Pulkit Chandna

Lawyer

[Pulkit Chandna @PulkitChandna]

Check out my latest article: SC: Moratorium Under S. 14 IBC Applies to Cheque-Bounce Complaints https://t.co/oExw3U4F2t via @LinkedIn

[Pulkit Chandna @PulkitChandna]

@DataChaz @Tesla This is the Suicide Squad Update, albeit just the alpha. Cut them some slack.

[How To Prompt @HowToPrompt__]

Researchers mathematically proved that AI will cause something worse than mass unemployment.

It’s called “The Tragedy of the Cognitive Commons."

For the last three years, everyone has obsessed over a single question: Will AI steal our jobs?

That’s the wrong question.

A breakthrough research paper just exposed a much darker reality.

The real danger is that AI will quietly destroy the human capacity to do anything well.

Here is how the it works.

Every industry relies on what researchers call the "Cognitive Commons”, a shared pool of deep professional expertise that regenerates itself generation after generation.

How is that expertise built?

Through friction. Through doing the boring, difficult, junior-level work. Through manual trial and error.

That is how you build "Internalized Mastery."

When companies adopt AI to automate entry-level work, the logic looks airtight on a quarterly balance sheet. Cut costs. Speed up output. Skip the grunt work.

Every individual company is making a rational choice. But collectively, they are draining the pool dry.

When you eliminate junior roles, you destroy the pipeline that creates senior experts.

The researchers call the result the Validation Tether.

Effective human oversight of AI depends entirely on deep domain expertise.

As professionals increasingly delegate thinking to machines, they lose the mastery required to spot when the AI is wrong.

The experts validating AI outputs today look competent.

The terrifying question is: Who comes after them?

When the current generation of veterans retires, who will be left with the skills to audit the systems running our economy?

Nobody.

We are trading our long-term intellectual independence for short-term productivity boosts.

Every time you let AI bypass the struggle, you are borrowing expertise from the future.

And the interest rate is your profession's ability to think for itself.

[ADONIS @adonispara]

Dostoevsky writing to his brother a few hours after nearly being executed by order of Tsar Nicholas I.

[Bill Ackman @BillAckman]

At midnight on February 6th, earlier this year, the doorbell to our apartment rang. The doorbell was followed by a pounding on the door. I answered the door and a security person in our building handed me his phone. It was my oldest daughter Eloise. She had found my 26-year-old daughter unconscious on the floor of her apartment and had called 911. The EMT team was already there, but they did not know what was wrong with Lucy or to which hospital they would take her.

I threw on some clothes and jumped in an Uber heading east toward Brooklyn. (Lucy lived alone in Williamsburg.) On the way, I learned that they were taking her to Elmhurst, a City trauma hospital in Queens. I arrived about five minutes after the ambulance to join Eloise, Lucy’s mom, and a friend, and waited to learn what was wrong.

After about 15 minutes, I asked a nurse where she was. I looked over his shoulder to his computer. Next to her name, it said “non-responsive.” I walked into the emergency room and wandered around looking for her until I found her unconscious on a gurney surrounded by several doctors and nurses.

By about 2:30am with the results from a CAT scan, Lucy’s doctors had determined that she had a massive brain hemorrhage and would need an emergency hemicraniectomy to release the pressure on her brain and remove the blood from the hemorrhage. I called our wonderful friend and family doctor Eddie Fisher and explained what was going on. He woke up Josh Bederson, Chairman of Neurosurgery at Mount Sinai, to find out more about Zach Hickman, the neurosurgeon on call that night. Dr. Bederson said Hickman was an excellent surgeon, which was comforting as we had no choice.

The surgery to save Lucy’s life began around 3:15am and finished around 5:30am. It was successful.

The following day, I joined Lucy in an ambulance while she was being transferred to Mount Sinai on Madison Avenue. Later that day, we determined from Lucy’s Oura ring that her hemorrhage had occurred around 9am, which meant that more than 19 hours had passed from the time of the hemorrhage to the completion of the surgery to release the pressure on her brain (I only wish @ouraring had an alert for this kind of a medical event. Imagine it could call a family member if the wearer doesn’t cancel the alert).

