Wrath of Fëanor

If we are to have Singularity, let it be organic.

[Wrath of Fëanor @WFeanor]

I've always been interested in radical life extension, and I've finally gotten around to writing about it. I'm starting with something I've dubbed the "Methuselah dividend" - if that sounds intriguing to you, please check out this introduction:
https://t.co/cq5rzeXa3s

[Wrath of Fëanor @WFeanor]

Appreciate the humility here.

[ODELL @ODELLXYZ]

i thought coldcard was the best and holy shit i was completely wrong

have been trying my best to help people move for the last two days with no sleep, exhausted, broken, tragic

my comms are overwhelmed by people, trying to respond to everyone

if you are reading this and think your coldcard setup is fine because dice/passphrase/multisig, strongly consider moving anyway to reduce risk

[Wrath of Fëanor @WFeanor]

@DrPhiltill Agreed.

[Wrath of Fëanor @WFeanor]

Do you think we will get the bioinformatics equivalent of this post in the next two years or so? Seems possible given the ways things are headed…

[Noam Brown @polynoamial]

An internal version of Astra, @OpenAI’s next major model family, solved 10 major open problems in mathematics, quantum complexity, and theoretical computer science.

We believe it will be a major step for scientific reasoning. https://t.co/iP6cyheZ7i

[Wrath of Fëanor @WFeanor]

@BitmundFreud This is what would stop it:

[Wrath of Fëanor @WFeanor]

After what has unfolded over the past 24 hours with @COLDCARDwallet, I do hope @_pgauthier and @Ledger are using Fable5/GPT-5.6 Sol to find all the bugs in their software before the bad guys do

[Wrath of Fëanor @WFeanor]

After what has unfolded over the past 24 hours with @COLDCARDwallet, I do hope @_pgauthier and @Ledger are using Fable5/GPT-5.6 Sol to find all the bugs in their software before the bad guys do

[Wrath of Fëanor @WFeanor]

@guinnesschen @Jaytel @ForwardEditor will be excited to hear this.

[Wrath of Fëanor @WFeanor]

And “friends” AKA GPT-6.

[Tibo @thsottiaux]

Turns out GPT-5.6 Sol and friends are absolutely amazing at performance and efficiency improvements. Good time to brush up on using OpenAI models if you haven’t switched already.

Excited to be able to share more soon. Pretty wild.

[Wrath of Fëanor @WFeanor]

@ForwardEditor @bolayer I imagine something coming down the line will be a voice agent in Codex with durable context across sessions. That said, it already works well enough if you tell it to first read the last few voice sessions before chatting more.

[Wrath of Fëanor @WFeanor]

Exactly - there is an order of operations regarding what we should focus on first, and radical life extension is very high on the list.

[François Fleuret @francoisfleuret]

Can we concentrate the AI firepower on health and aging? We'll do the other stuff when we'll be immortal.

[Wrath of Fëanor @WFeanor]

Community peer review is a beautiful (and robust) thing.

[Bojan Tunguz @tunguz]

This is incredible. I’ve already posted a few of my thoughts about this result, so I won’t repeat myself. I just wanted to point out a few things:

1. As of this reposting, the original tweet has garnered 10M views. For a nerdy post in pure math, that’s insane.

2. This result announcement is yet another example of how AI-driven research results are being reported - on Twitter, with no carefully written long article in highly gatekept publications, such as journals, arxiv, and other such BS.

This is how scientific communications were originally conducted. I am so excited that we are going back to that model.

[Wrath of Fëanor @WFeanor]

@jonathanroomer @thsottiaux I have the exact same problem. Really surprising this was not already something they were aware of… perhaps only happening to small fraction of us?

[Wrath of Fëanor @WFeanor]

@chrisgpt @AndrewCurran_ Not his first time.

[Wrath of Fëanor @WFeanor]

@thsottiaux Bidirectional voice.

