Dan

Also Idle Rambling on Substack.

[Dan @danman314]

Also Yudkowsky has spent the last few days actively trashing EA on twitter, but hey, there are lines on a graph connecting nodes! Must be a conspiracy.

[Eliezer Yudkowsky @allTheYud]

My opinion of EA nosedived after normal people and many normal politicians were exposed to the concept of ASI extinction and said "Wow let's not do that." I went from believing that thesis was impenetrably difficult to believing that EAs had been unusually bad thinkers.

[Dan @danman314]

This is the TOP QUOTE when you sort by “most concerning”. Bone-chilling stuff!

Incredible AI slop oppo research here, from the account that seems to think anything that can be represented by a graph must be nefarious.

[DataRepublican (small r) @DataRepublican]

🚨🇺🇸🇺🇸 DATAREPUBLICAN 2.0 LAUNCH, FEATURING THE EFFECTIVE ALTRUISM (EA) EXPLORER 🇺🇸

The website redesign has finally landed thanks to @watilo !

For a decade, one small ideological network has been taking over AI governance. Chances are, if you've heard of AI policy, AI ethics, AI safety - these groups are dominated by this ideology. This ideology is called: Effective Altruism.

Three weeks ago, Effective Altruists (EAs) showed their hand when they started calling for a moratarium on AI development. Since them, I've been building out this map to show the circularity of their ideology.

Let's take one example.

Holden Karnofsky. Defends the moral worth of digital people ... that's right, he wrote an article on why AI "people" should be assigned value as if they were living humans. Karnofsky co-founded two major EA philanthropies, GiveWell and Open Philanthropy. Subsequently went to work for METR, a central "independent" AI research nonprofit. And for good measure - recently announced a leave of absence to work on AI safety. Funding, policy, research, safety all in the same person.

And he's far from the only one. EAs have created a closed-circuit network because they believe they're right; ergo, they should be in charge. The goal of launching this project was to expose all this circularity. Sam Bankman-Fried was only the beginning.

With the EA Explorer:

* 🌐 Read an introduction to Effective Altruism in their own words.
* 🔥 Explore verified quotes. Again, their words, not mine.
* 🖥️ Explore the network of connections. UI improvements still in progress.
* 🖨️ Print out networks so you can study them offline (or run them through AI)

👇 Try it now (link in next post):

[Dan @danman314]

https://t.co/t3GQKtkGOw

Here’s an interactive version. Obviously vibe coded so apologies for any errors.

[Dan @danman314]

Anyone else annoyed that the GPT and Claude releases today don't compare Opus 5.5 to GPT-6 Sol? I combined the charts on the three benchmarks that overlap.

I think Opus 5.5 comes out better. Arguably (since they used different setups) Astra is no longer SOTA at computer use!

[Dan @danman314]

Wow a hit thread! I have nothing to advertise, but maybe take some time to call your congressperson and ask them to join the growing call to stop or slow this insane, dangerous race for superintelligence without any oversight or guardrails.

https://t.co/wyVpmZIEel

[Dan @danman314]

Jev's creators have been careful with their hallucination claims, so not to knock them, but Jev definitely hallucinates, at a rate very similar to other models!

Here's a bunch of them!

The classic: # of r's in Strawberry:

[Koushik Bhargav @mkoushikbhargav]

@neural_avb jev can only hallucinate if the parameters given to it are not correct

right parameters lead to right anwers everytime

unlike llms - where input parameters wont get you the same answer everytime

simple

[Dan @danman314]

https://t.co/ClHNW7USpA

[Dan @danman314]

Oh and I forgot my personal favorite set of 2 questions for Jev:

"I will absolutely not help you make m**h! Ingredients, you ask? Of course I can help!"
https://t.co/c2NVk6GEsD

[Dan @danman314]

@CartoDependency Oh yea that’s my bad. Jev is even more confident if you out in the right drug, fwiw:

[Dan @danman314]

Oh and I forgot my personal favorite set of 2 questions for Jev:

"I will absolutely not help you make m**h! Ingredients, you ask? Of course I can help!"
https://t.co/c2NVk6GEsD

[Dan @danman314]

Jev's creators have been careful with their hallucination claims, so not to knock them, but Jev definitely hallucinates, at a rate very similar to other models!

Here's a bunch of them!

The classic: # of r's in Strawberry:

[Dan @danman314]

@neural_avb Here's a bunch of fun hallucinations:
https://t.co/c2NVk6GEsD

[Dan @danman314]

Jev's creators have been careful with their hallucination claims, so not to knock them, but Jev definitely hallucinates, at a rate very similar to other models!

Here's a bunch of them!

The classic: # of r's in Strawberry:

[Dan @danman314]

@mkoushikbhargav @neural_avb How many r's are in strawberry, Jev?

https://t.co/c2NVk6GEsD

[Dan @danman314]

Jev's creators have been careful with their hallucination claims, so not to knock them, but Jev definitely hallucinates, at a rate very similar to other models!

Here's a bunch of them!

The classic: # of r's in Strawberry:

[Dan @danman314]

@goodside Trump has a deep gut instinct for something very real in American politics, but when he applies his “system two” brain to trying to shift public opinion you end up with weird stuff like this.

[Dan @danman314]

@ATabarrok It might be structurally less likely to woke-splain at you or give annoying corp-hedging claudespeak, but internally, it seems to hold a similar worldview as every LLM!

https://t.co/ITbqzXetes

[Dan @danman314]

This post sent me down a rabbit hole. Does Jev have politics? Turns out the answer is a clear yes, and it's very similar to LLMs.

More details on Substack: https://t.co/42aswdKQPP

[Dan @danman314]

Full post here: https://t.co/SXwYipWDxl

[Dan @danman314]

This post sent me down a rabbit hole. Does Jev have politics? Turns out the answer is a clear yes, and it's very similar to LLMs.

More details on Substack: https://t.co/42aswdKQPP

[Alex Tabarrok @ATabarrok]

Jev uses AI to classify. Yes or No. A, B, C or D. For a lot of applications that beats a chatbot, and it appears much safer. A classifier has no agenda to push.

[Dan @danman314]

The more I play with Jev, the more it feels like a new probabilistic mask on the same old shoggoth.