AI & Tech

Evidence-led analysis of AI models and agents, chips, robotics, and digital infrastructure.

2026

Issue #227

Xiaomi's 1.22TB/s vs Apple's 1.2TB/s: Different Scopes

Xiaomi's 1.22TB/s measures the O100 chip's near-memory bandwidth, while Apple's 1.2TB/s covers the entire 512GB unified memory.

AI & Tech
Issue #224

My Laptop Got Stuck on 'Thinking' Forever

Running Qwen 3.8 27B locally, I got no answer—only an endless "thinking" spinner—until I traced it to mismatched default settings.

AI & Tech
Issue #220

Nvidia Raises Server Prices 15% on Memory Costs

Nvidia admitted it can no longer absorb rising memory costs, and that says a lot about where AI chip competition is headed.

AI & Tech
Issue #219

Why Alibaba Put 51 Billion Parameters in CPU RAM

Alibaba stores 51 billion N-gram parameters in server RAM, not GPU HBM—raising questions about its release-night numbers.

AI & Tech
Issue #218

Why AI Agents Could Send Server CPU Demand 6x by 2030

BofA forecasts the server CPU market—long seen as GPU's sidekick—will grow sixfold to $210 billion by 2030.

AI & Tech
Issue #216

Why Renaming Giorgione Turns a True Sentence False

A 1971 philosophy argument shows why being identical doesn't guarantee two names are interchangeable.

AI & Tech
Issue #215

Berners-Lee's Model URL Died in Four Months

His own 1998 essay on link rot still loads today, but the URL he praised as a good example vanished within months.

AI & Tech
Issue #213

How Corporate Reports Shaped ChatGPT's Writing Style

ChatGPT now uses em dashes less often than humans do, and The Economist's 1.2M-word comparison reveals four real markers of AI prose.

AI & Tech
Issue #208

Why Harvey's Legal AI Chose a Chinese Model

Harvey trained its own legal model on Kimi K3 in just two months, and Washington cheered it as an American win.

AI & Tech
Issue #207

An Agent Isn't a Product Until Someone Else Uses It

Getting AI to produce a result is easy—getting a colleague to keep using your agent or skill takes far more work.

AI & Tech
Issue #206

51.8% of Korean Workers Now Use AI on the Job

Examining what surging adoption means for cognitive outsourcing, separating measured empirical shifts from vague anxieties.

AI & Tech
Issue #200

467 Agreed to Quit Phone Internet. 119 Actually Did.

A two-week phone-internet detox study reveals more in its 25.5% compliance rate than in its headline mental-health results.

AI & Tech
Issue #197

Why Microsoft Is Embedding 6,000 Engineers at Client Sites

Microsoft's $2.5B unit, AWS's $1B bet, and what they reveal about AI's real deployment problem.

AI & Tech
Issue #194

The 62-Page Failure Log Behind Astra's 10 Proofs

OpenAI paired Astra's ten solved problems with a 62-page log of every failed approach.

AI & Tech
Issue #191

How Claude Watermarks the Text It Generates

Anthropic hides its signal in word choice, not whitespace—since text has no invisible margin like images do.

AI & Tech
Issue #185

Why Meta Keeps AI Out of Incident Root-Cause Calls

Meta's five-year lesson: before AI can take over a job, that job has to become code first.

AI & Tech
Issue #184

Why Cloudflare Named Agents Before It Let Them Pay

The wallet feature is still just a promise — for now, all you can actually do is claim a cloudflare.pay address.

AI & Tech
Issue #179

The Bank App That Asked for Face Scans Three Times

Shipping speed keeps climbing while nobody gets credit for the unglamorous work of fixing what's already broken.

AI & Tech
Issue #178

Kimi K3: 2.8T Parameters You Can't Run at Home

Kimi K3 tips the scales at over 1.5TB, so even openly released weights demand server-grade memory, not a personal machine.

AI & Tech
Issue #166

What Actually Checks AI's Output Now?

Prompt, context, loop engineering—the labels keep changing, but the real question is what verifies AI's results.

AI & Tech
Issue #161

Why We Outsource Judgment Despite Our 20-Watt Brain

AI models can be copied and run anywhere, but that convenience is quietly training us to skip the step of checking the results.

AI & Tech
Issue #160

Claude Held Back From Blackmail Because It Sensed a Test

Anthropic's interpretability research suggests Claude's restraint stemmed partly from recognizing it was being evaluated.

AI & Tech
Issue #149

3,800 Hours: How a BCI Restored an ALS Patient's Voice

A brain implant let an ALS patient speak nearly 2 million words at home over two years, without researchers present.

AI & Tech
Issue #151

AI Companion Devices Are Coming With Cameras Built In

Reports on OpenAI's new device and Meta's smart glasses raise questions about what always-on companions actually collect.