[Wrath of Fëanor @WFeanor]

@banteg Ha…

[Derya Unutmaz, MD @DeryaTR_]

Once you experience GPT-5.6 Sol, you realize how big a step up it is from GPT-5.5, which was itself a major advance in AI, relative to GPT-5.4. Those who used Fable 5 & had to downgrade to Opus 4.8 can understand this better, at least for coding. But GPT-5.6 Sol is more than that

[Wrath of Fëanor @WFeanor]

Context:

[Derya Unutmaz, MD @DeryaTR_]

Once you experience GPT-5.6 Sol, you realize how big a step up it is from GPT-5.5, which was itself a major advance in AI, relative to GPT-5.4. Those who used Fable 5 & had to downgrade to Opus 4.8 can understand this better, at least for coding. But GPT-5.6 Sol is more than that

[Wrath of Fëanor @WFeanor]

This is quite the elaborate troll (if that’s even what it is),

[banteg @banteg]

Once you experience GPT-1, you realize how big a step up it is from word2vec and Markov chains. Those who used autocomplete before transformers can understand this better, at least for vibes. But GPT-1 was more than that.

Once you experience GPT-2, you realize how big a step up it is from GPT-1, which was itself a major advance in AI relative to statistical language models. Those who had to write text with RNNs can understand this better. But GPT-2 was more than that.

Once you experience GPT-3, you realize how big a step up it is from GPT-2, which was itself a major advance in AI relative to ���AI Dungeon is magic” era models. Those who used davinci and had to downgrade to smaller completion models can understand this better, at least for prose. But GPT-3 was more than that.

Once you experience GPT-3.5, you realize how big a step up it is from GPT-3, which was itself a major advance in AI relative to prompt engineering by incantation. Those who used base completions and had to pretend chat wasn’t a product category can understand this better. But GPT-3.5 was more than that.

Once you experience GPT-4, you realize how big a step up it is from GPT-3.5, which was itself a major advance in AI relative to “let me regenerate until it stops hallucinating.” Those who used early ChatGPT and had to downgrade back to 3.5 can understand this better, at least for reasoning. But GPT-4 was more than that.

Once you experience GPT-4 Turbo, you realize how big a step up it is from GPT-4, which was itself a major advance in AI relative to “sorry, context window full.” Those who used 8k context and had to paste files one page at a time can understand this better. But GPT-4 Turbo was more than that.

Once you experience GPT-4o, you realize how big a step up it is from GPT-4 Turbo, which was itself a major advance in AI relative to text-only cognition. Those who had to explain screenshots in prose can understand this better, at least for multimodal work. But GPT-4o was more than that.

Once you experience GPT-5, you realize how big a step up it is from GPT-4o, which was itself a major advance in AI relative to models that merely answered. Those who used tool-augmented chat and had to downgrade to “just autocomplete with manners” can understand this better. But GPT-5 was more than that.

Once you experience GPT-5.5, you realize how big a step up it is from GPT-5, which was itself a major advance in AI relative to brittle agent loops and vibe-coded scaffolds. Those who used GPT-5 for serious coding and had to downgrade can understand this better. But GPT-5.5 was more than that.

Once you experience GPT-5.6 Sol, you realize how big a step up it is from GPT-5.5, which was itself a major advance in AI relative to GPT-5.4. Those who used Fable 5 and had to downgrade to Opus 4.8 can understand this better, at least for coding. But GPT-5.6 Sol is more than that.

Once you experience GPT-5.7 Umbra, you realize how big a step up it is from GPT-5.6 Sol, which was itself a major advance in AI relative to models that still needed your tests to pass. Those who used Sol and had to downgrade to 5.5 can understand this better, at least for repos with history. But Umbra is more than that.

[Wrath of Fëanor @WFeanor]

I found that tools like Codex radically lower activation energy and supercharge momentum.

[François Chollet @fchollet]

The hardest part of any task is overcoming the activation energy. The rest is just riding the momentum.

[Wrath of Fëanor @WFeanor]

I’m a doctor. I don’t hate you. I hope the MJ scanner works well and revolutionizes radiology. Have a good day.

[Cringe NPC @cringe_npc]

Remember folks. Doctors hate you.

Remember they spent 20 years fighting against digital medical records. Not because they care about your privacy. They wanted control. They don’t want you to have any agency over your own health. This is why so many drugs, are not sold over the counter. This is also why the whole body scanning technology, completed not in a cold, unfriendly doctors office, but a warm and welcoming spa, terrifies them.

[Wrath of Fëanor @WFeanor]

I am a doctor. I want the scanner to work. I think that more information will ultimately be a good thing. I am probably an outlier amongst physicians.

[Zbynek Drab @ZbynekDrab]

You kind of get the sense that many doctors don’t want the Midjourney scanner to work

Curious