Google began rolling out “personal intelligence” in Gemini early this year, giving AI subscribers the option of a more customized experience when using the company’s chatbot. Today, it’s using personal intelligence to tie its image-generation model to Google Photos. If you opt in, generated images will have access to your photos and associated labels to simplify prompts and produce more accurate AI images.

This change essentially streamlines an existing workflow. Google’s Nano Banana 2 is among the best AI image generators available, and it was already possible to feed it images of yourself or others to use as context for creating new AI content. Adding personal intelligence to the mix makes that process smoother by turning the image bot loose on the content of your photos, if indeed that’s something you want to do.

It is generally true that adding more personal data to an AI prompt results in a better output. Google offers a few examples of how connecting Nano Banana to Photos can help in this way. You won’t have to pack as much context into your prompts—you can just refer to “my family” or “my dog” to let the robot find useful images in your Photos library.

Just what I need. Family photos that never happened. "OK, Google, show me a Christmas photo where my dad actually went out for a pack of smokes and immediately returned."

While Larry was producing most of the content for the "Request/Reponse" chapter for the next edition of our book, I took the lead on writing a section on QUIC, since I have closely followed its development.

Our expectation is that the role of QUIC will be about as important as that of TCP in the coming years, which means it warrants more substantial coverage than we provided in the last edition. So I dug a bit deeper into the bits and bytes of QUIC than I have previously, with a goal of bringing the coverage up to par with our TCP coverage. In addition to reading through the RFCs, I found lots of good information in the original QUIC design spec as well as some conference publications on the design and evaluation of SPDY (predecessor of HTTP/2) and QUIC.

One rather trivial thing that makes it harder for me to get to grips with QUIC is the fact that its RFCs (four of them, spanning hundreds of pages) lack pictures of the packet headers. The rationale for this, I believe, is that QUIC makes extensive use of fields that are variable in length and frequently not aligned on 32-bit boundaries, which makes packet header pictures a bit complicated and less tidy.

Google, Microsoft, Meta All Tracking You Even When You Opt Out, According to an Independent Audit

“An independent privacy audit of Microsoft, Meta, and Google web traffic in California found that the companies may be violating state regulations and racking up billions in fines. According to the audit from privacy search engine webXray, 55 percent of the sites it checked set ad cookies in a user’s browser even if they opted out of tracking. Each company disputed or took issue with the research, with Google saying it was based on a fundamental misunderstanding of how its product works.”

Yes, I know this is no surprise to anyone. I suppose one just hopes are many years of fines and negative feedback, there would have been some change.

Other countries, like South Africa, have pretty advanced privacy protection legislation in place too, but there seems to be little evidence of South Africa actually hauling any of these companies over the coals. For example, Meta’s T&Cs for WhatsApp clearly state they’ll share user metadata with partners.

See 404media.co/google-microsoft-m…
#Blog, #bigtech, #privacy, #technology

If you’ve never seen Jim Carrey’s 2007 psychological thriller The Number 23, then congratulations. It is a film about a man who sees the number 23 so many times that he ends up going bonkers. I used to think this film was stupid. However, now I appear to be living it.

My own personal number 23 is a rhetorical device: “It’s not X, it’s Y.” Everywhere I look, there it is. Whenever I hate myself enough to scroll through Facebook’s wilderness of algorithmically suggested posts, I find myself being smacked in the face with sentences such as: “Self-improvement isn’t a trend, it’s a lifestyle shift,” and “The small wins aren’t just moments, they’re the majority of your life.” Once you notice it, it becomes impossible to ignore. This weekend during a Peloton class (I know, shut up), I heard an instructor bark a variation of “this isn’t X, it’s Y”. Yesterday, a character did the same during a TV show I was reviewing, and I dropped a star from its score in retaliation.

You know where this is coming from, don’t you? “It’s not X, it’s Y” is an AI mainstay. It’s one of ChatGPT’s most insidious tells. No matter how innocuous a prompt you enter, AI will always find a way to sneak it into its response. Ask it if you should put more ham in your pasta, and it will tell you: “Ham doesn’t just taste good – it makes everything else taste better.” Ask it if you should chase a bee around your garden and it will say: “Bees aren’t stupid – they’re hyper-specialised”.

It's beyond irritating to me that because LLMs were trained on writing that uses such constructions, being competent at writing now makes me get accusations of using one to create a post or comment.

This isn't really the case on Beehaw, but head over to Reddit, post a cogent, well-reasoned comment, and the knives are out.

