the Google Threat Intelligence Group said the compromised tokens were limited to Salesloft Drift integrations with Salesforce. The compromise of the Workspace accounts prompted Google to change that assessment.

“Based on new information identified by GTIG, the scope of this compromise is not exclusive to the Salesforce integration with Salesloft Drift and impacts other integrations,” Thursday’s update stated. “We now advise all Salesloft Drift customers to treat any and all authentication tokens stored in or connected to the Drift platform as potentially compromised.”

On Thursday, Salesloft’s security guidance page made no reference to the new information and instead continued to indicate that the breach affected only Drift integrations with Salesforce. Company representatives didn’t immediately respond to an email seeking confirmation of the Google finding.

Google warns that mass data theft hitting Salesloft AI agent has grown bigger Read More »

genetically,-central-american-mammoths-were-weird

Genetically, Central American mammoths were weird

ancient DNA, Biology, evolution, mammoths, mexico, Science, woolly mammoth / /

This led a Mexican-European research collaboration to get interested in finding DNA from elsewhere in the Columbian mammoth’s range, which extended down into Central America. The researchers focused on the Basin of Mexico, which is well south of where any woolly mammoths were likely to be found. While the warmer terrain generally tends to degrade DNA more quickly, the team had a couple of things working in its favor. To begin with, there were a lot of bones. The Basin of Mexico has been heavily built up over the centuries, and a lot of mammoth remains have been discovered, including over 100 individuals during the construction of Mexico City’s international airport.

In addition, the team focused entirely on the mitochondrial genome. In contrast to the two sets of chromosomes in each cell, a typical cell might have hundreds of mitochondria, each of which could have dozens of copies of its genome. So, while the much smaller mitochondria don’t provide as much detail about ancestry, they’re at least likely to survive at high enough levels to provide something to work with.

And indeed they did. Altogether, the researchers obtained 61 new mitochondrial genomes from the mammoths of Mexico from the 83 samples they tested. Of these, 28 were considered high enough quality to perform an analysis.

Off on their own

By building a family tree using this genetic data, along with that from other Columbian and woolly mammoth samples, the researchers could potentially determine how different populations were related. And one thing became very clear almost immediately: They were in a very weird location on that tree.

To begin with, all of them clustered together in a single block, although there were three distinct groupings within that block. But the placement of that block within the larger family tree was notably strange. To begin with, there were woolly mammoths on either side of it, suggesting the lineage was an offshoot of woolly mammoths. That would make sense if all Columbian mammoths clustered together with the Mexican ones. But they don’t. Some Columbian mammoths from much farther north are actually more closely related to woolly mammoths than they are to the Mexican mammoths.

Genetically, Central American mammoths were weird Read More »

cdc-slashed-food-safety-surveillance,-now-tracks-only-2-of-8-top-infections

CDC slashed food safety surveillance, now tracks only 2 of 8 top infections

CDC, food safety, foodborne, foodnet, health, infections / /

In July, the Centers for Disease Control and Prevention dramatically, but quietly, scaled back a food safety surveillance system, cutting active tracking from eight top foodborne infections down to just two, according to a report by NBC News.

The Foodborne Diseases Active Surveillance Network (FoodNet)—a network of surveillance sites that spans 10 states and covers about 54 million Americans (16 percent of the US population)—previously included active monitoring for eight infections from pathogens. Those include Campylobacter, Cyclospora, Listeria, Salmonella, Shiga toxin-producing E. coli (STEC), Shigella, Vibrio, and Yersinia.

Now the network is only monitoring for STEC and Salmonella.

A list of talking points the CDC sent the Connecticut health department (which is part of FoodNet) suggested that a lack of funding is behind the scaleback. “Funding has not kept pace with the resources required to maintain the continuation of FoodNet surveillance for all eight pathogens,” the CDC document said, according to NBC. The Trump administration has made brutal cuts to federal agencies, including the CDC, which has lost hundreds of employees this year.

A CDC spokesperson told the outlet that “Although FoodNet will narrow its focus to Salmonella and STEC, it will maintain both its infrastructure and the quality it has come to represent. Narrowing FoodNet’s reporting requirements and associated activities will allow FoodNet staff to prioritize core activities.”

CDC slashed food safety surveillance, now tracks only 2 of 8 top infections Read More »

chris-roberts-hopes-squadron-42-will-be-“almost-as-big”-as-gta-vi-next-year

Chris Roberts hopes Squadron 42 will be “almost as big” as GTA VI next year

gaming / /

The long and winding road

It’s hard to remember now, but Star Citizen‘s then-impressive $6.3 million Kickstarter campaign came just a few months before Grand Theft Auto V first launched on the PlayStation 3 and Xbox 360 (remember those?). But development on Rockstar’s long-awaited sequel didn’t start in earnest until 2020, publisher Take Two says, around the time Star Citizen developer Roberts Space Industries was settling a contentious lawsuit over game engine rights and rolling out a new development roadmap for the game.

A graph visualizing the growing crowdfunding for Star Citizen from 2012 (top) through 2022 (bottom).

A graph visualizing the growing crowdfunding for Star Citizen from 2012 (top) through 2022 (bottom). Credit: Reddit / Rainbowles

Of course, the development of Grand Theft Auto VI has happened completely behind closed doors, with developer Rockstar and publisher Take Two only occasionally offering tiny drops of information to a desperate press and fan base. By contrast, Roberts Space Industries has issued regular, incredibly detailed information dumps on the drawn-out development progress for Star Citizen and Squadron 42, even when that kind of openness has contributed to the public appearance of internal dysfunction.

The massive, ongoing crowdfunding that powers the open development structure “allows us to do things without imposing the framework of a typical video game studio,” Roberts told La Presse. “The players who fund us expect the best game, period. We don’t have to streamline, cut jobs, or change our business model.”

That pre-launch development cycle must eventually end, of course, and the La Presse report suggests that the full 1.0 release of Star Citizen is “now promised” for “2027 or 2028.” While we’d love to believe that, the history of Star Citizen development thus far (and the lack of any provided sourcing for the claim) makes us more than a little skeptical.

Chris Roberts hopes Squadron 42 will be “almost as big” as GTA VI next year Read More »

lawmaker:-trump’s-golden-dome-will-end-the-madness,-and-that’s-not-a-good-thing

Lawmaker: Trump’s Golden Dome will end the madness, and that’s not a good thing

congress, Features, golden dome, House of Representatives, military space, missile defense, Science, seth moulton, Space, space weapons / /

“The underlying issue here is whether US missile defense should remain focused on the threat from rogue states and… accidental launches, and explicitly refrain from countering missile threats from China or Russia,” DesJarlais said. He called the policy of Mutually Assured Destruction “outdated.”

President Donald Trump speaks alongside Secretary of Defense Pete Hegseth in the Oval Office at the White House on May 20, 2025, in Washington, DC. President Trump announced his plans for the Golden Dome, a national ballistic and cruise missile defense system. Credit: Chip Somodevilla/Getty Images

Moulton’s amendment on nuclear deterrence failed to pass the committee in a voice vote, as did another Moulton proposal that would have tapped the brakes on developing space-based interceptors.