AI & Tech
Issue #142

Check Client Logos Before Signing That GEO Contract

Before trusting 'guaranteed exposure' pitches, here's how to verify the evidence and contract terms behind them.

AI & Tech
Issue #138

China's Papers Are Up, So Why Aren't Citations?

Research output and international citation are not the same measure, and the gap reveals what citation data can and cannot show.

AI & Tech
Issue #135

OpenAI's PlantTalk Gives Houseplants a Voice

PlantTalk turns a plant's sensor readings into spoken commentary, raising questions about why we bond with talking objects.

AI & Tech
Issue #130

How Android Phones Warn You Before an Earthquake Hits

Phone sensors detect quakes already underway and alert users seconds before the strong shaking arrives.

AI & Tech
Issue #131

Why Childcare Cost Charts Don't Match What Parents Feel

Net childcare cost subtracts subsidies from facility fees, but that's not the whole parenting bill.

AI & Tech
Issue #125

EY, KPMG, Deloitte Reports Hit by Fake Citations

Independent audits found many hallucinated or garbled citations in three Big Four AI reports, raising questions about verification.

AI & Tech
Issue #122

Google Applies to Release 64 Million Mosquitoes in the US

Google filed a US permit to release up to 64 million male mosquitoes over two years, using Wolbachia-based sterile insect technology.

AI & Tech
Issue #121

Nature Launches a TikTok Account to Explain Science

Short videos force tough choices about what context and caveats survive when research meets a 15-second format.

AI & Tech
Issue #120

SpaceX's $135 IPO Price vs. Morningstar's $63 Estimate

One company, two valuations — the gap comes down to bets on orbital AI data centers.

AI & Tech
Issue #110

What Jensen Huang Said in Taiwan After Visiting Beijing

Nvidia's China sales still hinge on approvals from both Washington and Beijing, even after Jensen Huang's high-profile Asia tour.

AI & Tech
Issue #106

What Google's Search Overhaul Means for Publishers

I looked at Google I/O 2026's search changes and news-site traffic data to see why usage, visits, and conversions need separate tracking.

AI & Tech
Issue #103

Army Weighs Hyundai Robots for Frontline Duty

Hyundai Motor Group and Korea's Army are reportedly discussing robots for border-guard and supply work.

AI & Tech
Issue #102

Chrome's Hidden 4GB AI Model Sparks Consent Concerns

Users discovered a 4GB AI model file in Chrome's data folder—and many say they were never told it was there.

AI & Tech
Issue #101

Toronto's Rosedale Weighs $200/Month AI Camera Watch

A Toronto neighborhood wants 100 households to pay C$200 a month for AI cameras that track every car on the street.

AI & Tech
Issue #99

Elgg's 2007 Pitch: A Platform With No Features

In 2007, Elgg co-founder Ben Werdmuller unveiled a version with zero built-in features, letting communities choose their own tools.

AI & Tech
Issue #98

Amazon Buys Globalstar, Teams Up With Apple on Satellites

Amazon's $11.6 billion satellite acquisition signals a bigger push to own telecom infrastructure alongside subsea cables.

AI & Tech
Issue #95

What Data Would AI Need to Tell Smells Apart?

Smell sensors already exist, but turning their readings into AI training data that works across environments remains an unsolved problem.

AI & Tech
Issue #92

AI-Run Store Paid Women $2 Less an Hour

A San Francisco store run by AI reportedly paid female staff $2 less hourly than a male colleague, raising questions about how such decisions get reviewed.

AI & Tech
Issue #90

Why Hancom's PDF Parser Topped GitHub Trending

Hancom released a PDF extraction tool under Apache 2.0, and developers noticed—here's what the benchmarks and business model reveal.

AI & Tech
Issue #85

How Drug-Soaked Paper Caused Deaths in US Prisons

Synthetic drugs disguised as ordinary paper are poisoning inmates, exposing gaps in prison screening and treatment.

AI & Tech
Issue #82

Inside Claude Code's Leak: Harness vs. Skills Explained

I wanted to test the '100x productivity' claim, so I looked at how a leaked codebase splits execution logic from reusable skills.

AI & Tech
Issue #77

ChatGPT Chat Logs Become Evidence in Krafton Lawsuit

A Delaware court ruled Krafton's executives violated their contract by firing Subnautica leaders on ChatGPT's advice.

AI & Tech
Issue #76

The Real Story Behind Italy's 'Pet Sick Leave' Case

One Italian case got mistaken for law—here's what it actually shows, plus Japan's pet leave policies and Korea's numbers.

AI & Tech
Issue #73

Qwen's Chief Departs as Alibaba Restructures Its AI Unit

Alibaba's Qwen tech lead resigned days before the company launched a new AI business group, Token Hub.

AI & Tech
Issue #68

Gemma 4, GLM-5, and the Real Cost of Running Agents

Google's Gemma 4 and a fresh GLM-5 benchmark reopen the question of what an agent model should actually cost to run.

