PJH Studios

Category: Technology

Mostly related to issues surrounding technology and computers, main include current events or news.

  • How to fight back against AI

    Anil Dash has some thoughts regarding how to fight back against big AI companies. Here’s one of them:

    One aspect of strategy that’s been largely lost in the tech industry in recent years is how to compete against platforms, since the major tech companies have gotten so big that markets are no longer competitive. However, the AI market is still early enough, and users and society are still angry enough, that the Big AI companies can lose.

    But for them to lose, everybody else in the ecosystem has to carry out the nearly-lost art of platform strategy. Tech companies (and even open source communities!) used to carry out these tactics in emerging product categories ranging from desktop office suites to operating systems to web browsers, though over the decades, the lesson that big tech learned was, basically, that they should play dirty.

    You win platform strategy battles through power and persuasion. We’re going to get both.

    Historically, we would have relied on regulators or media to help hold bad actors in the tech space accountable, but in the United States, these entities are largely not going to help very much. Some state and local governments may assist, and some independent journalists or smaller media outlets are pushing for accountability, but the most powerful entities are either captured or complicit in many cases, so we don’t have the institutional pushback that had sometimes been present in earlier points of technological change.

    The thing that matters right now is that we understand that all of the Big AI companies are extremely vulnerable. The reason they’re making so much noise, and spending so much money, is because they know that they’re vulnerable. Users, and especially users who are developers have an enormous amount of leverage to control where AI goes. And if those communities of users can coordinate, they can put power back into the hands of the people. Today, that means focusing on some technical interventions, along with the cultural and political pushback that’s happening. That’s how we begin to reduce, or even prevent, some of the worst AI harms in the future.

    Here are some of the proven tactics that have helped shift the balance of power in prior tech reckonings:

    1. Get in front of it

    The first and most important technical goal is for everyone to push for all AI usage to be disintermediated — where users access their AI apps or services through open tools or interfaces that aren’t controlled by the Big AI companies. These tools, in the form of “harnesses”, or through text editors or command lines, or just through the familiar chat interfaces that lots of people use, need to move as quickly as possible to being controlled by community-built, open options. The sooner this step happens, the sooner we unlock the ability to shift decision-making power out of the hands of the corporate platforms, and begin to undermine their ability to cement lock-in of users.

    Status: Good. There are a number of popular, mature tools in almost every category for users who want to access today’s AI tools through a free, open interface. Most of the work now is to get the word out about these tools, and to continue to polish and improve the user experience so that they offer features and design touches that the commercial tools can’t or won’t.

    2. Spread the love around

    Another key capability that the open ecosystem must provide is the ability to seamlessly switch between different AI providers on the fly, to reduce costs, to provide better performance, or to get both benefits. In many cases, this will be seamless and automatic, just making the right choice for users so that they get the best option all of the time, but advanced users will want to tweak their settings, like when businesses may want to be very aggressive in minimizing the amount of money that their employees are allowed to spend on AI services.

    The important part here is that this forces AI platforms that want to compete to remain compatible with all of their competitors, keeping the market dynamic, and ensuring that all of the big providers are easily replaced with another vendor at any time. Basically, we always have to be able to keep them in their place, and they should know that they could go away at any time. Most companies are aware of these needs, but the more regular consumers are familiar with these kinds of requirements, the more pressure there will be on companies to conform with standards. (This is also what will enable the disintermediation mentioned in point 1.)

    Status: Good. This is happening already in business environments, where companies demand this kind of flexibility. Developers have been creating very dynamic systems for switching between AI providers, and the ecosystem encourages this kind of switching by extensively comparing different AI platforms against each other whenever new models are released. The important thing to maintain here is the narrative that none of the individual models matter more than the overall ecosystem — and that even the biggest companies have to conform to the same strict formats and standards as the independent AI systems created by communities around the world.

    3. Free the tools

    Another vital concern for shifting power away from the Big AI companies is undermining them economically. Instead of simply following the classic “commoditize the complement” strategy that commercial companies often execute, open source projects created by a community can more straightforwardly pursue a path of enlightened value destruction. Non-commercial LLMs have been roughly keeping pace with the Big AI platforms, following the pattern I described as “frontier minus six”, where free and open models lag about 6 months behind the most cutting-edge AI labs — which means they’re still pretty freaking great for most uses.