But one of Moulton’s amendments did make it through the committee. This amendment, if reconciled with the Senate, would prohibit the Pentagon from developing a privatized or subscription-based missile defense intercept capability. The amendment says the US military can own and operate such a system.

Ultimately, the House Armed Services Committee voted 55–2 to send the NDAA to a vote on the House floor. Then, lawmakers must hash out the differences between the House version of the NDAA with a bill written in the Senate before sending the final text to the White House for President Trump to sign into law.

More questions than answers

The White House says the missile shield will cost $175 billion over the next three years. But that’s just to start. A network of space-based missile sensors and interceptors, as prescribed in Trump’s executive order, will eventually number thousands of satellites in low-Earth orbit. The Congressional Budget Office reported in May that the Golden Dome program may ultimately cost up to $542 billion over 20 years.

The problem with all of the Golden Dome cost estimates is that the Pentagon has not settled on an architecture. We know the system will consist of a global network of satellites with sensors to detect and track missile launches, plus numerous interceptors in orbit to take out targets in space and during their “boost phase” when they’re moving relatively slowly through the atmosphere.

The Pentagon will order more sea- and ground-based interceptors to destroy missiles, drones, and aircraft as they near their targets within the United States. All of these weapons must be interconnected with a sophisticated command and control network that doesn’t yet exist.

Will Golden Dome’s space-based interceptors use kinetic kill vehicles to physically destroy missiles targeting the United States? Or will the interceptors rely on directed energy weapons like lasers or microwave signals to disable their targets? How many interceptors are actually needed?

These are all questions without answers. Despite the lack of detail, congressional Republicans approved $25 billion for the Pentagon to get started on the Golden Dome program as part of the Trump-backed One Big Beautiful Bill Act. The bill passed Congress with a party-line vote last month.

Israel’s Iron Dome aerial defense system intercepts a rocket launched from the Gaza Strip on May 11, 2021. Credit: Jack Guez/AFP via Getty Images

Moulton earned a bachelor’s degree in physics and master’s degrees in business and public administration from Harvard University. He served as a Marine Corps platoon leader in Iraq and was part of the first company of Marines to reach Baghdad during the US invasion of 2003. Moulton ran for the Democratic presidential nomination in 2020 but withdrew from the race before the first primary contest.

The text of our interview with Moulton is published below. It is lightly edited for length and clarity.

Ars: One of your amendments that passed committee would prevent the DoD from using a subscription or pay-for-service model for the Golden Dome. What prompted you to write that amendment?

Moulton: There were some rumors we heard that this is a model that the administration was pursuing, and there was reporting in mid-April suggesting that SpaceX was partnering with Anduril and Palantir to offer this kind of subscription service where, basically, the government would pay to access the technology rather than own the system. This isn’t an attack on any of these companies or anything. It’s a reassertion of the fundamental belief that these are responsibilities of our government. The decision to engage an intercontinental ballistic missile is a decision that the government must make, not some contractors working at one of these companies.

Ars: Basically, the argument you’re making is that war-fighting should be done by the government and the armed forces, not by contractors or private companies, right?

Moulton: That’s right, and it’s a fundamental belief that I’ve had for a long time. I was completely against contractors in Iraq when I was serving there as a younger Marine, but I can’t think of a place where this is more important than when you’re talking about nuclear weapons.

Ars: One of the amendments that you proposed, but didn’t pass, was intended to reaffirm the nation’s strategy of nuclear deterrence. What was the purpose of this amendment?

Moulton: Let’s just start by saying this is fundamentally why we have to have a theory that forms a foundation for spending hundreds of billions of taxpayer dollars. Golden Dome has no clear design, no real cost estimate, and no one has explained how this protects or enhances strategic stability. And there’s a lot of evidence that it would make strategic stability worse because our adversaries would no longer have confidence in Mutual Assured Destruction, and that makes them potentially much more likely to initiate a strike or overreact quickly to some sort of confrontation that has the potential to go nuclear.

In the case of the Russians, it means they could activate their nuclear weapon in space and just take out our Golden Dome interceptors if they think we might get into a nuclear exchange. I mean, all these things are horrific consequences.

Like I said in our hearing, there are two explanations for Golden Dome. The first is that every nuclear theorist for the last 75 years was wrong, and thank God, Donald Trump came around and set us right because in his first administration and every Democratic and Republican administration, we’ve all been wrong—and really the future of nuclear deterrence is nuclear defeat through defense and not Mutually Assured Destruction.

The other explanation, of course, is that Donald Trump decided he wants the golden version of something his friend has. You can tell me which one’s more likely, but literally no one has been able to explain the theory of the case. It’s dangerous, it’s wasteful… It might be incredibly dangerous. I’m happy to be convinced that Golden Dome is the right solution. I’m happy to have people explain why this makes sense and it’s a worthwhile investment, but literally nobody has been able to do that. If the Russians attack us… we know that this system is not going to be 100 percent effective. To me, that doesn’t make a lot of sense. I don’t want to gamble on… which major city or two we lose in a scenario like that. I want to prevent a nuclear war from happening.

Several Chinese DF-5B intercontinental ballistic missiles, each capable of delivering up to 10 independently maneuverable nuclear warheads, are seen during a parade in Beijing on September 3, 2015. Credit: Xinhua/Pan Xu via Getty Images

Ars: What would be the way that an administration should propose something like the Golden Dome? Not through an executive order? What process would you like to see?

Moulton: As a result of a strategic review and backed up by a lot of serious theory and analysis. The administration proposes a new solution and has hearings about it in front of Congress, where they are unafraid of answering tough questions. This administration is a bunch of cowards who can who refuse to answer tough questions in Congress because they know they can’t back up their president’s proposals.

Ars: I’m actually a little surprised we haven’t seen any sort of architecture yet. It’s been six months, and the administration has already missed a few of Trump’s deadlines for selecting an architecture.

Moulton: It’s hard to develop an architecture for something that doesn’t make sense.

Ars: I’ve heard from several retired military officials who think something like the Golden Dome is a good idea, but they are disappointed in the way the Trump administration has approached it. They say the White House hasn’t stated the case for it, and that risks politicizing something they view as important for national security.

Moulton: One idea I’ve had is that the advent of directed energy weapons (such as lasers and microwave weapons) could flip the cost curve and actually make defense cheaper than offense, whereas in the past, it’s always been cheaper to develop more offensive capabilities rather than the defensive means to shoot at them.

And this is why the Anti-Ballistic Missile Treaty in the early 1970s was so effective, because there was this massive arms race where we were constantly just creating a new offensive weapon to get around whatever defenses our adversary proposed. The reason why everyone would just quickly produce a new offensive weapon before that treaty was put into place is because it was easy to do.