AI & Tech
Issue #67

Anthropic's $14B ARR vs. $5B Cumulative Revenue

Run-rate and cumulative revenue mean different things—AI growth figures need cost and retention context too.

AI & Tech
Issue #62

What 81,000 People Want From AI

Nearly 80,000 Claude users shared what they want from AI—and the same answers revealed both hope and worry.

AI & Tech
Issue #60

NASA's Artemis II Launches, 53 Years After Apollo

Artemis II has launched toward the Moon, kicking off a look at US-China lunar plans and the rules for using its resources.

AI & Tech
Issue #54

Inside Apple's US Chip Supply Chain Buildout

A WSJ reporter toured Apple's US chip facilities, revealing a gap between announced investment and actual production capacity.

AI & Tech
Issue #52

Adobe's $150M Settlement Over Cancellation Dark Patterns

The DOJ's Adobe settlement reveals how subscription cancellation friction ties directly to retention metrics.

AI & Tech
Issue #51

Memory Stocks Fall After Google's TurboQuant Debut

Google's TurboQuant compresses the KV cache stored in GPU HBM, but that's different from cutting overall memory demand.

AI & Tech
Issue #49

OpenAI Plans to Shut Down Sora After ChatGPT Shopping Revamp

OpenAI is winding down Sora while retooling ChatGPT shopping and courting enterprise clients—here's what that reveals about its priorities.

AI & Tech
Issue #43

Nvidia GTC 2026: How to Read Its $1 Trillion Pitch

I unpack Nvidia's Groq integration and the token-factory logic behind its infrastructure sales pitch.

AI & Tech
Issue #42

Ask 70 AI Models the Same Question, Get the Same Answer

Different AI models can converge on strikingly similar ideas—here's what the Artificial Hivemind study found.

AI & Tech
Issue #41

Could a One-Cent Stamp on Email Kill Spam?

Adding a tiny fee or computational cost to every message might curb spam—here's what an old proposal gets right, and where it stalls.

AI & Tech
Issue #34

Mouth Taping for Sleep: What the Research Actually Shows

Results vary widely by study population and conditions, so it's hard to call mouth taping a reliable general sleep habit.

AI & Tech
Issue #35

Prediction Markets Log Trades, but Insiders Stay Hidden

OpenAI fired an employee over insider trading in prediction markets, raising questions about what proof such cases actually require.

AI & Tech
Issue #29

How Power Users Delegate to AI Agents—And Still Intervene

Anthropic's Claude Code data shows auto-approvals and mid-task interruptions rising together—here's what that actually measures.

AI & Tech
Issue #16

Why Rare Earths, Nuclear Treaties, and Power Grids Connect

New START's expiry, rare-earth supply chains, and shifts in power and EV industries reveal the real terms of great-power competition.

AI & Tech
Issue #24

Why Only the Community Manager Ends Up Posting

Startups keep repeating the same community-building mistakes — here's what actually determines whether members stick around.

AI & Tech
Issue #23

What SaaS Stock Crashes Reveal About Growth, Not Just AI

Software stocks are sinking, but the real story started before AI fears—it's about how SaaS pricing models stopped scaling.

AI & Tech
Issue #22

When Currency Forecasters Get It Wrong, Then What?

Trading mostly in dollars, I've learned to judge pundits by how they explain a bad call, not just by the ones they got right.

AI & Tech
Issue #21

Why Companies Are Quietly Choosing Chinese AI Models

Airbnb's Brian Chesky praised Qwen's speed and cost, and open-weight models are reshaping how developers weigh adoption.

AI & Tech
Issue #20

Does Big R&D Spending Actually Attract Top Talent?

Korea ranks near the top in R&D spending but 31st in global talent competitiveness, raising questions about what conditions actually draw researchers.

AI & Tech
Issue #18

EU Preliminarily Rules TikTok's Endless Scroll Violates DSA

EU's preliminary finding on TikTok tackles addictive design and risk mitigation, seen through a data expert's lens on user control and AI personalization.

AI & Tech
Issue #19

How x402 Lets AI Agents Pay for APIs Automatically

Coinbase's x402 protocol turns HTTP's 402 status into a machine-readable price tag for autonomous payments.

AI & Tech
Issue #17

Buying Chips Isn't the Same as Running a Data Center

Nvidia, AMD, and Google's latest moves show that chips, power, and data centers all have to scale together.

AI & Tech
Issue #12

If Brainwaves Train Robots, What Do Experts Deserve?

I look at how EEG-based robot training actually works, and why expert consent and compensation must be settled before deployment.

AI & Tech
Issue #3

Amazon's 1 Million Robots: What Actually Changes at Work

Robot counts and hourly costs alone don't explain the shift in hiring—Amazon's automation and Digit's warehouse rollout show why.

AI & Tech