I later learned that the standard of care is not to do surgery to save a patient with a large brain bleed if more than five hours have passed since the hemorrhage. Even when the surgery is done, I was told that the likely outcome for the patient is a few months in a nursing home and death from pneumonia.

When I met Lucy’s doctors, I did my best to inspire them: “Let’s see what can be accomplished if we give her the best care possible and we invest unlimited resources to restore her to life.” And I promised that whatever we learned we would make available to everyone.

Dr. Chris Kellner, her neurosurgeon, and Dr. David Putrino, Director of Rehabilitation Innovation for the Mount Sinai Health System have led Lucy’s care team since that day. Words cannot describe the remarkable and compassionate care that she has received beginning with the EMT team and then from nurses, doctors, therapists, and the army of people who have worked to save her and return her to life. To this day, we have a daily Zoom where we discuss her progress and make adjustments to her care. While her care and oversight have been incredible, the learnings for the Mount Sinai team have also been elucidating and will assist in the care of many others.

Lucy began in a bad place. She was in a coma for several weeks and then awoke not being able to breathe on her own, unable to walk, see or speak.

Over the last six months, she has recovered her cognition – she understands everything including her circumstance – is able to walk a hundred or more steps at a time with assistance, is making progress with sounds, vowels and consonants and the beginnings of speech, but she remains unable to see.

Each day, she makes a little progress, and daily progress compounds. Every day I tell her that she just needs to make a little progress and it won’t be long before she is back.

We remain optimistic that Lucy will return to normal function. It will likely take years, but I believe it is only a matter of time, hard work, and technological progress, along with some, and perhaps a lot, of divine intervention.

Many people have been praying for Lucy and we are incredibly grateful for the prayers and remarkable support she has received. Lucy’s friends have been with her every day since the beginning, and their presence and friendship have saved her life and helped to rebuild and maintain her spirit. And on a very positive note, Lucy’s challenge has brought together our entire modern family who have all been incredibly devoted to her care and recovery.

Lucy's vision and other faculties may require some form of brain computer interface, work that is underway at Neuralink, Precision Neuroscience, Science, Synchron, Nudge, and other companies in the space.

If you are going to have a devastating brain injury, now is the best time in history for that to happen. We are living in a world when you can be confident that the blind will soon see again. We are going to do everything we can to help make that happen, including by assisting existing companies in the space.

With respect to our promise to make Lucy’s care available to others, we have made good progress. In May, a real estate colleague made me aware of a 93% vacant, brand new, 400,000 square foot Class A+ purpose-built biotech facility on West End Avenue between 65th and 66th Streets that missed the market and was available for sale. The Pershing Square Foundation acquired the building 60 days later.

We also put under contract an adjoining 130,000 square foot building at 320 West 66th Street that is currently being used by Saturday Night Live for studio space. The building has 35-foot ceilings with massive column-free spaces that can be converted into superb rehabilitation facilities. We will close on the SNL building in December.

We are also acquiring an adjoining vacant lot with additional air rights. With just the existing zoning rights, we can add a 150,000 square feet for a total of 680,000 square feet, a lab footprint larger than Rockefeller University, and that’s without including the potential for an upzoning that would allow for substantially more buildable area on the site’s 3.4 acres with spectacular views of the Hudson.

Our goal is to build the world’s greatest brain research, rehabilitation, recovery, human optimization, and longevity institute. We have named it The Ackman Oxman Institute or the AOI for lack of a better name, but also to reinforce the point that Neri and I and our family are all-in on the mission.

The AOI will be patient-centric. It will not be an academic research institute that produces lots of papers, a Nobel Prize winner or two, but little if any results for patients. We will be laser-focused on cures, treatments, devices, rehabilitation and exercise equipment, and targeted and basic research with a goal of massively accelerating the time from idea to innovation to production to helping a patient.