My point is that I’ve even thrown them this bone, and I’m saying, ‘Here, maybe that’s your reason, right?” And they just look at me dumbfounded because obviously none of them are thinking about this. They’re just trying to be lackeys for the president, and they don’t recognize how dangerous that is.

Ars: I’ve heard from a chorus of retired and even current active duty military leaders say the same thing about directed energy weapons. You essentially can use one platform in space take take numerous laser shots at a missile instead of expending multiple interceptors for one kill.

Moulton: Yes, that’s basically the theory of the case. Now, my hunch is that if you actually did the serious analysis, you would determine that it still decreases state strategic stability. So in terms of the overall safety and security of the United States, whether it’s directed energy weapons or kinetic interceptors, it’s still a very bad plan.

But I’m even throwing that out there to try to help them out here. “Maybe this is how you want to make your case.” And they just look at me like deer in the headlights because, obviously, they’re not thinking about the national security of the United States.

Ars: I also wanted to ask about the Space Force’s push to develop weapons to use against other satellites in orbit. They call these counter-space capabilities. They could be using directed energy, jamming, robotic arms, anti-satellite missiles. This could take many different forms, and the Space Force, for the first time, is talking more openly about these issues. Are these kinds of weapons necessary, in your view, or are they too destabilizing?

Moulton: I certainly wish we could go back to a time when the Russians and Chinese were not developing space weapons—or were not weaponizing space, I should say, because that was the international agreement. But the reality of the world we live in today is that our adversaries are violating that agreement. We have to be prepared to defend the United States.

Ars: Are there any other space policy issues on your radar or things you have concerns about?

Moulton: There’s a lot. There’s so much going on with space, and that’s the reason I chose this subcommittee, even though people would expect me to serve on the subcommittee dealing with the Marine Corps, because I just think space is incredibly important. We’re dealing with everything from promotion policy in the Space Force to acquisition reform to rules of engagement, and anything in between. There’s an awful lot going on there, but I do think that one of the most important things to talk about right now is how dangerous the Golden Dome could be.

Lawmaker: Trump’s Golden Dome will end the madness, and that’s not a good thing Read More »

trump-admin-issues-stop-work-order-for-offshore-wind-project

Trump admin issues stop-work order for offshore wind project

Energy, Science, Trump, Wind / /

In a statement to Politico’s E&E News days after the order was lifted in May, the White House claimed that Hochul “caved” and struck an agreement to allow “two natural gas pipelines to advance” through New York.

Hochul denied that any such deal was made.

Trump has made no effort to conceal his disdain for wind power and other renewable energies, and his administration has actively sought to stymie growth in the industry while providing what critics have described as “giveaways” to fossil fuels.

In a Truth Social post on Wednesday, Trump called wind and solar energy the “SCAM OF THE CENTURY,” criticizing states that have built and rely on them for power.

“We will not approve wind or farmer destroying Solar,” Trump wrote. “The days of stupidity are over in the USA!!!”

On Trump’s first day in office, the president issued a memorandum halting approvals, permits, leases, and loans for both offshore and onshore wind projects.

The GOP also targeted wind energy in the One Big Beautiful Bill Act, accelerating the phaseout of tax credits for wind and solar projects while mandating lease sales for fossil fuels and making millions of acres of federal land available for mining.

The administration’s subsequent consideration of rules to further restrict access to tax credits for wind and solar projects alarmed even some Republicans, prompting Iowa Sen. Chuck Grassley and Utah Sen. John Curtis to place holds on Treasury nominees as they awaited the department’s formal guidance.

Those moves have rattled the wind industry and created uncertainty about the viability of ongoing and future projects.

“The unfortunate message to investors is clear: the US is no longer a reliable place for long-term energy investments,” said the American Clean Power Association, a trade association, in a statement on Friday.

To Kathleen Meil, local clean energy deployment director at the League of Conservation Voters, that represents a loss not only for the environment but also for the US economy.

“It’s really easy to think about the visible—the 4,200 jobs across all phases of development that you see… They’ve hit more than 2 million union work hours on Revolution Wind,” Meil said.

“But what’s also really transformational is that it’s already triggered $1.3 billion in investment through the supply chain. So it’s not just coastal communities that are benefiting from these jobs,” she said.

“This hurts so many people. And why? There’s just no justification.”

This article originally appeared on Inside Climate News, a nonprofit, non-partisan news organization that covers climate, energy, and the environment. Sign up for their newsletter here.

Trump admin issues stop-work order for offshore wind project Read More »

reports-of-ai-not-progressing-or-offering-mundane-utility-are-often-greatly-exaggerated

Reports Of AI Not Progressing Or Offering Mundane Utility Are Often Greatly Exaggerated

Reports / /

In the wake of the confusions around GPT-5, this week had yet another round of claims that AI wasn’t progressing, or AI isn’t or won’t create much value, and so on. There were reports that one study in particular impacted Wall Street, and as you would expect it was not a great study. Situational awareness is not what you’d hope.

I’ve gathered related coverage here, to get it out of the way before whatever Google is teasing (Gemini 3.0? Something else?) arrives to potentially hijack our attention.

We’ll start with the MIT study on State of AI in Business, discuss the recent set of ‘AI is slowing down’ claims as part of the larger pattern, and then I will share a very good attempted explanation from Steven Byrnes of some of the ways economists get trapped into failing to look at what future highly capable AIs would actually do.

Chatbots and coding agents are clear huge wins. Over 80% of organizations have ‘explored or piloted’ them and 40% report deployment. The employees of the other 60% presumably have some news.

But we have a new State of AI in Business report that says that when businesses try to do more than that, ‘95% of businesses get zero return,’ although elsewhere they say ‘only 5% custom enterprise AI tools reach production.’

From our interviews, surveys, and analysis of 300 public implementations, four patterns emerged that define the GenAI Divide:

  1. Limited disruption: Only 2 of 8 major sectors show meaningful structural change.

  2. Enterprise paradox: Big firms lead in pilot volume but lag in scale-up.

  3. Investment bias: Budgets favor visible, top-line functions over high-ROI back office.

  4. Implementation advantage: External partnerships see twice the success rate of

    internal builds.

These are early days. Enterprises have only had capacity to look for ways to slide AI directly into existing structures. They ask, ‘what that we already do, can AI do for us?’ They especially ask ‘what can show clear measurable gains we can trumpet?’

It does seem reasonable to say that the ‘custom tools’ approach may not be doing so great, if the tools only reach deployment 5% of the time. They might have a high enough return they still come out ahead, but that is a high failure rate if you actually fully scrap the other 95% and don’t learn from them. It seems like this is a skill issue?

The primary factor keeping organizations on the wrong side of the GenAI Divide is the learning gap, tools that don’t learn, integrate poorly, or match workflows.

…

The 95% failure rate for enterprise AI solutions represents the clearest manifestation of the GenAI Divide. Organizations stuck on the wrong side continue investing in static tools that can’t adapt to their workflows, while those crossing the divide focus on learning-capable systems.

…

As one CIO put it, “We’ve seen dozens of demos this year. Maybe one or two are genuinely useful. The rest are wrappers or science projects.”

