Shashank

Making AI avatars talk, listen & clone voices. All local, all on cheap hardware. Documenting what I learn. 🇮🇳

[Shashank @randomshashank]

My AI avatar just joined a Google Meet call.

It listened. It replied. It made eye contact. It even noticed when the other person looked confused and simplified its answer.

It's running on a MacBook Air M1. 8GB RAM.

No cloud GPU. Everything local. Thread 🧵

[Shashank @randomshashank]

@reach_vb Needed this today. Not a deep-technical guy, just shipping local AI avatars on a cheap M1 and documenting the mess as I learn. "Release more than you refactor" is going on a sticky note.

[Shashank @randomshashank]

@ollama @NVIDIAAI @Zai_org Ollama is literally how a non-expert like me got local models running in the first place, built a voice-cloning avatar on an 8GB M1 with it.

[Shashank @randomshashank]

@pcuenq Wild to see frontier models running on Macs now. I'm at the opposite end. Squeezing AI avatars onto an 8GB M1 Air.

[Shashank @randomshashank]

@AnkurPhani Long live the HDD

[Shashank @randomshashank]

Is it as good as what Google or Meta are shipping with unlimited H200 clusters?

No. Honestly, no.

But those can't run on the laptop half of us Indians bought because it was the best value Mac available. I think the more interesting question is: what CAN you ship with just 8GB?

[Shashank @randomshashank]

My AI avatar just joined a Google Meet call.

It listened. It replied. It made eye contact.
It even noticed when the other person looked confused and simplified its answer.

It's running on a MacBook Air M1. 8GB RAM. No cloud GPU. Everything local.
Thread 🧵

[Shashank @randomshashank]

It also clones my voice. Locally.

MeloTTS + OpenVoice V2 running on the M1's Metal GPU.

~0.78x realtime.

Meaning it generates audio faster than you can listen to it.
On 8 gigs of unified memory.

[Shashank @randomshashank]

Is it as good as what Google or Meta are shipping with unlimited H200 clusters?

No. Honestly, no.

But those can't run on the laptop half of us Indians bought because it was the best value Mac available.

I think the more interesting question is: what CAN you ship with just 8GB?

[Shashank @randomshashank]

My AI avatar just joined a Google Meet call.

It listened. It replied. It made eye contact. It even noticed when the other person looked confused and simplified its answer.

It's running on a MacBook Air M1. 8GB RAM.

No cloud GPU. Everything local. Thread 🧵

[Shashank @randomshashank]

It also clones my voice. Locally.

MeloTTS + OpenVoice V2 running on the M1's Metal GPU.

~0.78x realtime. Meaning it generates audio faster than you can listen to it.

On 8 gigs of unified memory.

[How To Prompt @HowToPrompt__]

Yann Lecun published the most heretical AI paper of the year.

He opens by arguing Magnus Carlsen isn't good at chess and only gets more unhinged from there.

The Turing Award winner and his co-authors dropped a paper demanding the AI industry abandon its biggest obsession, AGI.

Right now, everyone from Silicon Valley CEOs to politicians assumes AGI is the ultimate goal. A machine that can do everything a human can do.

LeCun argues that this entire concept is a biological illusion.

Humans do not possess "general" intelligence. We are highly specialized biological machines, tuned by evolution simply to survive in the physical world.

We only think our intelligence is "general" because we are completely blind to the millions of cognitive tasks we are incapable of comprehending.

Which brings us to the chess argument.

Magnus Carlsen is the greatest human chess player in history. But compared to a modern computer? He is fundamentally terrible.

Our belief that Carlsen is "good" at chess is pure human-centric bias. He isn't objectively good. He's just better than the rest of us, who are biologically awful at it.

LeCun says we need to stop building AI to mimic human generality.

Instead, he proposes a new North Star: SAI.

Superhuman Adaptable Intelligence.

Instead of trying to build a machine that mimics our flawed, biologically-limited brains, we need to embrace extreme specialization.

SAI is about the speed of adaptation.

It is an intelligence that can learn to exceed humans at any specific, economically important task.

More importantly, it is designed to fill the vast skill gaps where humans are fundamentally incapable.

Things like managing global energy grids in real-time. Or predicting complex molecular structures.

The entire AI industry is obsessed with building a digital reflection in our own image.

LeCun's paper is a brutal wake-up call.

[Shashank @randomshashank]

Here's another one.

[Shashank @randomshashank]

Most AI videos optimize for quality.

I'm optimizing for cost.

The goal: generate watchable long-form videos for a fraction of what they're supposed to cost.

Here's a recent experiment 👇

[Shashank @randomshashank]

Mythos released today:

Next day:
Open the pod bay doors Claude... I'm sorry Dave, I'm afraid I can't do that.