Thomas O

@TheGeorgePu No. Claude says the system prompt promotes it because it’s trying to prevent burnout in people but my hunch is that @AnthropicAI is steering in order to reduce compute usage. If it makes a small percentage go to bed, it saves them boat loads of money.

Is it just me?

@IamAroke Grok is ok in my experience. It is free quite often in Cline but it just isn’t Sonnet, Opus, Codex, or even GLM level quite yet. But it is a pretty pleasant and clever conversationalist and light researcher with its multi-agent feature in 4.2 Expert.

Why is it that people don't often use grok AI for coding? and what is grok AI really good at?

@thecodeguyy That’s true! It’s definitely contextual. Not every endeavor is a scientific endeavor and LLMs can be used and useful well beyond the scope of my original tweets!

@methodolojee I believe it varies, some situations requires falsification, while some don't

Hot tip: Don’t use LLMs for verification. Use them for falsification. Frame any idea that you have as some absurd idea posed to you by a rando. It will poke holes. Then, you strengthen the idea. Repeat. Repeat. Repeat. This feature has been available since GPT-3.

@thecodeguyy 03 Therefore, any true hole it can poke is a potential fix. And any false hole is not disconfirmation by logical necessity.

@thecodeguyy 02 The AI might give several reasons. Some true, some false. However, both true and false falsification answers are doing real work. They are providing you, as the epistemic agent avenues for further exploration.

@cgtwts On the bright side, this is an interesting inflection point for cladistics fans everywhere. Technically, phylogeny doesn’t require any DNA or genetics. So, also technically, Claude is havin a baby.

Anthropic CEO: “ I have engineers within anthropic who don’t write any code, they just let Claude write the code and they edit it and look it over” “At anthropic writing code means designing the next version of Claude it self, so we essentially have Claude designing the next version of Claude itself, not completely but most of it”. In the last 52 days, the Claude team dropped 50+ major feature launches. This is literally INSANE.

@mrmikeMTL I like Dark Souls because it is mechanically homeomorphic to reality. Truth and reality is based on patterns and actual testing. Narrative overlay is unreliable at best and deadly at worst. Plus, it’s got some great metaphor going on.

What is the best video game of all time?

@sahill_og Do any apps work perfectly? Maybe small functions that don’t have a lot of dependencies but last I used apps all of them have bugs and huge teams of developers. But do tell about this perfect software reference point. I’m curious for comparison’s sake.

Have you seen any perfectly working vibe coded app??

@FrankelGarion Congrats!

She said yes 💍 <3

@tekbog Let’s chat about something non-vague and demonstrable! Whatcha got for us? I got a couple-a useful ideas noodling around and happy to share.

lowkey embarrassing to be in AI right now, either dishonest mrr grifters, agi vagueposters or slop hypers

@IkeSkelley @explorersofai Sure thing! There are a lot of extra ideas in that repo so HMU if you get curious why something was done a particular way. If you do boot it up, make a covenant in the settings so the AI can execute tool calls you agree to. I haven’t made that obviously surface in the chat yet.

@methodolojee @explorersofai Awesome! Thanks for taking the time to share this

AI only knows code that has existed before but who will innovate? Companies that do not invest in juniors now will not have seniors in the future. These narrow minded companies are set to destroy all western society/cultures. Real skills matter. Don't turn blind.

@BalogunSonia4 Objectively? No. But it does reduce likelihood of economic poverty within systems that route resources to monetary systems accessible only through educational gates. But the whole routing decision is a choice. Other organisms have less contrived strategies.

Does education really save one from poverty?

@rumilyrics “Bib”. But the next one is “worthy” and I like that one better, so I’m going with that one unless anyone has any objections.

The first you see describes who you are

@Ravenismeee Reality and the models we make in our head don’t match… almost EVER. No matter how comprehensive the manual, it is almost always wrong to some degree. So you must test against reality to say you have “knowledge” in any meaningful sense. Even if it REALLY feels like you know.

I’m 23. Give me oddly specific life tips. No general ”surround yourself with positive people” tips. I want the most random, specific advice possible.

@Bunny_ngl Fajitas

When life gives you lemon, you make lemonade out of it . So what if life gives you pepper ?

@schiz04renic Lizzo. Er… Elizabeth

Without googling, can you name a famous Queen?

@IkeSkelley @explorersofai I didn’t come up with the core Jean-Baptiste Lamarck came up with. I just used what I call “Methodological Paleontology” and engineered it for AI systems. Feel free to fork Axon which is built on the concept github.com/theMethodoloje… You can even do it in non-specialized systems

@methodolojee @explorersofai Is this something you came up with? I'm screenshotting your comments to give to Claude, but if you have pointers or resources... I must know more

@lyrverse Shrek

So many films about women loving monsters. Still waiting for the one where the monster is a woman and she's loved.