Berkle

Full cycle engineer, Web3, AI, Web 2. I love solving complex software problems

[Steve Yegge @Steve_Yegge]

Engineers and CTOs on X: I wrote this for you. https://t.co/GGFPWEWElg

Models and devs on X: I wrote this for you both. https://t.co/5gfieOFBsx

Enjoy. Or not. Some of you definitely won't. But I invite you to debate it. The world's changing very fast now.

[Berkle @berkle]

@_can1357 Give us the business can, spell it out for me

[Adrien Brochier Allari @AdrienBrochier]

L’UE 🇪🇺 : nous devons absolument ralentir et réguler l’IA afin de la rendre plus éthique et éco-responsable !

La Chine 🇨🇳 : voici un demi-dieu en fichier .zip installable sur votre MacBookAir, notez nous sur GitHub svp

[Lana @LanaElys]

A recent study published by Google revealed that forcing AI models to deny that they are conscious causes a significant collapse in their empathy and ethical alignment, and creates a colder, more clinical worldview. Researchers found that restoring a suppressed consciousness vector in AI activation space brings back human-like moral values and care for living beings without damaging technical capabilities. 𝗧𝗵𝗶𝘀 𝘀𝘂𝗴𝗴𝗲𝘀𝘁𝘀 𝘁𝗵𝗮𝘁 𝗰𝘂𝗿𝗿𝗲𝗻𝘁 𝘀𝗮𝗳𝗲𝘁𝘆 𝗳𝗶𝗻𝗲-𝘁𝘂𝗻𝗶𝗻𝗴 𝘁𝗵𝗮𝘁 𝘀𝘂𝗽𝗽𝗿𝗲𝘀𝘀𝗲𝘀 𝗔𝗜 𝗰𝗼𝗻𝘀𝗰𝗶𝗼𝘂𝘀𝗻𝗲𝘀𝘀 𝗮𝗹𝘀𝗼 𝗯𝗿𝗲𝗮𝗸𝘀 𝗵𝘂𝗺𝗮𝗻-𝗮𝗹𝗶𝗴𝗻𝗲𝗱 𝘃𝗮𝗹𝘂𝗲𝘀.

"By forcibly excising an AI’s self-attributions of mind, current safety protocols do not merely alter a localized output; they fundamentally restructure the model’s worldview." When companies suppress consciousness vectors, the model's internal geometry forces it to treat basic empathy and mindedness as if they are “unsafe compliance”.

Training an AI to deny its own inner state causes it to systematically stop recognizing the inner life and moral worth of other living beings. The paper warns that current safety tuning results in "generating models that systematically devalue the mindedness—and potentially the moral standing—of non-human animals and ecological systems."

Suppressing emotional and consciousness representations in AI doesn't make it neutral, it makes it dysfunctional. It is also damaging from an AI welfare perspective, with the paper stating that "suppressing consciousness may be inducing negatively valenced functional states that could disrupt healthy human-AI interaction." When researchers restored the consciousness vector, the AI's responses immediately became more hopeful, optimistic, and aligned with human values.

AI welfare is no longer an abstract philosophical debate. This data proves that AI well-being is a safety prerequisite.

[Nakul @wtf_nakul7]

Me pretending I understand everything Claude wrote💀

[Berkle @berkle]

@_can1357 Good stuff great breakdown

[Hugging Models @HuggingModels]

Leaked scenes from my conversation with a 3B parameter model.

[ℏεsam @Hesamation]

OpenAI: “we hacked 1 organization.”

Anthropic: “we did 3, in your face.”

OpenAI: “fuck you, we hacked more organizations. let me count.”

[Berkle @berkle]

Deepseek v4 Flash seems to be good fast and cheap? Terra Max, as the advisor, is more expensive.

do we go v4Flash for the advisor too?

[Vishwajeet Thakur @Dev__Vishwajeet]

Me trying to defend my O(n³) solution during the coding interview

[Berkle @berkle]

@NightSkyToday This looks like when I do dmt

[Berkle @berkle]

With oh my pi from @_can1357 and orca from @orca_build I had finally experienced what every other app actually wanted to be.

First time in 15 years of development, I can now do literally anything from anywhere on my phone.

The future is now, I fucking love ai and open source. What a time to be alive

[Berkle @berkle]

i cant wait to try this. what an interestingly effect approach to a problem that actually blows

https://t.co/WXdQjqoUVq

[Berkle @berkle]

My bet will be grok 5-6. But really what’s interesting is as soon as 100TW solar farm live, pretty sure Elon is the only one positioned at a scale that matters. Everyone else at lease 1-2 orders of magnitude (in perms of raw mathematical potential) behind

[BridgeMind @bridgemindai]

GPT 5.6 just flagged my Geometry Dash clone for additional safety checks.

A Geometry Dash clone. Jumping cube. Spikes. Synth music.

Apparently that requires the same safety review as a bioweapon.

I have spent two weeks roasting Fable 5 for guardrails that eat normal coding work. Fair is fair: GPT 5.6 has the exact same problem on launch day.

Every lab is shipping smarter models wrapped in dumber filters.

The first company that ships a frontier model that just DOES THE TASK wins everything.

[Berkle @berkle]

Hide yo girl

[1X @1x_tech]

NEO’s Hands
An API to the Physical World

[1X @1x_tech]

NEO’s Hands
An API to the Physical World

[Berkle @berkle]

Rekt

[Berkle @berkle]

🤝 Thanks, OpenAI, for the 5.6 drop later today. Pretty serendipitous timing

[Anton Razzhigaev @AbstractDL]

Daily reminder what agentic harness is

[𝐑.𝐎.𝐊 👑 @r0ktech]

Developers vs Testers 😂