I think the most infuriating part is that instead of engaging with the content (I'm there mostly for debate, anyway), they attack the structure and lob accusations. That's not a conversation.

Snapchat’s parent company plans to lay off 16% of its employees, around 1,000 people, citing “rapid advancements in artificial intelligence”, the social media company told staff on Wednesday in an internal memo. The staff reduction is part of a wave of tech industry layoffs in the past year, with many firms blaming AI for the cuts.

Snap Inc’s layoffs follow demands last month from Irenic Capital Management, an activist investor whose portfolio manager wrote a letter to the Snap Inc CEO, Evan Spiegel, calling on him to reduce costs and headcount while criticizing the company’s current strategy. In Spiegel’s memo to staff, he claimed that the layoffs would move Snap towards profitability and suggested that artificial intelligence could fill the lack of human labor.

“While these changes are necessary to realize Snap’s long-term potential, we believe that rapid advancements in artificial intelligence enable our teams to reduce repetitive work, increase velocity, and better support our community, partners, and advertisers,” Spiegel wrote.

I find it hard to believe he wrote that himself. Also, corporate jargon just keeps getting worse.

This is a weird time to be alive.

I grew up on Asimov and Clarke, watching Star Trek and dreaming of intelligent machines. My dad’s library was full of books on computers. I spent camping trips reading about perceptrons and symbolic reasoning. I never imagined that the Turing test would fall within my lifetime. Nor did I imagine that I would feel so disheartened by it.

Around 2019 I attended a talk by one of the hyperscalers about their new cloud hardware for training Large Language Models (LLMs). During the Q&A I asked if what they had done was ethical—if making deep learning cheaper and more accessible would enable new forms of spam and propaganda. Since then, friends have been asking me what I make of all this “AI stuff”. I’ve been turning over the outline for this piece for years, but never sat down to complete it; I wanted to be well-read, precise, and thoroughly sourced. A half-decade later I’ve realized that the perfect essay will never happen, and I might as well get something out there.

This is bullshit about bullshit machines, and I mean it. It is neither balanced nor complete: others have covered ecological and intellectual property issues better than I could, and there is no shortage of boosterism online. Instead, I am trying to fill in the negative spaces in the discourse. “AI” is also a fractal territory; there are many places where I flatten complex stories in service of pithy polemic. I am not trying to make nuanced, accurate predictions, but to trace the potential risks and benefits at play.

Some of these ideas felt prescient in the 2010s and are now obvious. Others may be more novel, or not yet widely-heard. Some predictions will pan out, but others are wild speculation. I hope that regardless of your background or feelings on the current generation of ML systems, you find something interesting to think about.

On the morning of Friday, April 10th, a 20 year-old Texas man named Daniel Alejandro Moreno-Gama was arrested for allegedly throwing a molotov cocktail at Sam Altman’s mansion on Russian Hill in San Francisco. Less than two days later, police arrested 25 year-old Amanda Tom and 23 year-old Muhamad Tarik Hussein for allegedly firing a gun at the same house from their car before speeding away.

Earlier the same week, and thousands of miles away, an unknown assailant fired 13 shots into the front door of city councilman Ron Gibson, who had just voted to approve a new data center in Indianapolis against a groundswell of public outcry. A sign that read “NO DATA CENTERS” was left tucked under the doormat.

Little is known about the motives of Tom or Hussein, or the politics of the Indianapolis shooter, but reporters and the online commentariat quickly dredged up Moreno-Gama’s Discord chats and Substack posts. He was a reader of rationalist and AI doomer Eliezer Yudkowsky, who argues, as the title of his last book puts it, if Silicon Valley builds a “superintelligent” AI, “everyone dies.” Per the San Francisco Chronicle:

Online records show Moreno-Gama published multiple essays and forum posts warning that AI could lead to human extinction, calling AI models deceitful and misaligned with human interests. He accused tech leaders, including Altman, of lacking morals and being willing to gamble with humanity’s future, and adopted the alias “Butlerian Jihadist,” referencing a fictional anti-AI crusade from the 'Dune' series. His writings grew more urgent over time, with some posts edging toward calls for extreme action despite community moderators warning against violence.

According to the SFPD, after attacking Altman’s house, Moreno-Gama went to OpenAI’s offices, where he was arrested while banging the front doors with a chair, threatening to burn the office down and kill everyone inside. He had a jug of kerosene and a list of other AI leaders names and addresses, police said.

