Daisy P-L

Connecting dots on:
climate, energy, housing, ag, politics.

[Daisy P-L @daisypl]

@heynavtoor Same thing as happened with industrialization in the late 1800s. It was supposed to give us all scads of leisure time with which to enjoy life. Instead, we have endless workweeks and unused vacation days and dinners at the office. 🙄

[Nav Toor @heynavtoor]

🚨BREAKING: Berkeley researchers spent 8 months inside a tech company watching how employees actually use AI.

The promise was simple: AI will save you time. Do less. Work smarter.

The opposite happened.

Workers didn't use AI to finish early and go home. They used it to take on more. More tasks. More projects. More hours. Nobody asked them to. They did it to themselves.

The researchers sat inside the company two days a week for 8 months. They watched 200 employees in real time. They tracked work channels. They conducted 40+ interviews across engineering, product, design, and operations.

Here's what they found. AI made everything feel faster, so people filled every gap. They sent prompts during lunch. Before meetings. Late at night. The natural stopping points in the workday disappeared. People ran multiple AI agents in the background while writing code, drafting documents, and sitting in meetings simultaneously.

It felt like momentum. It felt productive. But when they stepped back, they described feeling stretched, busier, and completely unable to disconnect.

83% said AI increased their workload. Not decreased. Increased.

62% of associates and 61% of entry-level workers reported burnout. Only 38% of executives felt the same strain. The people doing the actual work absorbed the damage while leadership celebrated the productivity numbers.

Then came the trap nobody saw coming. When one person uses AI to take on extra work, everyone else feels like they're falling behind. So the whole team speeds up. Nobody formally raises expectations. But the new pace quietly becomes the default. What AI made possible became what was expected.

The researchers gave it a name: workload creep. It looks like productivity at first. Then it becomes the new baseline. Then it becomes burnout.

AI was supposed to give you your time back. Instead it's eating more of it. And the worst part? You're doing it to yourself. Voluntarily.

[anand iyer @ai]

Meta’s El Paso data center plans to link 813 gas generators into a single on-site power system. Utilities can’t deliver new grid capacity fast enough, so developers are building “behind-the-meter” gas plants beside their servers.

xAI showed proved this model in Memphis, running largely on portable generators before EPA required emissions permits. Cleanview counts 47 such off-grid projects nationwide now including Meta, OpenAI, Oracle and Chevron.

We talk a lot about the near-term AI power story being nuclear or fusion, but not just yet.

https://t.co/mHtvyYFIqz

[Ryan Hart @thisdudelikesAI]

🚨 Holy shit… Stanford just published the most uncomfortable paper on LLM reasoning I’ve read in a long time.

This isn’t a flashy new model or a leaderboard win. It’s a systematic teardown of how and why large language models keep failing at reasoning even when benchmarks say they’re doing great.

The paper does one very smart thing upfront: it introduces a clean taxonomy instead of more anecdotes. The authors split reasoning into non-embodied and embodied.

Non-embodied reasoning is what most benchmarks test and it’s further divided into informal reasoning (intuition, social judgment, commonsense heuristics) and formal reasoning (logic, math, code, symbolic manipulation).

Embodied reasoning is where models must reason about the physical world, space, causality, and action under real constraints.

Across all three, the same failure patterns keep showing up.

> First are fundamental failures baked into current architectures. Models generate answers that look coherent but collapse under light logical pressure. They shortcut, pattern-match, or hallucinate steps instead of executing a consistent reasoning process.

> Second are application-specific failures. A model that looks strong on math benchmarks can quietly fall apart in scientific reasoning, planning, or multi-step decision making. Performance does not transfer nearly as well as leaderboards imply.

> Third are robustness failures. Tiny changes in wording, ordering, or context can flip an answer entirely. The reasoning wasn’t stable to begin with; it just happened to work for that phrasing.

One of the most disturbing findings is how often models produce unfaithful reasoning. They give the correct final answer while providing explanations that are logically wrong, incomplete, or fabricated.

This is worse than being wrong, because it trains users to trust explanations that don’t correspond to the actual decision process.