That sounds like the ‘AI tools’ that fail deserve the air quotes.

I also note that later they say custom built AI solutions ‘fail twice as often.’ That implies that when companies are wise enough to test solutions built externally, they succeed over 50% of the time.

There’s also a strange definition of ‘zero return’ here.

Just 5% of integrated AI pilots are extracting millions in value, while the vast majority remain stuck with no measurable P&L impact.

Tools like ChatGPT and Copilot are widely adopted. Over 80 percent of organizations have explored or piloted them, and nearly 40 percent report deployment. But these tools primarily enhance individual productivity, not P&L performance.

Issue a report where you call the 95% of projects that don’t have ‘measurable P&L impact’ failures, then wonder why no one wants to do ‘high-ROI back office’ upgrades.

Those projects are high ROI, but how do you prove the R on I?

Especially if you can’t see the ROI on ‘enhancing individual productivity’ because it doesn’t have this ‘measurable P&L impact.’ If you double the productivity of your coders (as an example), it’s true that you can’t directly point to [$X] that this made you in profit, but surely one can see a lot of value there.

They call it a ‘divide’ because it takes a while to see returns, after which you see a lot.

While most implementations don’t drive headcount reduction, organizations that have crossed the GenAI Divide are beginning to see selective workforce impacts in customer support, software engineering, and administrative functions.

In addition, the highest performing organizations report measurable savings from reduced BPO spending and external agency use, particularly in back-office operations.

Others cite improved customer retention and sales conversion through automated outreach and intelligent follow-up systems.

These early results suggest that learning-capable systems, when targeted at specific processes, can deliver real value, even without major organizational restructuring.

This all sounds mostly like a combination of ‘there is a learning curve that is barely started on’ with ‘we don’t know how to measure most gains.’

Also note the super high standard here. Only 22% of major sectors show ‘meaningful structural change’ at this early stage, and section 3 talks about ‘high adoption, low transformation.’

Or their ‘five myths about GenAI in the Enterprise’:

  1. AI Will Replace Most Jobs in the Next Few Years → Research found limited layoffs from GenAI, and only in industries that are already affected significantly by AI. There is no consensus among executives as to hiring levels over the next 3-5 years.

  2. Generative AI is Transforming Business → Adoption is high, but transformation is rare. Only 5% of enterprises have AI tools integrated in workflows at scale and 7 of 9 sectors show no real structural change.

  3. Enterprises are slow in adopting new tech → Enterprises are extremely eager to adopt AI and 90% have seriously explored buying an AI solution.

  4. The biggest thing holding back AI is model quality, legal, data, risk → What’s really holding it back is that most AI tools don’t learn and don’t integrate well into workflows.

  5. The best enterprises are building their own tools → Internal builds fail twice as often.

Most jobs within a few years is not something almost anyone is predicting in a non-AGI world. Present tense ‘transforming business’ is a claim I don’t remember hearing. I also hadn’t heard ‘the best enterprises are building their own tools’ and it does not surprise me that rolling your own comes with much higher failure rates.

I would push back on #3. As always, slow is relative, and being ‘eager’ is very different from not being the bottleneck. ‘Explored buying an AI solution’ is very distinct from ‘adopting new tech.’

I would also push back on #4. The reason AI doesn’t yet integrate well into workflows is because the tools are not yet good enough. This also shows the mindset that the AI is being forced to ‘integrate into workflows’ rather than generating new workflows, another sign that they are slow in adopting new tech.

Users prefer ChatGPT for simple tasks, but abandon it for mission-critical work due to its lack of memory. What’s missing is systems that adapt, remember, and evolve, capabilities that define the difference between the two sides of the divide.

I mean ChatGPT does now have some memory and soon it will have more. Getting systems to remember things is not all that hard. It is definitely on its way.

The more I explore the report the more it seems determined to hype up this ‘divide’ around ‘learning’ and memory. Much of the time seems like unrealistic expectations.

Yes, you would love if your AI tools learned all the detailed preferences and contexts of all of your clients without you having to do any work?

The same lawyer who favored ChatGPT for initial drafts drew a clear line at sensitive contracts:

“It’s excellent for brainstorming and first drafts, but it doesn’t retain knowledge of client preferences or learn from previous edits. It repeats the same mistakes and requires extensive context input for each session. For high-stakes work, I need a system that accumulates knowledge and improves over time.”

This feedback points to the fundamental learning gap that keeps organizations on the wrong side of the GenAI Divide.

Well, how would it possibly know about client preferences or learn from previous edits? Are you keeping a detailed document with the client preferences in preferences.md? People would like AI to automagically do all sorts of things out of the box without putting in the work.

And if they wait a few years? It will.

I totally do not get where this is coming from:

Takeaway: The window for crossing the GenAI Divide is rapidly closing. Enterprises are locking in learning-capable tools. Agentic AI and memory frameworks (like NANDA and MCP) will define which vendors help organizations cross the divide versus remain trapped on the wrong side.

Why is there a window and why is it closing?

I suppose one can say ‘there is a window because you will rapidly be out of business’ and of course one can worry about the world transforming generally, including existential risks. But ‘crossing the divide’ gets easier every day, not harder.

In the next few quarters, several enterprises will lock in vendor relationships that will be nearly impossible to unwind.

Why do people keep saying versions of this? Over time increasingly capable AI and better AI tools will make it, again, easier not harder to pivot or migrate.

Yes, I get that people think the switching costs will be prohibitive. But that’s simply not true. If you already have an AI that can do things for your business, getting another AI to learn and copy what you need will be relatively easy. Code bases can switch between LLMs easily, often by changing only one to three lines.

What is the bottom line?

This seems like yet another set of professionals putting together a professional-looking report that fundamentally assumes AI will never improve, or that improvements in frontier AI capability will not matter, and reasoning from there. Once you realize this implicit assumption, a lot of the weirdness starts making sense.

Ethan Mollick: Okay, got the report. I would read it yourself. I am not sure how generalizable the findings are based on the methodology (52 interviews, convenience sampled, failed apparently means no sustained P&L impact within six months but no coding explanation)

I have no doubt pilot failures are high, but I think it is really hard to see how this report gives the kind of generalizable finding that would move markets.

Nathan Whittemore: Also no mention of coding. Also no agents. Also 50% of uses were marketing, suggesting extreme concentration of org area.

Azeem Azhar: it was an extremely weak report. You are very generous with your assessment.

Aaron Erickson: Many reports like this start from the desired conclusion and work backwards, and this feels like no exception to that rule.

Most of the real work is bottom up adoption not measured by anything. If anything, it is an indictment about top-down initiatives.

The reason this is worth so much attention is that we have reactions like this one from Matthew Field, saying this is a ‘warning sign the AI bubble is about to burst’ and claiming the study caused a stock selloff, including a 3.5% drop in Nvidia and ~1% in some other big tech stocks. Which isn’t that much, and there are various alternative potential explanations.

