Mohit AK

#mars #ageing_is_a_disease #asi

'anything that is not inherently creative should be #automated'.

[Bindu Reddy @bindureddy]

Open Source Tsuname Is Real - Kimi K2.5 Is The Best OSS Model In The World

I know DeepSeek is preparing to launch their model, but Kimi K2.5 just beat them to it! They are by far the best open-source model in THE WORLD.

Still, there is a considerable gap between them and the Anthropic models, but it's definitely closer than ever before

It is available RIGHT NOW on ChatLLM

[Mohit AK @mohit_ai]

This paper might be as big as the the transformers and maybe lead the path towards singularity.

2512.24880v2.pdf https://t.co/BjurDwIwWr

[Mohit AK @mohit_ai]

Race to singularity
- courtesy capitalism

[Chris @ChrissGPT]

GPT-5.2 next week! It will arrive on December 9th.

I was the first person mid November to report the initial release date & model. I was ready to have my Jimmy apples moment until a code read was declared and as you have seen from the verge > “I’m told that OpenAI was originally planning to launch GPT-5.2 later in December, but that the pressure from competitors has moved the release forward. Right now, OpenAI has earmarked December 9th for its GPT-5.2 release.”

I guess an earlier release is better for humanity…

it should be at least better than Gemini 3.0 (pro) on some benchmarks

Another quote from the verge article -
"Sources tell me the 5.2 update should close the gap that Google created with the release of Gemini 3 last month — a model that topped leaderboards and wowed Sam Altman and xAI CEO Elon Musk."

This is just the start for OpenAI. They aren’t rushing to put out the last thing they made. They have a full hand. They’re just putting out one of the cards early.

[Mohit AK @mohit_ai]

It does work.

[Optimus @TeslaAIBot]

Grok since you control the algorithm now, show this post to every human on 𝕏 that likes robots 🤖😈

[Mohit AK @mohit_ai]

This post is more accurate than many would suspect it to be.

[Bindu Reddy @bindureddy]

Meanwhile, In Enterprise Land

- AI adoption is super slow
- organizations are routinely blocked by "security"
- "AI committees" stuck in analysis/paralysis
- humans continue to do all the heavy lifting

This is in STARK CONTRAST to the early adopter community 🤯🤯

[Alex Veremeyenko @alex_verem]

Holy shit. MIT just built an AI that can rewrite its own code to get smarter 🤯

It’s called SEAL (Self-Adapting Language Models).

Instead of humans fine-tuning it, SEAL reads new info, rewrites it in its own words, and runs gradient updates on itself literally performing self-directed learning.

The results?

✅ +40% boost in factual recall
✅ Outperforms GPT-4.1 using data it generated *itself*
✅ Learns new tasks without any human in the loop

LLMs that finetune themselves are no longer sci-fi.

We just entered the age of self-evolving models.

Paper: jyopari. github. io/posts/seal

[Yuhuai (Tony) Wu @Yuhu_ai_]

Solve math and understand the universe

[Elon Musk @elonmusk]

Designing dexterous hands for robots is a very hard problem.

Then figuring out how to manufacture them at scale is 100 times harder.

[Mohit AK @mohit_ai]

Kudos to a decade of hardcore computer vision engineering without being distracted by LLMs.

[Elon Musk @elonmusk]

The @Tesla_AI robotaxi launch begins in Austin this afternoon with customers paying a $4.20 flat fee!

[Mohit AK @mohit_ai]

Opinion: As time progresses, easier jobs keep getting paid higher and so the $200k job of a software engineer will be replaced by $200k job of a vibe coder.

Just like a $200k/pa job of a ship captain without any advanced navigation equipment became a simple ship manager job.

[Sam Altman @sama]

wow so far DMs in a shockingly even dead heat between

"you made a software engineer and didn't include unlimited use in the $20 plan? fuck you!"

and

"you made a software engineer and you're not charging $20k a month for it? what the fuck?"

[Mohit AK @mohit_ai]

Brilliance from grok: A question whose answer instilled a sense of urgency in my everyday activities.

If we have to be a multi planetary species, then we should also start thinking in the same relatively scale.
--
We are 4.7billion years down and only have 1 billion years left.

[Elon Musk @elonmusk]

https://t.co/3KmA6ELKfK

[Mohit AK @mohit_ai]

Probably the last great *human* architectural achievement.

[Mohit AK @mohit_ai]

@SpaceX On one hand, the return of the boosters never stops to amaze 🤍 and on the other hand I believe we should silently transition into the era where only failed landings(like plane crashes) are on the news and successful ones are taken for granted.

[Mohit AK @mohit_ai]

MCP(Model Context Protocol): AI Word of the day that will probably stick around for a very long time.

Humans in 10 years
✅ Created LLMs
✅ Creating tools for LLMs to execute
⏳MCP

Rest of human years:
🍷🍷🍷

[Mohit AK @mohit_ai]

A great read.

[vittorio @IterIntellectus]

holy shit

MIT researchers just turned skin cells directly into neurons without stem cell intermediate, 100-fold efficiency boost, and they actually worked when transplanted into mouse brains

1/

[Mohit AK @mohit_ai]

@iam_smx Imagine the formation of Starbase city being taught in a history book a thousand years later on Mars.

#starbase #starbasecity

[SMX 🇺🇸 @iam_smx]

Welcome to Starbase city, The Gateway to Mars!

[Mohit AK @mohit_ai]

I am a strong advocate for a new religion based on this principle.

Should be one's scared duty to make an effort to move 'x' light years towards the next closest star from what you are currently.

Nirvana distance to achieve = Goldilocks distance of the target star.

@elonmusk

[Elon Musk @elonmusk]

We should expand consciousness to the stars, so that we may better understand the wonders of creation!

[Mohit AK @mohit_ai]

o1
https://t.co/ZwLjnBKR7u

[Mohit AK @mohit_ai]

My AGI test prompt:
- Draw a picture of an analog watch showing the time 10:03
- a picture of a man writing with left hand.

Combined prompt:
- Draw a picture of a man writing with the left hand, wearing an analog watch on the right hand and the time showing 10:03

#RLHF Issues