AutoRickshaw Endless Runner mobile game developed in Unity using 100% AI (mostly Codex, some Claude Code, a little bit of Formy3D). More than 100k lines of code and not one character was written or even read by me. Check it out!
Fast paced, action packed AutoRickshaw driving game now live on Apple App Store and Android Play Store!
iPhone App Store: https://t.co/lzuK76xtIF
Android Play Store: https://t.co/GqFlGNAJoJ
@Mechramc Hmm. It's probably because the whole set of possible characters is only in the few hundreds, so learning the composition rule doesn't have a big payoff
AI is like a Natural Human Language interface/layer on other technologies. It’s powerful because Language is more powerful than many people believed. LLMs are good at intelligence because Language is fundamental to reality at a spiritual/ontological level. ‘In the Beginning there was the Word’. ‘Word’=Symbol. There is a Duality between Word and Form. This has always been recognized in theology of all major religions.
People like Yann LeCun don’t understand the spiritual significance of Language.
I typically use ChatGPT for brainstorming and Codex for execution, often asking ChatGPT for good prompts to give Codex. But the context sharing is a chore since the chat app does not have access to my project folder.
I asked ChatGPT if I should just use Codex for everything and it told me ‘No, keep me in the loop’.
But now once I have the basic outline of an idea settled through brainstorming in the chat app, working purely with Codex is much more efficient and sufficient. Codex is just as good at ideation and much less wordy and verbose. It cuts to the chase.
I agree with point 1.
LLMs work so well at intelligence because Language is fundamental to reality at a spiritual/ontological level. This has always been understood in theology of all major religions. ‘In the beginning there was the Word’. ‘Word’=Symbol (includes math and code). There is a Duality between Word and Form.
The human brain is not a computer or algorithmic machine. We are spiritual beings. Our thoughts and words are sometimes not even our own — they can travel from one person to another, they can change our reality.
There was a case of a human who was functioning with most of his brain missing (don’t know how authentic the story is).
Amidst all the SM noise, we did not celebrate Vimag labs enough. They built India’s 1st software-defined, magnet-free electric motor platform.
Standard EVs rely heavily on Permanent Magnet Synchronous Motors. These require physical, rare-earth magnets embedded directly into the motor's rotor to create a magnetic field. Vimag Labs completely eliminated the physical magnets.
This is a win for cheaper EV manufacturing, but the real, strategic importance of what they did goes far deeper into global geopolitics & structural engineering:
- The global processing capacity & supply chain for rare-earth materials is overwhelmingly controlled by China. By engineering a completely magnet-free motor, Vimag Labs quietly handed automotive OEMs an escape hatch from a massive geopolitical supply chain vulnerability.
- Vimag Labs designed this architecture to scale up into massive high-power systems ranging b/w 200 kW & 600 kW. This means the software-defined, magnet-free platform is directly targeted for critical, heavy backend sectors: defense applications, robotics & advanced cooling infra, allowing India to build high-performance military & industrial hardware entirely free from foreign mineral dependencies.
- The breakthrough is the result of 87600+ (~10 person yrs) engineering hrs spent by co-founders Manish Seth, Rahul Krishnamurthy & their team. They built a massive IP pipeline (including 5 granted patents, 10 active applications & 15 trademarks) & signed a manufacturing MoU with Jendamark to scale the physical production of these motors right out of India.
Vimag labs handed a rising nation the ultimate industrial escape hatch: a future where our engines run on Indian brainpower, while leaving the rest of the world fighting over the dirt. 🙏🙏
@psankar@Mechramc I will make my GitHub repositories public when I am done. I want to write an expository article and publish it on my website and twitter with all useful artifacts publicly available. I then plan to do several more technical experiments and write similar articles as a series.
Amidst all the SM noise, we did not celebrate Vimag labs enough. They built India’s 1st software-defined, magnet-free electric motor platform.
Standard EVs rely heavily on Permanent Magnet Synchronous Motors. These require physical, rare-earth magnets embedded directly into the motor's rotor to create a magnetic field. Vimag Labs completely eliminated the physical magnets.
This is a win for cheaper EV manufacturing, but the real, strategic importance of what they did goes far deeper into global geopolitics & structural engineering:
- The global processing capacity & supply chain for rare-earth materials is overwhelmingly controlled by China. By engineering a completely magnet-free motor, Vimag Labs quietly handed automotive OEMs an escape hatch from a massive geopolitical supply chain vulnerability.