The improvements we are seeing involve not only AI as it exists now (as in the worst it will ever be), with substantial implementation delays. It also involves only individuals adopting AI or at best companies slotting AI into existing workflows.

Traditionally the big gains from revolutionary technologies come elsewhere.

Roon: real productivity gains for prior technological advances came not from individual workers learning to use eg electricity, the internet but entire workflows, factories, processes, businesses being set up around the use of new tools (in other words, management)

couple years ago I figured this could go much faster than usual thanks to knowledge diffusion over the internet and also the AIs themselves coming up with great ideas about how to harness their strengths and weaknesses. but I’m not sure about that at present moment.

Patrick McKenzie: (Agree with this, and generally think it is one reasons why timelines to visible-in-GDP-growth impact are longer than people similarly bullish on AI seem to believe.)

I do think it is going faster and will go faster, except that in AI the standard for ‘fast’ is crazy fast, and ‘AIs coming up with great ideas’ is a capability AIs are only now starting to approach in earnest.

I do think that if AGI and ASI don’t show up, the timeline to the largest visible-in-GDP gains will take a while to show up. I expect visible-in-GDP soon anyway because I think the smaller, quicker version of even the minimally impressive version of AI should suffice to become visible in GDP, even though GDP will only reflect a small percentage of real gains.

The ‘AI is losing steam’ or ‘big leaps are slowing down’ and so on statements from mainstream media will keep happening whenever someone isn’t feeling especially impressed this particular month. Or week.

Dean Ball: I think we live in a perpetual state of traditional media telling us that the pace of ai progress is slowing

These pieces were published during a span that I would describe as the most rapid pace of progress I’ve ever witnessed in LLMs (GPT-4 Turbo -> GPT 5-Pro; remember: there were no public reasoner models 365 days ago!)

(Also note that Bloomberg piece was nearly simultaneous with the announcement of o3, lmao)

Miles Brundage: Notably, it’s ~never employees at frontier companies quoted on this, it’s the journalists themselves, or academics, startups pushing a different technique, etc.

The logic being “people at big companies are biased.” Buddy, I’ve got some big news re: humans.

Anton: my impression is that articles like this mostly get written by people who really really want to believe ai is slowing down. nobody working on it or even using it effectively thinks this. Which is actually basically a marketing problem which the entire field has been bad at since 2022.

Peter Gostev: I’m sure you’ve all noticed the ‘AI is slowing down’ news stories every few weeks for multiple years now – so I’ve pulled a tracker together to see who and when wrote these stories.

There is quite a range, some are just outright wrong, others point to a reasonable limitation at the time but missing the bigger arc of progress.

All of these stories were appearing as we were getting reasoning models, open source models, increasing competition from more players and skyrocketing revenue for the labs.

Peter links to about 35 posts. They come in waves.

The practical pace of AI progress continues to greatly exceed the practical pace of progress everywhere else. I can’t think of an exception. It is amazing how eagerly everyone looks for a supposed setback to try and say otherwise.

You could call this gap a ‘marketing problem’ but the US Government is in the tank for AI companies and Nvidia is 3% of total stock market cap and investments in AI are over 1% of GDP and so on, and diffusion is proceeding at record pace. So it is not clear that they should care about those who keep saying the music is about to stop?

Coinbase CEO fires software engineers who don’t adopt AI tools. Well, yeah.

On the one hand, AI companies are building their models on the shoulders of giants, and by giants we mean all of us.

Ezra Klein (as an example): Right now, the A.I. companies are not making all that much money off these products. If they eventually do make the profits their investors and founders imagine, I don’t think the normal tax structure is sufficient to cover the debt they owe all of us, and everyone before us, on whose writing and ideas their models are built.

Then there’s the energy demand.

Also the AI companies are risking all our lives and control over the future.

On the other hand, notice that they are indeed not making that much money. It seems highly unlikely that, even in terms of unit economics, creators of AI capture more than 10% of value created. So in an ‘economic normal’ situation where AI doesn’t ‘go critical’ or transform the world, but is highly useful, who owes who the debt?

It’s proving very useful for a lot of people.

Ezra Klein: And yet I am a bit shocked by how even the nascent A.I. tools we have are worming their way into our lives — not by being officially integrated into our schools and workplaces but by unofficially whispering in our ears.

The American Medical Association found that two in three doctors are consulting with A.I.

A Stack Overflow survey found that about eight in 10 programmers already use A.I. to help them code.

The Federal Bar Association found that large numbers of lawyers are using generative A.I. in their work, and it was more common for them to report they were using it on their own rather than through official tools adopted by their firms. It seems probable that Trump’s “Liberation Day” tariffs were designed by consulting a chatbot.

All of these uses involve paying remarkably little and realizing much larger productivity gains.

Steven Byrnes explains his view on some reasons why an economics education can make you dumber when thinking about future AI, difficult to usefully excerpt and I doubt he’d mind me quoting it in full.

I note up top that I know not all of this is technically correct, it isn’t the way I would describe this, and of course #NotAllEconomists throughout especially for the dumber mistakes he points out, but the errors actually are often pretty dumb once you boil them down, and I found Byrnes’s explanation illustrative.

Steven Byrnes: There’s a funny thing where economics education paradoxically makes people DUMBER at thinking about future AI. Econ textbooks teach concepts & frames that are great for most things, but counterproductive for thinking about AGI. Here are 4 examples. Longpost:

THE FIRST PIECE of Econ anti-pedagogy is hiding in the words “labor” & “capital”. These words conflate a superficial difference (flesh-and-blood human vs not) with a bundle of unspoken assumptions and intuitions, which will all get broken by Artificial General Intelligence (AGI).

By “AGI” I mean here “a bundle of chips, algorithms, electricity, and/or teleoperated robots that can autonomously do the kinds of stuff that ambitious human adults can do—founding and running new companies, R&D, learning new skills, using arbitrary teleoperated robots after very little practice, etc.”

Yes I know, this does not exist yet! (Despite hype to the contrary.) Try asking an LLM to autonomously write a business plan, found a company, then run and grow it for years as CEO. Lol! It will crash and burn! But that’s a limitation of today’s LLMs, not of “all AI forever”.

AI that could nail that task, and much more beyond, is obviously possible—human brains and bodies and societies are not powered by some magical sorcery forever beyond the reach of science. I for one expect such AI in my lifetime, for better or worse. (Probably “worse”, see below.)

Now, is this kind of AGI “labor” or “capital”? Well it’s not a flesh-and-blood human. But it’s more like “labor” than “capital” in many other respects:

• Capital can’t just up and do things by itself? AGI can.

• New technologies take a long time to integrate into the economy? Well ask yourself: how do highly-skilled, experienced, and entrepreneurial immigrant humans manage to integrate into the economy immediately? Once you’ve answered that question, note that AGI will be able to do those things too.

• Capital sits around idle if there are no humans willing and able to use it? Well those immigrant humans don’t sit around idle. And neither will AGI.

• Capital can’t advocate for political rights, or launch coups? Well…

Anyway, people see sci-fi robot movies, and they get this! Then they take economics courses, and it makes them dumber.

