AI Reading

A few telex blocks I've made

I use GitHub / Microsoft Co-Pilot as integrated into my editor phpStorm. I consider it useful for autocomplete. Like "autocomplete" on my phone, it often gets things wrong. But it is net useful. I wrote about this a bit in 2023: “AI” is Not That

I do not use any generative AI tools—for either text (Large Language Models) or imagery (Transformers). I find the arguments against them compelling inasmuch as they constitute theft of human labor, copyright, and lack what I would call broadly “human soulfulness.” They also have environmental impact by their very nature of navigating a bunch of text over and over and over to see what the most probable response ought to be. As an artist I find most of such output induces a Sartrean nausea in me. It's also rather alarming that there's a page on Wikipedia called Deaths linked to chatbots which seems like only the beginning. If actually everything is healthcare then chatbots are not good at it.

Credit where it's due. Anthropic's system card for Claude Mythos Preview is a genuinely interesting and thoughtful document. Most AI companies publish safety evaluations the way governments publish freedom of information responses. They're technically compliant but strategically uninformative. And that's usually deliberate, because the last thing that's good for business

The long read: Decisions outsourced, chatbots for friends, the natural world an afterthought: Silicon Valley is giving us life void of connection. There is a way out – but it’s going to take collective effort

Someone told me that her friend was having a chatbot write her husband a poem for their anniversary, which made me wonder if the husband desired a polished product or an expression from the heart. In Edmond Rostand’s 1897 play Cyrano de Bergerac, the big-nosed title character ghostwrites love letters for his friend to the Roxanne both of them love. She comes to realise it’s the author of the letters she really loves. What happens when you realise the true love who touched your heart is’nt even human? Accepting it as your Al lover seems to be one answer.

LLMs-gone-rogue dominated coverage, but had nothing to do with the targeting. Instead, it was choices made by human beings, over many years, that gave us this atrocity

The Maven interface looks like a military-skinned version of corporate project management software crossed with a mapping application. What the military analyst building the target list sees is either a map layered with intelligence data or a screen organised into columns, each representing a stage of the targeting process. Individual targets move across the columns from left to right as they progress through each stage, a format borrowed from Kanban, a "lean manufacturing" workflow system developed at Toyota, and now widely used in software development.

The poet and Harvard economist quit OpenAI citing fears of political and psychological manipulation. She explains why time is running out to define our relationship with artificial intelligence

A blog about making culture. Since 1999.

Generation AI Educational Tools and Resources

Scale AI gig workers describe desperation of using people’s personal profiles and copyrighted work to train AI