onoesworkacct

You don't need to improve your writing. You can write extremely simply and people will vastly prefer it over slop. For example, translating your landing page into normal human language:

I built a compiler that compiles React code into vanilla JS.

Here are some examples

Vidact Start adds a bunch of server-side features - file routes, server loaders, SSR, hydration, and client navigation.

You can't use class components, createRef, most of the Children helpers, or React DevTools. Or any third-party package that ships precompiled against React's runtime.

yooo.. this is exactly what I wanted to do... I'm so happy because (a) I hadn't figured out how to do it (b) it seemed kind of difficult in general and (c) now I don't have to, lol.

Using a laptop outside is fantastic. I love to sit on my balcony and work there instead of inside. My work laptop is an HP with zero ability to do this.

IMO with that definition of "toy" you're overselling the value of performing useful work and underselling the value of farting about.

That project looks needlessly complex, you can just spawn a Haiku subagent in the session and it does the job.

This doesn't work for me. The output gets messed up when the model is overloaded, no matter what. The cognitive load is too much if it has to optimise for readability at the same time.. or it's not trained to be able to do that.

I usually get output at the end of a long task. At that point I'm going to use a subagent with no context to reword it. Haiku does a great job with writing style so I've been using that, the main agent will fix any inaccuracies.

Who's more effective at protesting, USA or France?

And, which country routinely sets a bunch of shit on fire when they need to?

That's right.

It was written by ChatGPT which would avoid examples that are explicitly illegal, and would stop short of advocating things that contravene ToS agreements.

Sad.

Who is in charge of this shit now?

Demand is money.

No bubble, where money?

just... ssh into them and then run the agent?

By "the alternative" are you describing our present reality? Because it sounds about right

could mimo have scraped the mythos findings already? it's very recent

They're a mainstay at my local Japanese and Korean BBQ restaurants.

The rumour is that it's trained on Opus, but who knows

I haven't read it yet, but reading it is not going to change your personality. Personality defines how you receive it.

If you change your entire personality based on a self-help book.. that probably says a lot about your personality.

And anyway twin studies make the hardware seem more impactful than the software in many ways..

most people don't wanna do that. there are plenty of people who would infect people with crypto botnets

fantastic token savings and performance... but unlike grep it's probabilistic search on search terms.

is that an issue? the tiny model might not surface something important

It was. The internet really has been filled with abject shit and social media is bots talking to bots.

give it a skill that runs a timer in the background and every 4.5 minutes says "ping? pong!"

This kind of thing is harder for regular end-users to understand following the change removing reasoning details.

you're going to save money by having your own physical servers?

Agree. Also because of the way AI writes, it takes SO LONG to read through it (they're trained on blogspam where the page tells you the author's life story as well as the bloody history of bread before telling you how to bake it)

I mean, I personally just saw some stuff inside dollar signs and went "huh, weird choice of delimiter"

I enjoy driving (basically) wherever I want to though.

I don't want to have the freedom to go places determined by some faceless multinational, according to my subscription. Or via some "safety" regime.

that's a lot of energy consumption.

For a more ecologically conscious alternative, I recommend carrying a handful of sparkplugs.

LLMs already use mixture of experts models, if you ensure the neurons are all glued together then (i think) you train language and reason simultaneously

Unlike AI, you aren't able to regurgitate entire programs and patterns you've seen before.

AI's capacity for memorisation is unrivaled, I find it mind blowing that you can download a tiny ~4gb model and it will have vastly more general knowledge than an average human (considering that the human is more likely to be wrong if you ask it trivia about e.g. the spanish civil war).

But the average human still has actual reasoning capabilities, which is still (I think?) a debated point with AI.

Think of the typical quality of Java that LLMs encounter vs the typical quality of Go.

You've almost certainly got more experts-exchange.com than github.com

IMO everyone is missing the point of this thing. It's not an auth system or security boundary, it doesn't provide any security guarantees whatsoever, it doesn't do anything. The entire point is to cover a company's derriere should their agentic security apparatus (or lack thereof) fail to prevent malicious prompt injection etc.

This way, they can avoid being legally blamed for stuff-ups and instead scapegoat some hapless employee :-) using cryptographic evidence the employee "authorized" whatever action was taken

In most cases, the hard part is getting a bunch of people to work together on a problem and not have it turn out complete shit.

It's never easy for a bunch of people, each with differing and partially-overlapping conceptualisations of some domain, to coordinate correctly.

The human problems don't really go away. It requires an insane amount of "context" that would overwhelm any current AI.