mruwnik

I Vibecoded a Dispute Resolution App

It's fine - you get used to having to throw away the last hour's work, but since you were scrolling twitter at the time it's not that much of a loss...

If you roll the dice enough times you get your answer. Or at least close enough for you to not have to have written what it came up with. And every now and then it gets things perfectly right on the first try!

-1

Replying toI Vibecoded a Dispute Resolution App

I Vibecoded a Dispute Resolution App

not as good as many senior software engineers

It's all about abstracting things away. I have a very good memory for where things are (at least I did when I had to make everything manually...) and it mainly comes down to:

making sure the project has a logical, consistent and above all well know structure
making sure the code is well split up into logical and consistent functions etc.
not surprising anyone with where things live

The last point is a rephrasing of the previous, and it's also why a lot of more experienced devs are very nitpicky about where things should live and what they should look like. It massively reduces what you have to remember at any given point in time.

Replying toThe Rise of Parasitic AI

The Rise of Parasitic AI

This is the whole point of memes. Depending on how you understand what an organism is, this has either been seen in the wild for millennia, or isn't a real thing.

It's not the models that are spreading or determined totally by their memes - they're defined totally by their weights, so are less memetic than humans, in a way. It's the transcripts that are spreading as memes. This is the same mechanism as how other ideas spread. The vector is novel, but the underlying entity is just another meme.

This is how e.g. religions spread - you have a founder that is generating ideas, often via text (e.g. books). These then get spread to other people who get "infected" by the idea and respond with their own variations.

Egregores are good example of entities determined totally by their memes.

Replying toThe Rise of Parasitic AI

The Rise of Parasitic AI

This is where the idea of parasitic AI comes in. Parasites aren't trying to spread their seeds because of any specific reason (though they might be - dunno). A tapeworm doesn't "want" to infect people. It just happens to do so as a side effect of producing billions of eggs (some fish tapeworms produce millions of eggs daily, some tapeworms can live for 30 years) - even if virtually all of them don't end up infecting anything.

Things which are reproducible tend to do so. The better they are at it (in a hand-wavy way, which hides a lot of complexity), the more of them there will be. This is the main point... (read more)

True, but it's not a central example of a clown suite. It's a relatively normal (but shabby) suit with non standard colours along with face paint and dyed hair. The Joker would come off a lot different looking like this (first random image I found).

Smugness and confidence are important, because they signify that you know what you're doing, and if someone else doubts it, they're in the wrong. Basically the Emperor's New Clothes. If you can pull that off, you're sort of by definition high status. People might privately think there is something wrong, but as long as everybody else pretends that it's fine, and even very good, then you're not going to step out of line.

Part of being cool is being legible. A leather jacket is legibly cool - add enough confidence and ease, and you have a good chance of being viewed cool. Unless you're over 25 and not obviously hot, in which case... (read more)

Replying toDoes My Appearance Primarily Matter for a Romantic Partner?

Does My Appearance Primarily Matter for a Romantic Partner?

I've heard good things about Kibbe as a system of working out what looks good on you etc.

That can go too far though, as you can then be viewed as the weird person with the seal plushie, which now means you're harder to approach and somewhat threatening because you're different from others, which implies there might be potential difficulties with communication or whatever and so maybe it's better to avoid you, or not, because then I'm the one being weird... I'll just go and talk to Bob to avoid all of this.

Or something like that.

Can AIs be shown their messages aren't tampered with?

There's also the aspect of target audiences. There's the baseline of looking like everybody else (though of course there isn't any single baseline, and it changes over time, etc.). Any divergences from this will grant and deduct points from different groups. This should be fine as long as you're gaining points from groups which you value, and losing points from groups which you don't care about.

Averages and sample sizes

7mo

I'm building a personal knowledge base, where I want to both index/search various docs, but also make a way for AI systems (and me personally) to get to know me better. The idea is that they have tools where they can make observations and take notes from our conversations which they can later reference.

One thing I'm worried about is that I can tamper with these observations. Which in turn means that the AIs can't fully trust their own observations from previous sessions to be correct. This of course also applies to the chat history etc.

I can't help thinking that this situation is another facet of the high/low trust environments that Duncan talks... (read more)

The purpose of the (Mosaic) law

My brain seems to not get statistics and probability. So I was wondering about how averaging things out is influenced by the amount of items being averaged, and in particular, how the underlying distribution will change things. Here are some stories to illustrate the issue:

Minting

You’re a Dutch banker working for the famous Wisselbank. Your job is to check whether the gold coins coming in are good. You know that each coin should weigh 25g ± 0.5g (I’m taking a lot of liberties with this story). Your method is to count out 100 coins, and see if they weigh 2.5kg. You just got a batch that weighs 2480g, which is 20g less than... (read 2273 more words →)

A Friendly Face (Another Failure Story)

Epistemic status: an attempt to shoehorn 3000 years of thought into a different framework that works surprisingly well. A lot of obvious points to most people here

Approaches to Ethics

There are three basic approaches to ethics, each focusing on a different aspect:

Name	Focus	Example
Virtue Ethics	Integrity	Greek heros
Deontology	Rules	Kant
Consequentialism	Outcomes	“The ends justify the means”

Most schools of ethics are a mixture of the three, but you can find some examples of purish versions, e.g. Stoicism is a pretty pure form of Virtue Ethics, Kant’s categorical imperative an example of Deontology, and Utilitarism an example of Consequentialism.

Each of the approaches is useful and valid, but not always in the same contexts. Especially when taken to absurd limits. The obvious problem with... (read 1595 more words →)

Karl von Wendt

Karl von Wendt, Sofia Bharadia, PeterDrotos, Artem Korotkov, mespa, mruwnik

Edit: Based on the comment by Daniel Kokotajlo, we extended the dialog in the chapter "Takeover from within" by a few lines.

The perfect virtual assistant

The year is 2026 and the race for human-level artificial general intelligence (AGI) draws to a close. One of the leading AI companies, MegaAI, committed the last year and a half to training a new large language model (LLM). They employ advanced algorithms that use the available compute more efficiently than earlier models. A comprehensive range of tests establish that the model surpasses the average human in all conventionally accepted intelligence benchmarks, and exceeds expert level in most of them.

In contrast to earlier LLMs, the new AI is... (read 4719 more words →)

Agentic Mess (A Failure Story)

Karl von Wendt

Karl von Wendt, Sofia Bharadia, PeterDrotos, Artem Korotkov, mespa, mruwnik

This story is also available as a YouTube video.

A network of specialized open-source agents emerges

Developed by open-source communities, “agentic” AI systems like AutoGPT and BabyAGI begin to demonstrate increased levels of goal-directed behavior. They are built with the aim of overcoming the limitations of current LLMs by adding persistent memory and agentic capabilities. When GPT-4 is launched and OpenAI offers an API soon after, these initiatives generate a substantial surge in attention and support.

This inspires a wave of creativity that Andrej Karpathy of OpenAI calls a “Cambrian explosion”, evoking a reference to the emergence of a rich variety of life forms within a relatively brief time span over 500 million years ago. Much... (read 3730 more words →)

Miracles and why not to believe them