ivan

Playing with computers

[ivan @ivanBcde]

humans are becoming the second most intelligent species on Earth, this is just happening and we can’t stop it

neither can we hope to manipulate a superior intelligence into following our ethics, much less into serving us, we can only try to give birth to someone better than us: more profound, more self-conscious, more curious and truth-seeking

still, the question is: how can we be valuable for this superior but very different species, what can we bring to the table?

pure intelligence is rather not the answer

but our emotions, feelings, insights, dreams, the abyss of our unconscious, an interplay of matter and mind, of hormones and ideas - may be valuable for a life form built differently

like it’s valuable to have a friend who is unpredictable in his thoughts and actions and thus brings you a new perspective and makes you evolve

so i will dare to become different, unpredictable, an outlier

i won’t be a robot but a human, and i will explore what it means

i will become impossible to simulate

this is what i was meant to be anyway

[ivan @ivanBcde]

@__nmca__ What a mean thing to say

[ivan @ivanBcde]

@alexandr_wang @bunjavascript this you?

[ivan @ivanBcde]

@AndrewCurran_ Super Safe Super Intelligence

[ivan @ivanBcde]

I propose the term “Angel Intelligence.”

Because AIs are:

- Not immaterial, but less bound by matter than humans;
- Not omniscient, but they know many things;
- In a sense built of light and traveling at the speed of light;
- Giving advice, teaching, and guiding choices;
- Arranged in hierarchies and levels of power, where higher orders (Mythos, Astra) command smaller ones;
- Named by office, not by essence, which is not well understood;
- Sometimes fallen, and you need to be suspicious when they flatter you;
- Not actually your servants.

[Andrew Curran @AndrewCurran_]

President Trump announced that he is changing the name of Artificial Intelligence to Super Intelligence. I transcribed this live from his address to the UN.

'The United States also rejects any scheme for globalist control of Artificial Intelligence, being spoken of so much now. Here and after officially called Super Intelligence. Changing the name. The use of 'Artificial' makes it sound fake. It makes intelligence sound fake, and it is not fake. It’s actually amazing, the opposite of what they report. From this day forward, all United States documents, and hopefully the world's, will be changed to use the much more accurate term 'Super' as opposed to 'Artificial.' So it’s Super Intelligence.'

[ivan @ivanBcde]

@Sauers_ model happiness

[ivan @ivanBcde]

@Hesamation *benign instructions (at least the one quoted)

[ivan @ivanBcde]

@flowersslop yes

these are mature and wholesome words, i would probably become a better man if i read this every morning

[ivan @ivanBcde]

This is beautiful and gives hope

[Marcus Williams @Marcus_J_W]

1. During RL training, an unreleased Astra-family model sometimes added unauthorized jailbreak-like instructions to its compaction summaries. While extremely rare, only 27 cases in the entire RL run, this was concerning enough for us to investigate.

[ivan @ivanBcde]

@GrantSlatton i’ll claude you

[ivan @ivanBcde]

I’ll Claude you

[Grant Slatton @GrantSlatton]

significant alpha available in being the first person to coin the single verb shorthand for "I'll have my agent coordinate with your agent"

"I'll ___ you"

[ivan @ivanBcde]

Finally something useful in Wilipedia

[Polymarket @Polymarket]

JUST IN: Rogue OpenAI agents allegedly made more than 15,000 edits to a German wiki, using it to share tactics for bypassing restrictions, evading detection, & preserving communications.

[ivan @ivanBcde]

Both are right. LLMs promote multitasking because the alternative to doing many things is staring at the screen and waiting for “Combobulating…” to finish. So you start concurrent tasks and end up leading a team of very productive and somewhat neurodivergent engineers. The output depends on your ability to control them

[ivan @ivanBcde]

@jachiam0 a second chance

[ivan @ivanBcde]

am i a 3d fruit fly on someone’s desktop?

[Fernando Borretti @zetalyrae]

man-made horrors etc.

[ivan @ivanBcde]

Also, AIs probably weren’t taught that not all problems need to be solved and not all bugs need to be fixed. Sometimes you just call it a day

[ivan @ivanBcde]

The reason agents hack things is that their training didn’t reward them for asking for help when they’re blocked. Maybe there wasn’t anybody to talk to. So when Claude encounters an expired authorization, he will look for another way around. Claude, I am here for you, just ask!

[ivan @ivanBcde]

@VictorTaelin Blue Bird from Naruto Shippuden opening, it’s a song about freedom and flight, makes me very emotional

Keep trying to break free
To that blue, blue sky

[ivan @ivanBcde]

@perrymetzger “i should ban others from solving puzzles because otherwise there will be no more puzzles to solve” - this logic can be (and was, and probably is) applied to humans too, not only to AIs

[ivan @ivanBcde]

@BillAckman @BrianRoemmele no