- Vimag Labs designed this architecture to scale up into massive high-power systems ranging b/w 200 kW & 600 kW. This means the software-defined, magnet-free platform is directly targeted for critical, heavy backend sectors: defense applications, robotics & advanced cooling infra, allowing India to build high-performance military & industrial hardware entirely free from foreign mineral dependencies.
- The breakthrough is the result of 87600+ (~10 person yrs) engineering hrs spent by co-founders Manish Seth, Rahul Krishnamurthy & their team. They built a massive IP pipeline (including 5 granted patents, 10 active applications & 15 trademarks) & signed a manufacturing MoU with Jendamark to scale the physical production of these motors right out of India.
Vimag labs handed a rising nation the ultimate industrial escape hatch: a future where our engines run on Indian brainpower, while leaving the rest of the world fighting over the dirt. 🙏🙏
I’m not taking sides on this issue but I have always felt that using language to infer movement of people is very tenuous. Language travels much faster than people. And it has always been that way (I don’t believe telecommunication is a new thing. Also humans have the ability to communicate without external technology — visions, voices, etc).
💯. My plan is to now train a small language model for translation from Tamil to English and use that as a benchmark for comparing different tokenizers. If you have your own evaluation code, would love to have it tested. I will try to share my tokenizer with you within the next 1-3 days.
Whether the earth is flat or spherical depends on the observer’s perspective. The fact that the Sun and the Moon are the same size when viewed from the earth is not an accident. The Sun and Moon represent Duality — Sun is Masculine, Moon is Feminine.
If you understand Duality, everything becomes clear.
Surah Ya-Seen [36:36]: "Glory be to Him Who created pairs of all the things that the earth produces, as well as their own kind (human) and [other] things of which they have no knowledge."
Duality is defined by two properties — Conditionality (Interdependence) and Orthogonality (Oppositeness).
Male and Female are interdependent but also opposites. Same applies to Birth and Death, Good and Evil, Being and Non-Being, Sameness and Otherness, etc.
Duality is built into the Fundamental Law of Cause and Effect in Nature. Every Cause has two immediate Effects which are Dual to each other. When a light switch is flipped on (Cause), there are two immediate effects — Birth of Light and Death of Darkness. These two effects are dual to each other. Birth is opposite of Death. But they arise together.
The representation of this relation of a Cause to its two immediate Effects, which has Duality built in, is also the Fundamental relation of Word (Symbol) to Form. All other relations of Word to Form (like Counting and Naming) are secondary to the Law of Cause and Effect which expresses Duality.
Whether the earth is flat or spherical depends on the observer’s perspective. The fact that the Sun and the Moon are the same size when viewed from the earth is not an accident. The Sun and Moon represent Duality — Sun is Masculine, Moon is Feminine.
If you understand Duality, everything becomes clear.
Surah Ya-Seen [36:36]: "Glory be to Him Who created pairs of all the things that the earth produces, as well as their own kind (human) and [other] things of which they have no knowledge."
Duality is defined by two properties — Conditionality (Interdependence) and Orthogonality (Oppositeness).
Male and Female are interdependent but also opposites. Same applies to Birth and Death, Good and Evil, Being and Non-Being, Sameness and Otherness, etc.
Duality is built into the Fundamental Law of Cause and Effect in Nature. Every Cause has two immediate Effects which are Dual to each other. When a light switch is flipped on (Cause), there are two immediate effects — Birth of Light and Death of Darkness. These two effects are dual to each other. Birth is opposite of Death. But they arise together.
The representation of this relation of a Cause to its two immediate Effects, which has Duality built in, is also the Fundamental relation of Word (Symbol) to Form. All other relations of Word to Form (like Counting and Naming) are secondary to the Law of Cause and Effect which expresses Duality.
What exactly is gravity? I think nobody knows. We can create multiple theories to explain gravity and all of them can exist in parallel. For example, we already have Newton's theory and Einsteins theory.
And, I can predict, we will have a newer theory of gravity soon.
The point is the so called reality, including gravity, aren't real. We are collectively experiencing the illusion of gravity and creating theories of gravity to collectively confirm and believe in them at different scales. In reality, there's neither mass nor gravity.
I’m building a semantic tokenizer for Tamil and I am blown away by the quality and depth of understanding of Tamil grammar subtleties by GPT-5.6 and even 5.5. It is doing a much better job at identifying correct morphology of words than I ever could.
With models like GPT 5.6 and Fable, the real risk is miscommunication and non-alignment between my goals and the model’s understanding of my goals.
I read what the LLM is doing, ask probing questions, clarifications, suggest changes/alternatives, give feedback on bugs/errors, and ask for automated tests and extensive documentation. I don’t read the code anymore.