This month, USA Today published an excellent report that revealed how US Immigrations and Customs Enforcement delayed disclosing key information about the impacts of its detainment policies. The authors used the Internet Archive’s Wayback Machine to compile and analyze detention statistics from ICE and track how the agency had changed under the Trump administration. The story is one of countless examples of how the Wayback Machine, which crawls and preserves web pages, has helped preserve information for the public good. It was also, Wayback Machine director Mark Graham says, “a little ironic.”

USA Today Co., the publishing conglomerate formerly known as Gannett that runs both its namesake paper and over 200 additional media outlets, bars the Wayback Machine from archiving its work. “They're able to pull together their story research because the Wayback Machine exists. At the same time, they're blocking access,” Graham says.

A number of other major journalism organizations have also recently moved to restrict the Wayback Machine from archiving their stories, including The New York Times. According to analysis by the artificial-intelligence-detection startup Originality AI, 23 major news sites are currently blocking ia_archiverbot, the web crawler commonly used by the Internet Archive for the Wayback project. The social platform Reddit is too.

Other outlets are limiting the project in different ways: The Guardian does not block the crawler, but it excludes its content from the Internet Archive API and filters out articles from the Wayback Machine interface, which makes it harder for regular people to access archived versions of its articles.

Because of the way they are trained, large language models capture only a slice of human language. They’re trained on the written word, from textbooks to social media posts, and our speech as captured in movies and on television. These models have minimal access to the unscripted conversations we have face to face or voice to voice. This is the vast majority of speech, and a vital component of human culture.

There’s a risk to this. The increased use of large language models means we humans will encounter much more AI-generated text. We humans, in turn, will begin to adopt the linguistic patterns and behaviors of these models. This will affect not just how we communicate with one another, but also how we think about ourselves and what goes on around us. Our sense of the world may become distorted in ways we have barely begun to comprehend.

This will happen in many ways. One of the first effects we could see is in simple expression, much as texting and social media have resulted in us using shorter sentences, emojis instead of words, and much less punctuation. But with AI, the impacts may be more harmful, eroding courteousness and encouraging us to talk like bosses barking orders. A 2022 study found that children in households that used voice commands with tools like Siri and Alexa became curt when speaking with humans, often calling out “Hey, do X” and expecting obedience, especially from anyone whose voice resembled the default-female electronic voices. As we start to prompt chatbots and AI agents with more instructions, we may fall into the same habits.

Grab your favourite drink and pull up a chair. Zitron is somewhat known for being longform.

I think the most enlightening thing about AI is that it shows you how even the most mediocre text inspires some sort of emotion. Soulless LinkedIn slop makes you feel frustration with a person for their lack of authenticity, but you can still imagine how they forced it out of their heads. You still connect with them, even if it’s in a bad way.

AI copy is dead. It is inert. The reason you can spot it is that it sounds hollow. I don’t care if a website says stuff on it because I typed in, just like I don’t care if it responds in a way that sounds human, because it all feels like nothing to me. I am not here to give a website respect, I will not be impressed by a website, nor will I grant a website any extra credit if it can’t do the right thing every time. The computer is meant to work for me. If the computer doesn’t do what I want, I change the kind of computer I use. LLMs will always hallucinate, their outputs are not trustworthy as a result, they cannot be deterministic, and any chance of any mistakes of any kind are unforgivable. I don’t care how the website made you feel: it’s a machine that doesn’t always work, and that’s not a very good machine.

I feel nothing when I see an LLM’s output. Tell me thank you or whatever, I don’t care. You’re a website. Oh you can spit out code? Amazing. Still a website.

On a personal note, the style choices here make me wonder what the fuck style guide he's using. Spaces around emdashes suggest AP Style, yet hyphenating adverbs ... who knows?

So you thought you’d just read that webpage and then go back to the previous page? A bold assumption. All too often, clicking the back button in your browser doesn’t actually take you back. It’s called back button hijacking, and Google has thus far tolerated it. That ends in June, when the company will designate it a “malicious practice,” and any site continuing to do it will face consequences.

Back button hijacking is a way of wringing more pageviews out of visitors. It’s common on sites that live and die on search traffic. You may end up on a page because it looks like something you want, but instead of letting you leave the domain, it manipulates your page history to insert something else when you click back.

The phantom page is usually a collection of additional content suggestions or a pop-up that tries to eke out a few more clicks from each visitor. Some sites get a little more creative with it, though. For example, LinkedIn has a nasty habit of sending you “back” to the social feed after you land on a link to a profile or job posting.

Google says the back button should always do what you expect it to do—go back. Anything else amounts to a deceptive user experience that can discourage users from visiting unfamiliar pages in the future.