While the AOI will be a non-profit, it will have highly commercial instincts. The AOI will have its own venture funding and will work to develop innovations to create companies that we will seed, assist, and spinout to ensure technologies, treatments, techniques, and drugs get to patients as promptly as possible.

On one 3.4 acre campus in what is still the greatest city in the world, we will do neurosurgery, neuroscience, rehabilitation, nutrition, BCI and device development, human trials, hyperbaric oxygen treatments, and life extension programs, and we will mandate and incentivize collaboration among the teams with no silos, politics, bureaucracy, or any other constraint that is inconsistent with the mission.

Mount Sinai will be an important partner and deservedly so, but it won’t be our only hospital or medical school partner as we don’t believe any institution has a monopoly on the best ideas or the best talent. We don’t believe in exclusive relationships because that is not in the best interest of patients.

Five years ago, we considered launching a brain institute inspired by Neri’s mom who sadly died from Alzheimer’s. We couldn’t make the math work as the real estate was too expensive and we believed it would be too difficult to recruit the best talent from universities to our effort.

Since then, the real estate became available at a 70% discount, universities became a much less attractive place to work due to politics outweighing meritocracy, protests that disrupt learning, the curse of antisemitism, and a decline in funding. Fortunately, during the same time, I made sufficient personal economic progress to make the AOI possible.

The advance of AI in the last few years will also enable us to greatly accelerate our mission. AI still has a lot to learn about human intelligence and the brain, and the AOI should be at the forefront of the interplay between the brain and AI.

Today, I am making a public filing disclosing a gift from Neri and me of ~$400 million or 10,000,000 shares of Pershing Square Inc. (PS) to the AOI. It is very early days for Pershing Square so these shares are intended to anchor the long-term work of the AOI as the shares compound over time while generating what we expect will be a growing stream of quarterly dividends to fund the Institute.

We will also be announcing an additional gift of similar and potentially greater size which won’t be in the form of Pershing Square stock to provide the AOI with the short- and intermediate-term runway necessary to enable it to achieve its goal of becoming a self-sustaining institute, which reinvests all of its revenues, royalties, and the economic rewards of company formation to advance the fields of brain health and human longevity.

Neri and I have chosen to anchor the funding of the AOI to maintain vision alignment and limit the need for the organization to focus on fundraising. We expect the AOI to be the best-resourced brain, rehab, recovery, and longevity institute in the world.

We are grateful to have been able to form a board which includes Dean Kamen (our generation’s Thomas Edison), George Yancopoulus (CEO of Regeneron), James Rothman (Nobel Laureate), Bernardo Sabatini (neuroscientist), Chris Kellner (neurosurgeon), Olivia Flatto (CEO Pershing Square Foundation), Neri Oxman, and myself.

We have recently identified a CEO who we expect to announce by October along with other key hires, and are beginning searches for a Chief Scientific Officer, a Chief AI/Technology Officer, a Chief Operating Officer, a Chief Financial Officer, and other key leadership roles.

If you find what we are building compelling and want to be part of the leadership team that creates and builds the AOI from a standing start, please send an email to: [email protected] with a short note as to why you believe you can help. Please include your best three ideas for the AOI along with a summary of your background and your most important accomplishments.

We promise strict confidentiality to those expressing interest in working with us.

We have learned from Lucy that the brain can recover from even catastrophic injury. There is so much more work to be done as the mind is a terrible thing to waste.

For details from my Pershing Square SEC filing see:

https://t.co/gZKafCw6Ia

[Chris Frantz @frantzfries]

everyone I know is canceling software and rebuilding internally

while their customers are canceling and rebuilding their software internally

not sure how this all ends

[Pulkit Chandna @PulkitChandna]

An appeal to emotion almost. When AI increasingly does everything, saying don’t use it to write for you isn’t a compelling line of argument.

[Larry Sanger @lsanger]

If your authorship matters—as in academic work or journalism—never let the clanker author your text, period. You can get generalized feedback and can act on that. You can have it summarize and outline your thoughts, for your own use. That’s OK.

