"The War on General Purpose Computing."

people engage when they have autonomy, competence, and purpose.

Boalar’s research confirms this: when work is relevant and challenging, cheating drops dramatically.

AI cheating highlights that much of what passes for education today has no value.  So let AI burn down and reveal how inane this work is.

Technological innovation is outpacing our ability to thoughtfully apply it

opinionated

L’esperimento mentale è stato inizialmente pubblicato da Less Wrong — un forum e blog su razionalità, psicologia e intelligenza artificiale in generale — nel luglio 2010 da un utente chiamato Roko. Al livello più base del tutto, l’esperimento mentale parla delle condizioni in cui sarebbe razionale per una futura super-intelligenza artificiale uccidere gli umani che non hanno contribuito alla sua creazione.

se una super-intelligenza decide le sue azioni per fare sì che siano quanto più positive possibili per il benessere degli esseri umani, allora non si fermerà mai perché le cose possono sempre andare un po’ meglio.

Chiunque non stia lavorando per creare quella macchina si sta frapponendo al progresso e dovrebbe essere eliminato per riuscire a raggiungere più rapidamente quell’obiettivo.

ora che lo avete letto, siete tecnicamente coinvolti in esso. Non avete più una scusa per non contribuire alla creazione di questa super-intelligenza artificiale e se scegliete di non farlo, sarete uno dei primi obiettivi dell’intelligenza artificiale.

Quando Roko ha pubblicato la sua teoria del basilisco su Less Wrong, Yudkowsky si è piuttosto spazientito, a tal punto da cancellare il post e vietare ogni tipo di discussione sul basilisco dal forum per cinque anni. Quando Yudkowsky ha spiegato le sue azioni sulla subreddit Futurology qualche anno dopo il post originale, ha affermato che l’esperimento mentale di Roko dava per scontato di riuscire a superare una serie di ostacoli tecnici nella teoria delle decisioni, e anche se ciò fosse possibile, Roko stava efficacemente diffondendo delle idee pericolose, proprio come i cattivi della storia di Langford.

31% of responses showed serious sourcing problems – missing, misleading, or incorrect attributions.

20% contained major accuracy issues, including hallucinated details and outdated information.

Imagine if every time you got into your car, you had to manually approve the engine's use of oil, the tires' use of air, and the radio's use of electricity. It’s absurd, right? You’d set your preferences once, and the car would just work.

It Doesn't Actually Give Us Control: The illusion of choice is not choice. When the options are "Accept All" or "Spend Five Minutes in a Menu of Legalese," the system is designed to push you toward the path of least resistance.

COLUMN "It's always DNS" is a long-standing sysadmin saw, and with good reason: a disproportionate number of outages are at their heart DNS issues. And so today, as AWS is still repairing its downed cloud as this article goes to press, it becomes clear that the culprit is once again DNS. But if you or I know this, AWS certainly does.

• June, 2025: the Apple reasoning paper confirmed that even with “reasoning”, LLMs still can’t solve distribution shift, the core Achille’s heel in neural networks that I have been writing about for nearly 30 years.

But with conventional software, someone with inside knowledge can usually deduce what’s going on,

worked for a dozen years — will have a good idea why. “Here’s what really terrifies me” about the current breed of artificial intelligence (AI), he says: “there is no such understanding”, even among the people building it.

Martin Wattenberg, a computer scientist at Harvard University in Cambridge, Massachusetts, says that understanding the behaviour of LLMs could even help us to grasp what goes on inside our own heads.

stochastic parrots

some say more is going on, including reasoning and other startlingly human-like abilities

The researchers described the model’s behaviour as role-playing — doing more than parroting but less than planning.

When they asked their LLM whether it consented to being shut down, they found it drew on several source materials with the theme of survival to compose a compelling response (see ‘Lust for life’).

trained an LLM from scratch to play the board game Othello,

The team successfully trained a smaller model to interpret the internal activations of the AI, and discovered that it had constructed an internal map of the discs based on the text descriptions of the gameplay2

Because chatbots can chat, some researchers interrogate their workings by simply asking the models to explain themselves. This approach resembles those used in human psychology. “

The researchers first intentionally biased their study models by, say, giving them a series of multiple-choice questions for which the answer was always option A. The team then asked a final test question. The models usually answered A — whether correct or not — but almost never said that they chose this response because the answer is usually A

“It’s a little weird to study [LLMs] the way we study humans,” Bau says. But although there are limits to the comparison, the behaviour of the two overlaps in surprising ways.

“It is nonsensical to say that an LLM has feelings,” Hagendorff says. “It is nonsensical to say that it is self-aware or that it has intentions. But I don’t think it is nonsensical to say that these machines are able to learn or to deceive.”

Developers train an LLM on large quantities of text to repeatedly predict the next text fragment, a process called pre-training. Then, when the LLM is given a text prompt, it generates a continuation. Offered a question, it predicts a plausible answer. Most LLMs are then fine-tuned to align with developers’ goals

the interface might append a ‘system prompt’ to each user prompt

external documents

LLMs don’t copy-paste (or cut and paste) code. For instance, when you ask them to refactor a big file into smaller ones, they’ll "remember" a block or slice of code, use a delete tool on the old file, and then a write tool to spit out the extracted code from memory. There are no real cut or paste tools.

Is there a seahorse emoji? Let's ask GPT-5 Instant:

The real work usually happens alongside coding, as the developer learns the domain, narrows down requirements, maps out relevant abstractions, considers side effects, tests features incrementally, and finally squashes bugs that survived this rigorous process. It looks something like this:

most software lives within complex systems, and since LLMs can't yet hold the full context of an application in memory at once, human review, testing, and integration needs will remain