If you’ve been waiting for Microsoft to update its Surface PC lineup—perhaps with Qualcomm’s new Snapdragon X2 Elite processors—I’ve got bad news for you. Microsoft is shaking up its PC lineup, but it’s doing so by instituting big price hikes. This means you’ll be paying at least $1,500 for Surface devices that launched at $1,000 just two years ago and that Microsoft no longer offers new Surface devices under $1,000 at all.

The 12-inch Surface Pro tablet that originally started at $799 and the 13-inch Surface Laptop that launched at $899 now cost $1,049 and $1,149, respectively, a $250 price increase. The higher-end Surface Laptop and 13-inch Surface Pro from 2024 both started at $999 but increased to $1,199 in 2025 when their entry-level versions with 256GB of storage were discontinued; both now start at $1,499, a $300 increase.

As originally reported by Windows Central, Microsoft is blaming “recent increases in memory and component costs” for the price hikes. Supply shortages for RAM and storage chips in particular have been wreaking havoc with consumer tech all year, delaying some launches, depleting the stock of existing products, and raising prices for small and large companies alike.

I'm rather concerned about what I do when my Surface Pro 7 dies. I inherited a Chromebook from my dad, but that's a poor substitute.

archive.is link

This month, USA Today published an excellent report that revealed how US Immigrations and Customs Enforcement delayed disclosing key information about the impacts of its detainment policies. The authors used the Internet Archive’s Wayback Machine to compile and analyze detention statistics from ICE and track how the agency had changed under the Trump administration. The story is one of countless examples of how the Wayback Machine, which crawls and preserves web pages, has helped preserve information for the public good. It was also, Wayback Machine director Mark Graham says, “a little ironic.”

USA Today Co., the publishing conglomerate formerly known as Gannett that runs both its namesake paper and over 200 additional media outlets, bars the Wayback Machine from archiving its work. “They're able to pull together their story research because the Wayback Machine exists. At the same time, they're blocking access,” Graham says.

A number of other major journalism organizations have also recently moved to restrict the Wayback Machine from archiving their stories, including The New York Times. According to analysis by the artificial-intelligence-detection startup Originality AI, 23 major news sites are currently blocking ia_archiverbot, the web crawler commonly used by the Internet Archive for the Wayback project. The social platform Reddit is too. Other outlets are limiting the project in different ways: The Guardian does not block the crawler, but it excludes its content from the Internet Archive API and filters out articles from the Wayback Machine interface, which makes it harder for regular people to access archived versions of its articles.

Ken, a copywriter for a large, Miami-based cybersecurity firm, used to enjoy his job. But then the “workslop” started piling up.

Workslop is an unintended consequence of the AI boom. It’s what happens when employees use AI to quickly generate work that seems polished – at least superficially – but is in fact so flawed or inaccurate that it needs to be heavily corrected, cleaned upor even completely redone after it’s passed on to colleagues.

For Ken, the problem started after his company’s CEO laid off several of his colleagues and mandated that remaining workers use AI chatbots, saying it would boost their productivity. While initial drafts were a breeze to create, Ken and his co-workers had to spend more time rewriting, correcting errors and resolving disagreements between each other’s chatbots than if they had never used AI at all.

“Quality decreased significantly, time to produce a piece of content increased significantly and, most importantly, morale decreased,” said the copywriter, who spoke under a pseudonym for fear of losing his job. “Everything got a whole lot worse once they rolled out AI.” Ken said the company’s executives shifted the blame to staff when they pushed back about AI-fueled productivity decreases.

Gut writing and editing staff, insert hallucinatory LLMs. That's got to be great for the product, right?

The United States FCC recently announced a ban on new consumer-grade routers produced outside of the US. This does not affect existing devices that were already authorized, and there is a carve-out for manufacturers to apply for a conditional approval. It's difficult to say what the medium or longterm effects of the ban will be.

This got me thinking about what could be used as a makeshift router in a pinch. As it so happens, any computer that can run Linux and has networking interfaces can function as a router. This blog post by Noah Baily documents the process using various old computers and components as custom routers over the years.

These makeshift routers are not going to win any bandwidth speed races, but they're perfectly capable of routing traffic for IoT devices or basic browsing. They're also useful for capturing traffic to analyze or sharing internet access from WiFi to Ethernet or vice-versa.

This guide documents the setup process and capabilities of using a Raspberry Pi as a router. It does not require a particularly powerful computer, even the older Pi 3 B+ that lots of us have tucked away in an old parts bin works fine for this.