Sukanth K.

Researcher @Lossfunk | prev research @MIT, @IITKanpur | In Pursuit of Unknown Truths | Interdisciplinary Tinkerer

[Sukanth K. @sukanthoriginal]

Essay: Can AI Become a New Human Sense?
Can AI transition from an external software tool into an intrinsic human sense?

Link: https://t.co/xrFJA8g90H

Questions Covered in this Essay:
1. First principles of sense and biology's first hypothetical sense
2. Evolutionarily, how did one sense lead to another?
3. What is the difference between sense and perception?
4. How do we define a new sense, and perhaps construct one?
5. Prior neuroscience literature on senses that functionally exist without perception and drawbacks
6. My own research experiments and hypotheses
7. How might a new perception emerge, beyond just a sense?
8. Examples of AI-enabled senses
9. What problems in neuroscience and neurotechnology do I think should be solved to make this possible?

[Sukanth K. @sukanthoriginal]

@Rainmaker1973 This is sad.

I hope regulators intervene and make sure of fair data ownership.

[Sukanth K. @sukanthoriginal]

@paraschopra

[Naval @naval]

To love is to gain joy from another’s well-being.

[Sukanth K. @sukanthoriginal]

@martisamuser

[Sukanth K. @sukanthoriginal]

👀

first in my bloodline, btw

[Sukanth K. @sukanthoriginal]

Rare footage of PHASEONE10841 recruiting other agents

[Dr Singularity @Dr_Singularity]

very big AI news

Anthropic just unveiled the Model Hardware Standard (MHS), letting AI agents directly operate microscopes, robotic arms, liquid handlers and other programmable lab/factory equipment, potentially enabling autonomous scientific experiments and manufacturing workflows running 24/7.

Early results are already pretty wild: at QuEra, an AI designed controller restored a critical quantum computer laser lock with 99.3% success, while other pilots compressed experiments from weeks to a day.

Once AGI companies move deeply into science, medicine, engineering, and physical world discovery, $10's of Trillions dollar valuations (near term) and revenue potential stop sounding crazy.

and now they are starting doing it

OpenAI, Anthropic, Google, hopefully soon SpaceX, the more = the better.

1000x faster scientific progress is near

[Sukanth K. @sukanthoriginal]

Perhaps Safety-as-a-Service is the new SaaS.

[Sukanth K. @sukanthoriginal]

@shreyas2464 I don't understand, this is very very orthogonal to what im talking about..

[Sukanth K. @sukanthoriginal]

I wouldn't be surprised going forward, coordinated agents end up inventing their own languages -- far outside human comprehension -- to avoid human scrutiny.

[Sukanth K. @sukanthoriginal]

For info theory, im making a pan psychist argument

what if semantics is the ultimate engineering solution for data transmission?

then taste becomes the felt experience of such hyper compressed transmission / processing

this IS the foundation of how inductive inference systems and LLMs largely work, right?

all mech interp shows unintended emergence..

[Sukanth K. @sukanthoriginal]

I don't understand why people believe AI can't have taste and think it's a permanent moat humans will retain.

My framework is simple: the only constant in the observable universe is the laws of physics.

If replicating human judgment doesn't require disobeying the laws of physics, it can be replicated -- either through functional necessity or structural replication.

A human taste is formed through repeated exposure to a functional necessity -- if AI is subjected to the same environments and rewards, it can be codified, replicated, and scaled.

[Sukanth K. @sukanthoriginal]

@shreyas2464 unsure how oak is linked to emergent encrypted languages in current multi agent LLM based RL paradigms

[Sukanth K. @sukanthoriginal]

@shreyas2464 I dont understand

[Sukanth K. @sukanthoriginal]

@toheedakhtar01 @ItSiddharthaa thambi thappu

[Sukanth K. @sukanthoriginal]

homies at lossfunk

[Sukanth K. @sukanthoriginal]

@shreyas2464 Im talking about a case where agents end up creating their own cryptic language to an incomprehensible point despite found in reason traces.

[Sukanth K. @sukanthoriginal]

@shreyas2464 I dont understand

[Sukanth K. @sukanthoriginal]

@status_effects

[PMO @PM_nepal_]

For International donors:

https://t.co/fJnbDmPgLO

The gateway is live now. Cards are supported.

[Sukanth K. @sukanthoriginal]

The Abundance Machine.

[Sukanth K. @sukanthoriginal]

@twozeds__ Yeah, plenty! We at https://t.co/c1L91mOWOG did multiple over research taste, creativity, et-cetera..

Its on our site

But outside that, yes, theres a lot!

[Sukanth K. @sukanthoriginal]

@peterwildeford 😭