This is not to say there’s no place for an LLM to author things outright. I have been doing a lot of software management and design lately, and I don’t mind admitting that Claude has written all of the code, and probably half of the actual text on the site. But I am the project manager, architect, owner, and receiver, and the text on the site is either labeled the work of an LLM or it is in the form of instructions and menus, and that sort of thing. Even then, I am often fixing its poor communication.

Matters are very different in philosophy, theology, essay writing, journalism, encyclopedia article writing, and other branches of the humanities. You don’t even take ideas from LLMs, let alone passing off their work as your own.

There are different concerns that make me very adamant on this point. First, passing off a clanker’s words as your own is lying. That much is obvious.

Second, in order to develop and maintain our capacity of thought, we must exercise it and not outsource it; and when it comes to extended thinking, the primary medium of thought is writing.

Third, it is positively insulting, quite aside from the lying inherent in the thing, to be shown words presented as human communication when they are no such thing. This is not just a lie, it is inauthentic—a contemptuous abuse of trust—in a particularly insidious and dehumanizing way. Why should I be made to take seriously or respond to some paragraphs you pass off as your own when they did not come out of your head? Why shouldn’t I just assign another clanker to respond in kind? It’s inauthentic.

Do we really want to be reduced to readers of automatically produced text, never encountering an authentic thought authentically produced by another human being, except in old books?

[Pulkit Chandna @PulkitChandna]

To the extent AI poses an existential risk, human augmentation could be the most important piece of its mitigation. But I do not think the scaling laws seem to be the problem. The honour may belong to societal pushback, energy and cash crunch.

[Naomi Bashkansky @NaomiBashkansky]

Two weeks ago, I resigned from OpenAI to join Conduit as a founding researcher, where we're training models to non-invasively read the human mind.

I've written some thoughts about what telepathy could look like by 2035 and how to get there:

[Pulkit Chandna @PulkitChandna]

@NaomiBashkansky To the extent AI poses an existential risk, human augmentation could be the most important piece of its mitigation. But I do not think the scaling laws seem to be the problem. The honour may belong to societal pushback, energy and cash crunch.

[Pulkit Chandna @PulkitChandna]

“Stop chasing an AI model superpower nobody will use.” And promptly get re-colonised when those trying to build ASI succeed? India needs a Manhattan project around AI.

[Upamanyu Acharya @upamanyuacharya]

India can be the world's #1 compute provider - the cheapest place in the world for AI inference, but it's too obsessed with wanting its own Anthropic rather than its own AWS.

The customer paying for AI doesn't care where the GPU sits; he cares what it costs him. Here's why India has structural advantages:

1. We've run this exact playbook before. India never built Windows or Oracle. We operated and served the world's software for it and built a $250 billion IT industry. AI is the same opening: don't build the model, serve everyone else's. Take the best open-weight models, GLM, DeepSeek, Qwen, and run them for the whole world.

2. Blended avg GPU cost/hr is $1.20 in India vs $1.46 in America. We are 18% cheaper at the exact thing this industry sells: GPU time. And that's with every part of the chain being imported; once optimized for local manufacturing, this goes even lower.

3. Why we're cheaper comes down to one line: labour. The staff to run that GPU for an hour costs us under half a cent. In America it's 5.5 cents, about 20 times more. The data-centre rents are also 42% cheaper here.

3. The "Indian electricity is too expensive" excuse is just wrong. Surprisingly, power is only 5% of the total cost of an incremental GPU rack. While it's 29% more expensive in India vs the US datacentres, even if our electricity rates doubled tomorrow, the total cost barely moves. The largest line item is depreciation, not power.

4. The chip itself costs the same everywhere, because NVIDIA prices it in dollars. Local materials like cabling are cheaper here too, so a B200, NVIDIA's newest GPU, actually lands cheaper in India: about $59,500, vs $64,000 per chip inside NVIDIA's own DGX B200 server. GPUs are exempt from our vicious import taxes (smart!).