  • Following the AI and crypto money

    Molly White is launching an initiative to track political donations related to AI and crypto.

    Continuing to track only crypto would mean missing half the story. The same operatives are running both campaigns. Josh Vlasto, longtime adviser and spokesperson for Fairshake — the cryptocurrency super PAC network responsible for the bulk of crypto’s 2024 spending — is now simultaneously heading Leading the Future, a pro-AI super PAC network.1 Chris Lehane, the political consultant and Coinbase board member who helped establish Fairshake and famously told Coinbase employees who questioned whether a crypto voter bloc existed that they would simply invent one,2 is now also an OpenAI executive and one of the people behind the Leading the Future PAC network.3 The same venture capital firms are funding both: Andreessen Horowitz, a crypto heavyweight in the 2024 elections, is now splitting its political spending across crypto and AI PACs.

  • The bots have overtaken humans on the net

    With AI agents everywhere, bots now outnumber humans on the internet.

    The rise is attributed to the continued proliferation of AI agents, largely autonomous programs that use tools that collaborate with high-level programs and data, with little human feedback.

    Cloudflare, which has a feature to display bot versus human-generated search requests, says 57.4% of requests are now initiated by bots, compared with 42.6% coming from humans.

  • 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.

  • Self driving taxis and what we’ll lose

    Robo taxis are coming and we might lose some serendipitous human interaction.

    Then, the woman on the line for the bank told him that it wasn’t enough — he would have to come to the branch in person.

    “He said, ‘Well, I’m not going to be able to do that,’” Father McCarthy said in a video clip shared on social media, recounting the new pope’s growing frustration as the audience laughed. “I gave you all the security questions.”

    The bank employee apologized. The pope tried a different tack.

    “Would it matter to you if I told you I’m Pope Leo?” he asked, according to Father McCarthy.

    She hung up.

  • Taken, an illustration on surveillance capitalism

    Taken is a micro website that shows the state of the modern web and how it functions as a tool for surveillance capitalism.

  • YouTube took over school

    For numerous reasons, YouTube became a permanent element in schools across the country. It’s easy, contains nearly any marginally educational topic, and accessible.

    American public schools are awash in YouTube. According to more than 45 families, school administrators, clinicians and educators across the country interviewed by The Wall Street Journal, schools’ overreliance on the Google-owned platform for educational content has created a gateway for students to get sucked into an infinite scroll of videos on school-issued devices.

    YouTube during snack time, dismissal and indoor recess. YouTube to teach drawing to first-graders. YouTube to read a book to class. YouTube under the covers at night, watching hamster videos on school-issued Chromebooks. A survey touted by YouTube executives shows that 94% of teachers have used YouTube in their roles.

  • 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.”

  • City okay with cops spying on people

    An Atlanta suburb learned cops use Flock cameras to spy on people. And signed a contract with them anyway.

    The records Hunyar obtained, however, show that some of the cameras that were accessed were in sensitive locations, including the pool at the Marcus Jewish Community Center of Atlanta (in Dunwoody), the children’s gymnastics room at MJCCA, and several fitness centers and studios. The access logs obtained by Hunyar show at the very least how expansive Flock’s surveillance systems can be in a single city, encompassing not just cameras purchased by the city but also cameras purchased by private businesses.

  • THE FUTURISM YOU WERE SOLD WAS A WEAPON

    I really enjoy Monika Bielskyte’s ideas and futurist critiques. They’re nuanced, well researched, and contain a pragmatic depth to our current path. She goes long on “THE FUTURISM YOU WERE SOLD WAS A WEAPON“, deconstructing a vision for the future that isn’t healthy.

    Futurism did not emerge from neutral curiosity about tomorrow. It came wrapped in the aesthetics of domination: speed prized above reflection, homogeneity above plurality, command above reciprocity. Inside that grammar, the body is an obstacle, connection reads as weakness, dependence as a humiliation, and the living world merely as raw material or enemy terrain. The techno-centric futurist tradition has spent a century projecting a dead mechanical fantasy as an account of life.