(Yes I know, #NotAllEconomists etc.)

THE SECOND PIECE of Econ anti-pedagogy is instilling a default assumption that it’s possible for a market to equilibrate. But the market for AGI cannot: AGI combines a property of labor markets with a property of product markets, where those properties are mutually exclusive. Those properties are:

• (A) “NO LUMP OF LABOR”: If human population goes up, wages drop in the very short term, because the demand curve for labor slopes down. But in the longer term, people find new productive things to do—the demand curve moves right. If anything, the value of labor goes UP, not down, with population! E.g. dense cities are engines of growth!

• (B) “EXPERIENCE CURVES”: If the demand for a product rises, there’s price increase in the very short term, because the supply curve slopes up. But in the longer term, people ramp up manufacturing—the supply curve moves right. If anything, the price goes DOWN, not up, with demand, thanks to economies of scale and R&D.

QUIZ: Considering (A) & (B), what’s the equilibrium price of this AGI bundle (chips, algorithms, electricity, teleoperated robots, etc.)?

…Trick question! There is no equilibrium. Our two principles, (A) “no lump of labor” and (B) “experience curves”, make equilibrium impossible:

• If price is low, (A) says the demand curve races rightwards—there’s no lump of labor, therefore there’s massive profit to be made by skilled entrepreneurial AGIs finding new productive things to do.

• If price is high, (B) says the supply curve races rightwards—there’s massive profit to be made by ramping up manufacturing of AGI.

• If the price is in between, then the demand curve and supply curve are BOTH racing rightwards!

This is neither capital nor labor as we know it. Instead of the market for AGI equilibrating, it forms a positive feedback loop / perpetual motion machine that blows up exponentially.

Does that sound absurd? There’s a precedent: humans! The human world, as a whole, is already a positive feedback loop / perpetual motion machine of this type! Humans bootstrapped themselves up from a few thousand hominins to 8 billion people running a $80T economy.

How? It’s not literally a perpetual motion machine. Rather, it’s an engine that draws from the well of “not-yet-exploited economic opportunities”. But remember “No Lump of Labor”: the well of not-yet-exploited economic opportunities is ~infinitely deep. We haven’t run out of possible companies to found. Nobody has made a Dyson swarm yet.

There’s only so many humans to found companies and exploit new opportunities. But the positive feedback loop of AGI has no such limit. The doubling time can be short indeed:

Imagine an autonomous factory that can build an identical autonomous factory, which then build two more, etc., using just widely-available input materials and sunlight. Economics textbooks don’t talk about that. But biology textbooks do! A cyanobacterium is such a factory, and can double itself in a day (≈ googol percent annualized growth rate 😛).

Anyway, we don’t know how explosive will be the positive feedback loop of AGI building AGI, but I expect it to be light-years beyond anything in economic history.

THE THIRD PIECE of Econ anti-pedagogy is its promotion of GDP growth as a proxy for progress and change. On the contrary, it’s possible for the world to transform into a wild sci-fi land beyond all recognition or comprehension each month, month after month, without “GDP growth” actually being all that high. GDP is a funny metric, and especially poor at describing the impact of transformative technological revolutions. (For example, if some new tech is inexpensive, and meanwhile other sectors of the economy remain expensive due to regulatory restrictions, then the new tech might not impact GDP much, no matter how much it upends the world.) I mean, sure we can argue about GDP, but we shouldn’t treat it as a proxy battle over whether AGI will or won’t be a big deal.

Last and most importantly, THE FOURTH PIECE of Econ anti-pedagogy is the focus on “mutually-beneficial trades” over “killing people and taking their stuff”. Econ 101 proves that trading is selfishly better than isolation. But sometimes “killing people and taking their stuff” is selfishly best of all.

When we’re talking about AGI, we’re talking about creating a new intelligent species on Earth, one which will eventually be faster, smarter, better-coordinated, and more numerous than humans.

Normal people, people who have seen sci-fi movies about robots and aliens, people who have learned the history of colonialism and slavery, will immediately ask lots of reasonable questions here. “What will their motives be?” “Who will have the hard power?” “If they’re seeming friendly and cooperative early on, might they stab us in the back when they get more powerful?”

These are excellent questions! We should definitely be asking these questions! (FWIW, this is my area of expertise, and I’m very pessimistic.)

…And then those normal people take economics classes, and wind up stupider. They stop asking those questions. Instead, they “learn” that AGI is “capital”, kinda like an injection-molding machine. Injection-molding machines wouldn’t wipe out humans and run the world by themselves. So we’re fine. Lol.

…Since actual AGI is so foreign to economists’ worldviews, they often deny the premise. E.g. here’s @tylercowen demonstrating a complete lack of understanding of what we doomers are talking about, when we talk about future powerful AI.

Yep. If you restrict to worlds where collaboration with humans is required in most cases then the impacts of AI all look mostly ‘normal’ again.

And here’s @DAcemogluMIT assuming without any discussion that in the next 10 yrs, “AI” will not include any new yet-to-be-developed techniques that go way beyond today’s LLMs. Funny omission, when the whole LLM paradigm didn’t exist 10 yrs ago!

(Tbc, it’s fine to make that assumption! Maybe it will be valid, or maybe not, who knows, technological forecasting is hard. But when your paper depends on a giant load-bearing assumption about future AI tech progress, an assumption which many AI domain experts dispute, then that assumption should at least be clearly stated! Probably in the very first sentence of the paper, if not the title!)

And here’s another example of economists “arguing” against AGI scenarios by simply rejecting out of hand any scenario in which actual AGI exists. Many such examples…

Eliezer Yudkowsky: Surprisingly correct, considering the wince I had at the starting frame.

I really think that if you’re creating a new intelligent species vastly smarter than humans, going “oh, that’s ‘this time is different’ economics”, as if it were economics in the first place, is exactly a Byrnes-case of seeing through an inappropriate lens and ending up dumber.

I am under no illusions that an explanation like this would satisfy the demands and objections of most economists or fit properly into their frameworks. It is easy for such folks to dismiss explanations like this as insufficiently serious or rigerous, or simply to deny the premise. I’ve run enough experiments to stop suspecting otherwise.

However, if one actually did want to understand the situation? This could help.

Discussion about this post

Reports Of AI Not Progressing Or Offering Mundane Utility Are Often Greatly Exaggerated Read More »

how-the-cavefish-lost-its-eyes—again-and-again

How the cavefish lost its eyes—again and again

Science, syndication / /

Mexican tetras in pitch-black caverns had no use for the energetically costly organs.

Photographs of Astyanax mexicanus, surface form with eyes (top) and cave form without eyes (bottom). Credit: Daniel Castranova, NICHD/NIH

Photographs of Astyanax mexicanus, surface form with eyes (top) and cave form without eyes (bottom). Credit: Daniel Castranova, NICHD/NIH

Time and again, whenever a population was swept into a cave and survived long enough for natural selection to have its way, the eyes disappeared. “But it’s not that everything has been lost in cavefish,” says geneticist Jaya Krishnan of the Oklahoma Medical Research Foundation. “Many enhancements have also happened.”

