Pritesh Desai

Technology | Cinema | Consumer Behaviour | Mountains | Cyclist | Michael Clayton Evangelist | ex- @timesbridge | @MICA_Ahmedabad Alum

[Pritesh Desai @interpriteshan]

@theskindoctor13 Greatest paradox of our lives. Cinema imitates real life or other way around.

[Pritesh Desai @interpriteshan]

@ABiggerSpalash i mean, any theory why it took so much time for a fairly regionally popular dish to get such massy attention in its own country? partly it was lowkey in cyclical trend w/ indiranagar rameswaram which was twitter vanity darling till '22 (b4 obscure sourdough signalling took over)?

[Pritesh Desai @interpriteshan]

@ABiggerSpalash It blows my mind after spending almost two decades in bbay why benne dosa wasn't a thing while it's heavily commoditised in blr & otherwise in KA. I mean if vadapav could cross all regional boundaries, why was i only encountering to 8/10 times TN or KL prep x north to south bbay!

[Palash @ABiggerSpalash]

Agriculture is the only recent human invention that has had a genetic-level impact on humanity. We can now digest animal milk, starch and fend off animal pathogens. Human body has materially changed since agriculture, and it's only taken 12,000 years.

Now imagine what hyper-reduced stress on a child's brain muscles will do to human beings in the next 12,000 years. I think AI might be the next human invention that might change us at a genetic level.

We are definitely getting 'brain gyms' by the end of this decade, where children and adults go spend time solving riddles and math problems with a pen and a paper.

Physical gyms came up only after the Industrial Revolution. A man in the 1600s would have laughed at the notion of ordinary human beings going to a place to purposely do physical labour.

[Pritesh Desai @interpriteshan]

@ABiggerSpalash We are a species with amnesia - Graham Hancock

[Pritesh Desai @interpriteshan]

@ABiggerSpalash If we ground this thing in history, this was/has been the most employed parcel of men versus the other lots who were employed at best a quarter per year for their peak years. Only downside I see right now is men are isolated in their eco chembers and cosmopolitan outfits.

[pc @pcshipp]

Claude:
You've used 90% of session limit

Me instantly:

[ℏεsam @Hesamation]

“Expecting an algorithmic description to instantiate the quality it maps is like expecting the mathematical formula of gravity to physically exert weight.”

[Pritesh Desai @interpriteshan]

True! without clean kitchen, a house is a storage room for living and otherwise. without cleaner roads, cities of the nation are just concentration camps you are destined to succumb to.

[Mr Melancholy @chakravartiiin]

One look at average Indian street and you'll understand it has no business in being the top ten list of anything. GDP is a fake manipulated number to fool people into living like cockroaches while politicians settle their kids abroad. A service import driven economy is ngmi.

[Pritesh Desai @interpriteshan]

@ABiggerSpalash revolutions are always invoked by top 2% against top 1% by weaponising bottom 98%

[Pritesh Desai @interpriteshan]

Let me put it this way so I can come back to how I thought about it and if i thought about it right:

In short run, all such tools are great proxy ghosts for intern/jr level human talents and in the long run just extra tools in tool-sprawlness adding marginal upside to workflows

[Computer @AskPerplexity]

Perplexity Computer replaced $225K/yr in marketing tools in a single weekend.

We built an AI marketing agent that scans hourly, manages budgets, detects fatigue, and coordinates several campaigns end to end.

In one test run, it made 224 micro-optimizations to our ad stack.

[Pritesh Desai @interpriteshan]

@ABiggerSpalash My favourite scene Oscar playing Shoals of herring to his father. He starts it with saying " You used to like this one.." and after such a poignant play when he looks at his father, he looks away. Man, if only rejection had a visual expression

[Pritesh Desai @interpriteshan]

@Airtel_Presence Airtel WiFi in Tapovan, Rishikesh has been down for 3 days now. No technician has visited despite multiple requests. My 70 yr old mom is unable to do her work. Only our connection is affected. No local airtel support team has been responding for 3 days. Pls help.

[Pritesh Desai @interpriteshan]

@ABiggerSpalash I am going to take the other side. Job loss hysteria will be the biggest hoax in the history of humanity. Yes fat jobs are going to get trimmed but that's equalization rather than elimination. AI will be known as most efficient tool to complicate the fuck out simple workflows.

[Michael J. Miraflor @michaelmiraflor]

Either everyone on Twitter is right about the future of AI, or we’re observing a fantastic mass psychosis event.

[Naval @naval]

There is unlimited demand for intelligence.

[sphinx @protosphinx]

AGI is not coming.

We are nowhere near AGI. What we have today is inference, not learning.

Models get trained once on huge fixed datasets, then frozen. You ask questions, they remix patterns they already saw. Nothing updates. Nothing sticks. Talking to the model does not make it smarter. It does not learn from you. Ever.

Learning is still slow, expensive - and offline.

Look at self driving. You drive around a pothole, make a U turn, and come back. The car’s AI does not learn that you just solved that exact problem. It reacts the same way every time using sensors and rules. Do this 20 times a day and it still has zero memory that the pothole exists. It just re sees it. That is why edge cases never die. There is no local learning. No accumulation.

No 'oh yeah, I’ve seen this before'

LLMs work the same way. Tell it your name and it does not remember. The only reason it looks like memory is because scaffolding keeps shoving your name back into the prompt every time and sanitizing the output.

The model itself has no idea who you are and cannot learn from interaction. It is structurally incapable.

And the scaffolding is the worst part. It is pure duct tape. Just prompts on prompts on prompts around a frozen model. When something breaks, nobody fixes learning. They add another layer. Another rule. Another retry. Another evaluator model judging the first model.

So you end up with systems that are insanely complex but mentally shallow. Debugging is hell because behavior comes from hack interactions, not a learnable core. Tiny prompt tweaks cause wild behavior shifts. Latency goes up. Costs go up. Reliability goes down. None of this compounds into intelligence. It just hides the cracks.

Until we have real persistent learning and real memory inside the system, there is no AGI.

LLMs are not built for this. You cannot prompt your way out of it. You need a totally different architecture. Yann LeCun is right.

And even then, what architecture can actually learn online, store memory, and stay stable on today’s hardware?

Best case, maybe 5-10 yrs.

Right now it is all inference. It looks magical, but the emperor has no clothes. A lot of people see it. Almost nobody says it out loud.

[Pritesh Desai @interpriteshan]

I have never seen any manifestation of a poignant lyrics (here one in focus is sound of silence by Simon & Garfunkel) ever. Probably poetically best sermons on modernity penned ever.

[Pritesh Desai @interpriteshan]

@ABiggerSpalash Whatever gets imprinted in our teens, we live with it. We dont sense this but earlier social media internet ate millennial teen's psyche with, excessively seeded feeling of, default distrust. Hence simple bets scare us. We are gen which is in forever PTSD of stuff never happened.