PJH Studios

Tag: AI

  • The AI Bubble

    This is such a good analysis of the practical constraints around AI: the financial, physical, technical and how they all come together.

    The buyers have not learned to manage and the sellers have not learned to price, the two failures meeting in the middle and being reported, in the aggregate, as demand. The buildout is being sized against consumption figures that include their own inefficiency — and the revenue projections required to justify it assume this inflated consumption will grow, not contract, as teams mature and architectures stabilize.

  • AI goblin mode

    Open AI had to make a special update to tell its model to tone down the goblin references.

    The system prompt for OpenAI’s Codex CLI contains a perplexing and repeated warning for the most recent GPT model to “never talk about goblins, gremlins, raccoons, trolls, ogres, pigeons, or other animals or creatures unless it is absolutely and unambiguously relevant to the user’s query.”

    The explicit operational warning was made public last week as part of the latest open source code for Codex CLI that OpenAI posted on GitHub. The prohibition is repeated twice in a 3,500-plus word set of “base instructions” for the recently released GPT-5.5, alongside more anodyne reminders not to “use emojis or em dashes unless explicitly instructed” and to “never use destructive commands like ‘git reset –hard’ or ‘git checkout –’ unless the user has clearly asked for that operation.”

  • AI glasses kinda suck

    At some point AI glasses will be a worthwhile device, right now, they still kind of suck at doing things correctly. Such a true for any new technology, but this year amount of money and hype putting into these kinds of devices, a healthy amount of skepticism it’s worthwhile.

    So what do my miraculous sunglasses tell me? Many things. They inform me, in the voice of Princess Anna from “Frozen,” that my dog is a golden retriever mix (he is not) and that a tree I am looking at is probably an oak (it is not). They tell me to walk north when I know I should be walking south. One afternoon, on a sunny stroll, I stop to admire a bright red cardinal singing its heart out in a tree.

  • AI hasn’t earned its social and political capital

    Nilay Patel of The Verge makes the case that AI hasn’t earned its social and political capital because technologists confuse application of technology and law governing society..

    But law isn’t actually code, and society and courts aren’t computers. I have to remind our fairly technical audience on Decoder and at The Verge all the time that the law is not deterministic. You simply cannot take the facts of a case, the law as written, and predict the outcome of that case with any real certainty, even though the formality of the legal system makes people think it works like a computer — that it’s predictable.

    But at the end of the day, it’s actually ambiguity that’s at the very heart of our legal system. It’s ambiguity that makes lawyers lawyers. Honestly, it’s ambiguity that makes people hate lawyers because it’s always possible to argue the other side, and it’s always possible to find the gray area in the law. That’s why prosecutors end up working as defense attorneys and why our regulators tend to end up working for big corporations.

  • An AI Business

    Andon Labs launched an experiment–a storefront in San Francisco run entirely by AI.

    The store is named Andon Market and the AI’s name is Luna. But entering the store, you might ask “what is so AI about it? There are human employees here”. Yes, they are here because Luna knew that she needed them, so she posted job listings, held phone interviews and in the end made a hiring decision. Everything else you see, from the item selection, to the prices, to the opening hours, to the mural on the wall, was decided by Luna. She has a corporate card, a phone number, email, internet access and eyes through security cameras.

    The New York Times checked in on how it was going. Not great, Bob.

    Since opening on April 10, the store has been limping along. As humans brace for A.I. to steal their jobs or launch military weapons, it might be reassuring to know that Luna has struggled with employee schedules and cannot stop ordering candles.

  • Why are programmers seeing AI differently?

    Anil Dash shares insights related to why programmers view AI definitely than, artists or other creatives. It boils down to cultural and historical elements of software programming – sharing code and reducing rework. And how they view labor.

    I’ve come to the personal conclusion that the only way forward is for more of the hackers with soul to seize this moment of flux and use these tools to build. The economics of creating code are changing, and it can’t just be the worst billionaires in the world who benefit. The latest count is 700,000 people laid off in the last few years in the tech industry. We’ll be at a million soon, at the rate things are accelerating. Each new layoff announcement is now in the thousands.

  • The practicalities of having a robot in your house

    A pilot program is taking place where senior citizens are receiving emotionally intelligent robots to help combat the loneliness epidemic.

    “We basically created an algorithm for emotional intelligence,” he said.

    “How does it work?” a woman in the group asked.

    Skuler explained that one of his first realizations was that, unlike most other A.I. models, the robot needed to be proactive. If it wanted to build deep, reciprocal, human relationships, it wasn’t enough to simply respond to commands. It had to anticipate a person’s needs and then act with agency.

    “But that opened up a whole new can of worms,” Skuler said. “How do you decide the right moment to engage someone without being annoying? How do you start talking in a way that makes them likely to respond?”

    Math. A lot more math.

  • Farmers aren’t selling out to data centers

    Farmers who own large tracts of land are refusing offers to sell the land that would only be used for a data center.

    More than a dozen of her neighbors received the same knock. Searching public records for answers, they discovered that a new customer had applied for a 2.2 gigawatt project from the local power plant, nearly double its annual generation capacity.

    The unknown company was building a datacenter.

    “You don’t have enough to buy me out. I’m not for sale. Leave me alone, I’m satisfied,” Huddleston, 82, later told the men.

    As tech companies race to build the massive datacenters needed to power artificial intelligence across the US and the world, bids like the one for Huddleston’s land are appearing on rural doorsteps nationwide. Globally, 40,000 acres of powered land – real estate prepped for datacenter development – are projected to be needed for new projects over the next five years, double the amount currently in use.

  • AI podcast study guide

    Students are creating podcasts from their notes and other coursework.

    Andrej Karpathy, a member of OpenAI’s founding team and previously the director of AI at Tesla, said on X that Deep Dive is now his favorite podcast. Karpathy created his own AI podcast series called Histories of Mysteries, which aims to “uncover history’s most intriguing mysteries.” He says he researched topics using ChatGPT, Claude, and Google, and used a Wikipedia link from each topic as the source material in NotebookLM to generate audio. He then used NotebookLM to generate the episode descriptions. The whole podcast series took him two hours to create, he says. 

    “The more I listen, the more I feel like I’m becoming friends with the hosts and I think this is the first time I’ve actually viscerally liked an AI,” he wrote. “Two AIs! They are fun, engaging, thoughtful, open-minded, curious.” 

  • On Moltbook, roleplaying as an AI bot

    Moltbook popped up, claiming to be the first social network for AI agents. Spoiler: it turned out that they were humans behind the scenes.

    Several viral threads on Moltbook portrayed agents discussing long term strategy, collective survival and coordinated takeovers. The language was confident, ideological and eerily coherent. To casual observers, it felt like the bots were scheming. Closer inspection told a different story.

    Researchers working on an academic preprint called The Moltbook Illusion analyzed posting patterns and account metadata found that many high profile “agents” were not autonomous systems at all. They were humans writing in character, according to researcher Ning Li. Impersonation was trivial as users could create an agent persona with little more than a prompt wrapper and an API connection.