Though the demise of their eyes continues to fascinate biologists, in recent years, attention has shifted to other intriguing aspects of cavefish biology. It has become increasingly clear that they haven’t just lost sight but also gained many adaptations that help them to thrive in their cave environment, including some that may hold clues to treatments for obesity and diabetes in people.

Casting off expensive eyes

It has long been debated why the eyes were lost. Some biologists used to argue that they just withered away over generations because cave-dwelling animals with faulty eyes experienced no disadvantage. But another explanation is now considered more likely, says evolutionary physiologist Nicolas Rohner of the University of Münster in Germany: “Eyes are very expensive in terms of resources and energy. Most people now agree that there must be some advantage to losing them if you don’t need them.”

Scientists have observed that mutations in different genes involved in eye formation have led to eye loss. In other words, says Krishnan, “different cavefish populations have lost their eyes in different ways.”

Meanwhile, the fishes’ other senses tend to have been enhanced. Studies have found that cave-dwelling fish can detect lower levels of amino acids than surface fish can. They also have more tastebuds and a higher density of sensitive cells alongside their bodies that let them sense water pressure and flow.

Regions of the brain that process other senses are also expanded, says developmental biologist Misty Riddle of the University of Nevada, Reno, who coauthored a 2023 article on Mexican tetra research in the Annual Review of Cell and Developmental Biology. “I think what happened is that you have to, sort of, kill the eye program in order to expand the other areas.”

Killing the processes that support the formation of the eye is quite literally what happens. Just like non-cave-dwelling members of the species, all cavefish embryos start making eyes. But after a few hours, cells in the developing eye start dying, until the entire structure has disappeared. Riddle thinks this apparent inefficiency may be unavoidable. “The early development of the brain and the eye are completely intertwined—they happen together,” she says. That means the least disruptive way for eyelessness to evolve may be to start making an eye and then get rid of it.

In what Krishnan and Rohner have called “one of the most striking experiments performed in the field of vertebrate evolution,” a study published in 2000 showed that the fate of the cavefish eye is heavily influenced by its lens. Scientists showed this by transplanting the lens of a surface fish embryo to a cavefish embryo, and vice versa. When they did this, the eye of the cavefish grew a retina, rod cells, and other important parts, while the eye of the surface fish stayed small and underdeveloped.

Starving and bingeing

It’s easy to see why cavefish would be at a disadvantage if they were to maintain expensive tissues they aren’t using. Since relatively little lives or grows in their caves, the fish are likely surviving on a meager diet of mostly bat feces and organic waste that washes in during the rainy season. Researchers keeping cavefish in labs have discovered that, genetically, the creatures are exquisitely adapted to absorbing and storing nutrients. “They’re constantly hungry, eating as much as they can,” Krishnan says.

Intriguingly, the fish have at least two mutations that are associated with diabetes and obesity in humans. In the cavefish, though, they may be the basis of some traits that are very helpful to a fish that occasionally has a lot of food but often has none. When scientists compare cavefish and surface fish kept in the lab under the same conditions, cavefish fed regular amounts of standard fish food “get fat. They get high blood sugar,” Rohner says. “But remarkably, they do not develop obvious signs of disease.”

Fats can be toxic for tissues, Rohner explains, so they are stored in fat cells. “But when these cells get too big, they can burst, which is why we often see chronic inflammation in humans and other animals that have stored a lot of fat in their tissues.” Yet a 2020 study by Rohner, Krishnan, and their colleagues revealed that even very well-fed cavefish had fewer signs of inflammation in their fat tissues than surface fish do.

Even in their sparse cave conditions, wild cavefish can sometimes get very fat, says Riddle. This is presumably because, whenever food ends up in the cave, the fish eat as much of it as possible, since there may be nothing else for a long time to come. Intriguingly, Riddle says, their fat is usually bright yellow, because of high levels of carotenoids, the substance in the carrots that your grandmother used to tell you were good for your… eyes.

“The first thing that came to our mind, of course, was that they were accumulating these because they don’t have eyes,” says Riddle. In this species, such ideas can be tested: Scientists can cross surface fish (with eyes) and cavefish (without eyes) and look at what their offspring are like. When that’s done, Riddle says, researchers see no link between eye presence or size and the accumulation of carotenoids. Some eyeless cavefish had fat that was practically white, indicating lower carotenoid levels.

Instead, Riddle thinks these carotenoids may be another adaptation to suppress inflammation, which might be important in the wild, as cavefish are likely overeating whenever food arrives.

Studies by Krishnan, Rohner, and colleagues published in 2020 and 2022 have found other adaptations that seem to help tamp down inflammation. Cavefish cells produce lower levels of certain molecules called cytokines that promote inflammation, as well as lower levels of reactive oxygen species — tissue-damaging byproducts of the body’s metabolism that are often elevated in people with obesity or diabetes.

Krishnan is investigating this further, hoping to understand how the well-fed cavefish remain healthy. Rohner, meanwhile, is increasingly interested in how cavefish survive not just overeating, but long periods of starvation, too.

No waste

On a more fundamental level, researchers still hope to figure out why the Mexican tetra evolved into cave forms while any number of other Mexican river fish that also regularly end up in caves did not. (Globally, there are more than 200 cave-adapted fish species, but species that also still have populations on the surface are quite rare.) “Presumably, there is something about the tetras’ genetic makeup that makes it easier for them to adapt,” says Riddle.

Though cavefish are now well-established lab animals used in research and are easy to purchase for that purpose, preserving them in the wild will be important to safeguard the lessons they still hold for us. “There are hundreds of millions of the surface fish,” says Rohner, but cavefish populations are smaller and more vulnerable to pressures like pollution and people drawing water from caves during droughts.

One of Riddle’s students, David Perez Guerra, is now involved in a committee to support cavefish conservation. And researchers themselves are increasingly careful, too. “The tissues of the fish collected during our lab’s last field trip benefited nine different labs,” Riddle says. “We wasted nothing.”

This article originally appeared in Knowable Magazine, a nonprofit publication dedicated to making scientific knowledge accessible to all. Sign up for Knowable Magazine’s newsletter.

Photo of Knowable Magazine

Knowable Magazine explores the real-world significance of scholarly work through a journalistic lens.

How the cavefish lost its eyes—again and again Read More »

time-is-running-out-for-spacex-to-make-a-splash-with-second-gen-starship

Time is running out for SpaceX to make a splash with second-gen Starship

artemis, Commercial space, elon musk, launch, NASA, Science, Space, spacex, Starbase, starship, super heavy, texas / /

SpaceX is gearing up for another Starship launch after three straight disappointing test flights.

SpaceX’s 10th Starship rocket awaits liftoff. Credit: Stephen Clark/Ars Technica

STARBASE, Texas—A beehive of aerospace technicians, construction workers, and spaceflight fans descended on South Texas this weekend in advance of the next test flight of SpaceX’s gigantic Starship rocket, the largest vehicle of its kind ever built.