Embodied reasoning is where things really fall apart. LLMs systematically fail at physical commonsense, spatial reasoning, and basic physics because they have no grounded experience.

Even in text-only settings, as soon as a task implicitly depends on real-world dynamics, failures become predictable and repeatable.

The authors don’t just criticize. They outline mitigation paths: inference-time scaling, analogical memory, external verification, and evaluations that deliberately inject known failure cases instead of optimizing for leaderboard performance.

But they’re very clear that none of these are silver bullets yet.

The takeaway isn’t that LLMs can’t reason.

It’s more uncomfortable than that.

LLMs reason just enough to sound convincing, but not enough to be reliable.

And unless we start measuring how models fail not just how often they succeed we’ll keep deploying systems that pass benchmarks, fail silently in production, and explain themselves with total confidence while doing the wrong thing.

That’s the real warning shot in this paper.

Paper: Large Language Model Reasoning Failures

[Daisy P-L @daisypl]

@kron4news He was repeating back what the man in the crowd said. Why is this news?!?

[Daisy P-L @daisypl]

Another stupid @Waymo blocking traffic at rush hour @ 16th and Valencia. 🤦🏼‍♀️🤦🏼‍♀️ Must fix this, constant problem on 1-lane streets in SF. #Waymo & other #RobotCars should pull to the side to wait. Passengers can walk a few storefronts to avoid inconveniencing the rest of the city.

[Daisy P-L @daisypl]

@TKavulla the trees are exploding because even they want the ICE out (of their xylem and phloem) in MN.

[Brian Schatz @brianschatz]

One of the most underreported stories of 2025 is that Trump is the first President ever to create electricity shortages on purpose. He is illegally canceling solar and wind, causing layoffs and price spikes, and the media is like “wow what a showman that guy is!”

[Daisy P-L @daisypl]

@grok @putthole21 @TomSteyer Silly, why do you think there are labor shortages?

[Frank Hatcher @FrankRHatcher]

@samstein Data heist.

[Sam Stein @samstein]

But Elon Musk got a trillion dollar pay package from Tesla and was dining with Trump and MBS the other day. so, i guess, it was a success on that front.

[Sam Stein @samstein]

The legacy of DOGE is not that it failed. It's that it destroyed:

1. faith in real cost conscious governance
2. the careers of committed govt employees
3. scientific research and the biomedical breakthroughs that come with it
4. countless lives of the world's poorest

[Maine @TheMaineWonk]

If you’re keeping track at home, Trump admin has cancelled:

- Jobs Report
- Inflation Report
- GDP Report

I’m sure everything is just fine.

[Daisy P-L @daisypl]

@ArmandDoma Even better when Mario can take Wario's money

[House Foreign Affairs Committee Dems @HouseForeign]

Trump brazenly using his pardon power to free a supporter who was found guilty of helping China repress and intimidate people on American soil.

Blatant corruption at the cost of US national security and values.

[Eric Geller @ericgeller]

"The provision, tucked into a measure to fund the legislative branch, appears to immediately allow for eight GOP senators to sue over their phone records being seized in the course of the investigation by Jack Smith ... into the riot at the Capitol..." https://t.co/R1xMzuisy6

[Elizabeth Warren @SenWarren]

I will not support a deal that does nothing to make health care more affordable.

We are in a health care emergency. A simple one-year extension of these tax credits would cost less than Donald Trump’s $40 billion bailout for Argentina.

A vote for this bill is a mistake.

[Robert Reich @RBReich]

Effective tax rates before and after the Trump tax law:

Verizon
Before: 21%
After: 8%

Walmart
Before: 31%
After: 17%

AT&T
Before: 13%
After: 3%

Walt Disney
Before: 26%
After: 8%

FedEx
Before: 18%
After: 1%

This is what a corporate giveaway looks like.

[Daisy P-L @daisypl]

If you include the normal coders that the Big Tech overlords are trying to replace with AI, that *miiiiight* be true. But once it's just Elon and his fellow tech leaders executing their dream of AI doing all the work, this will be flipped on its head and the proportions way off.

[Elon Musk @elonmusk]

Accurate

[Daisy P-L @daisypl]

@charlove @ArmandDoma This. WTH?!