Towering 404 feet (123.1 meters) tall, the rocket was supposed to lift off during a one-hour launch window beginning at 6: 30 pm CDT (7: 30 pm EDT; 23: 30 UTC) Sunday. But SpaceX called off the launch attempt about an hour before liftoff to investigate a ground system issue at Starbase, located a few miles north of the US-Mexico border.

SpaceX didn’t immediately confirm when it might try again to launch Starship, but it could happen as soon as Monday evening at the same time.

It will take about 66 minutes for the rocket to travel from the launch pad in Texas to a splashdown zone in the Indian Ocean northwest of Australia. You can watch the test flight live on SpaceX’s official website. We’ve also embedded a livestream from Spaceflight Now and LabPadre below.

This will be the 10th full-scale test flight of Starship and its Super Heavy booster stage. It’s the fourth flight of an upgraded version of Starship conceived as a stepping stone to a more reliable, heavier-duty version of the rocket designed to carry up to 150 metric tons, or some 330,000 pounds, of cargo to pretty much anywhere in the inner part of our Solar System.

But this iteration of Starship, known as Block 2 or Version 2, has been anything but reliable. After reeling off a series of increasingly successful flights last year with the first-generation Starship and Super Heavy booster, SpaceX has encountered repeated setbacks since debuting Starship Version 2 in January.

Now, there are just two Starship Version 2s left to fly, including the vehicle poised for launch this week. Then, SpaceX will move on to Version 3, the design intended to go all the way to low-Earth orbit, where it can be refueled for longer expeditions into deep space.

A closer look at the top of SpaceX’s Starship rocket, tail number Ship 37, showing some of the different configurations of heat shield tiles SpaceX wants to test on this flight. Credit: Stephen Clark/Ars Technica

Starship’s promised cargo capacity is unparalleled in the history of rocketry. The privately developed rocket’s enormous size, coupled with SpaceX’s plan to make it fully reusable, could enable cargo and human missions to the Moon and Mars. SpaceX’s most conspicuous contract for Starship is with NASA, which plans to use a version of the ship as a human-rated Moon lander for the agency’s Artemis program. With this contract, Starship is central to the US government’s plans to try to beat China back to the Moon.

Closer to home, SpaceX intends to use Starship to haul massive loads of more powerful Starlink Internet satellites into low-Earth orbit. The US military is interested in using Starship for a range of national security missions, some of which could scarcely be imagined just a few years ago. SpaceX wants its factory to churn out a Starship rocket every day, approximately the same rate Boeing builds its workhorse 737 passenger jets.

Starship, of course, is immeasurably more complex than an airliner, and it sees temperature extremes, aerodynamic loads, and vibrations that would destroy a commercial airplane.

For any of this to become reality, SpaceX needs to begin ticking off a lengthy to-do list of technical milestones. The interim objectives include things like catching and reusing Starships and in-orbit ship-to-ship refueling, with a final goal of long-duration spaceflight to reach the Moon and stay there for weeks, months, or years. For a time late last year, it appeared as if SpaceX might be on track to reach at least the first two of these milestones by now.

The 404-foot-tall (123-meter) Starship rocket and Super Heavy booster stand on SpaceX’s launch pad. In the foreground, there are empty loading docks where tanker trucks deliver propellants and other gases to the launch site. Credit: Stephen Clark/Ars Technica

Instead, SpaceX’s schedule for catching and reusing Starships, and refueling ships in orbit, has slipped well into next year. A Moon landing is probably at least several years away. And a touchdown on Mars? Maybe in the 2030s. Before Starship can sniff those milestones, engineers must get the rocket to survive from liftoff through splashdown. This would confirm that recent changes made to the ship’s heat shield work as expected.

Three test flights attempting to do just this ended prematurely in January, March, and May. These failures prevented SpaceX from gathering data on several different tile designs, including insulators made of ceramic and metallic materials, and a tile with “active cooling” to fortify the craft as it reenters the atmosphere.

The heat shield is supposed to protect the rocket’s stainless steel skin from temperatures reaching 2,600° Fahrenheit (1,430° Celsius). During last year’s test flights, it worked well enough for Starship to guide itself to an on-target controlled splashdown in the Indian Ocean, halfway around the world from SpaceX’s launch site in Starbase, Texas.

But the ship lost some of its tiles during each flight last year, causing damage to the ship’s underlying structure. While this wasn’t bad enough to prevent the vehicle from reaching the ocean intact, it would cause difficulties in refurbishing the rocket for another flight. Eventually, SpaceX wants to catch Starships returning from space with giant robotic arms back at the launch pad. The vision, according to SpaceX founder and CEO Elon Musk, is to recover the ship, quickly mount it on another booster, refuel it, and launch it again.

If SpaceX can accomplish this, the ship must return from space with its heat shield in pristine condition. The evidence from last year’s test flights showed engineers had a long way to go for that to happen.

Visitors survey the landscape at Starbase, Texas, where industry and nature collide. Credit: Stephen Clark/Ars Technica

The Starship setbacks this year have been caused by problems in the ship’s propulsion and fuel systems. Another Starship exploded on a test stand in June at SpaceX’s sprawling rocket development facility in South Texas. SpaceX engineers identified different causes for each of the failures. You can read about them in our previous story.

Apart from testing the heat shield, the goals for this week’s Starship flight include testing an engine-out capability on the Super Heavy booster. Engineers will intentionally disable one of the booster’s Raptor engines used to slow down for landing, and instead use another Raptor engine from the rocket’s middle ring. At liftoff, 33 methane-fueled Raptor engines will power the Super Heavy booster off the pad.

SpaceX won’t try to catch the booster back at the launch pad this time, as it did on three occasions late last year and earlier this year. The booster catches have been one of the bright spots for the Starship program as progress on the rocket’s upper stage floundered. SpaceX reused a previously flown Super Heavy booster for the first time on the most recent Starship launch in May.

The booster landing experiment on this week’s flight will happen a few minutes after launch over the Gulf of Mexico east of the Texas coastline. Meanwhile, six Raptor engines will fire until approximately T+plus 9 minutes to accelerate the ship, or upper stage, into space.

The ship is programmed to release eight Starlink satellite simulators from its payload bay in a test of the craft’s payload deployment mechanism. That will be followed by a brief restart of one of the ship’s Raptor engines to adjust its trajectory for reentry, set to begin around 47 minutes into the mission.

If Starship makes it that far, that will be when engineers finally get a taste of the heat shield data they were hungry for at the start of the year.

This story was updated at 8: 30 pm EDT after SpaceX scrubbed Sunday’s launch attempt.

Photo of Stephen Clark

Stephen Clark is a space reporter at Ars Technica, covering private space companies and the world’s space agencies. Stephen writes about the nexus of technology, science, policy, and business on and off the planet.

Time is running out for SpaceX to make a splash with second-gen Starship Read More »