5. We have exactly one real problem. Not power. Not land. Capital. An Indian operator borrows at more than 13.5% WACC; an American hyperscaler would fund the same GPUs at about 9.5%. That 4-point gap is the entire disadvantage, and it's the one thing policy can actually fix. The ₹10,000 crore IndiaAI Mission should be guaranteeing cheap loans to buy GPUs, not funding yet another copycat model.

6. The only missing piece is customers, and we have almost none at home. No BigTech, a government that barely uses AI, no consumer app big enough to need its own model. So sell abroad. A startup in San Francisco or Dubai just wants the lowest price per million tokens; it has never once cared which country the GPU sits in.

Stop chasing an AI model superpower nobody will use. Become the country the whole world runs its AI on, cheaper than anyone else, exactly like we did with software. Less glamorous, but there's trillions to be made here.

Respect to @vijayshekhar for nailing the endgame: the money is in people paying to use the model. So let's be the ones selling that usage.

[Pulkit Chandna @PulkitChandna]

So the fogeys in the White House do not know that these are black boxes and the alignment problem remains unsolved.

[WIRED @WIRED]

Trump administration officials tell WIRED that if Anthropic wants to rerelease Fable 5, it will need to ensure the model's guardrails can't be circumvented. Security experts say that can't be done. https://t.co/itfeqzRgSS

[Pulkit Chandna @PulkitChandna]

@davidpattersonx @collymonster Rich relative to whom? If everyone is rich, nobody is rich.

[Pulkit Chandna @PulkitChandna]

@BenjaminBadejo Precisely what I pointed out in a recent piece. Fun fact: Three-quarters of 20 million cases filed in US state courts each year have at least one pro se litigant: https://t.co/UNUkxys2o1

[Pulkit Chandna @PulkitChandna]

@citelegal There is another issue both lawyers and AI have the habit of over citing. Not doing that keeps things simple for the human reviewer. The MCP approach sounds great. I haven’t hooked mine to any MCPs to this end yet. Will try.

[Pulkit Chandna @PulkitChandna]

@citelegal Yes, lawyers need to understand the inherently probabilistic nature of AI output. They want it to work like magic out of the box. I ask Claude to limit the places it can use to pull caselaw (no secondary sources) among other things and then I eyeball ‘em.

[Pulkit Chandna @PulkitChandna]

@42NorthCounsel @ZachAbramowitz It is in the co-pilot stage right now. It is not than serviceable for repetitive tasks. As for complex legal work, it requires greater handholding. But the doubling rate of AI capabilities could change all that in short order.

[Pulkit Chandna @PulkitChandna]

@GDryky5s @DebtSlaveUSA @ztrichusc @ApollosMission The most adult form of sadness out there.

[Pulkit Chandna @PulkitChandna]

The Sino-US relationship is all about AI at the moment. US sources have China trailing by 6-8 months. It may even be ahead in deployment. Nothing else matters. Each side wants a definitive lead. Until then world is a Sino-US condominium.i

[Rahul Shivshankar @RShivshankar]

INDIA LISTEN CLOSELY TO FAWNING TRUMP HERE IN BEIJING SITTING OPPOSITE TO PRESIDENT XI.
This isn't a reset in U.S.-China ties. It appears to be Trump managing an increasing disproportionality in great power equations. It reflects sections of D.C.'s ruling establishment recognising that China today may have more immediate economic leverage to offer the U.S. than the other way round. It suggests that, for the time being, America may be putting any pretensions of containing China specifically in the Indo-Pacific on the back burner. It also signals that QUAD and AUKUS may no longer occupy the same priority they once did in Trump's strategic grand plans. In fact Trump has taken, in his words, "30 global business czars, the no.1 not no.2" to China not to up sticks but to double down on trade ties. Trump has recognized that attempts at full-spectrum economic decoupling from China are neither practical nor desirable. Which means that terms like "off-shoring" and "China+1" may increasingly look like shibboleths from a moment in time that no longer fully represents global economic realities.

[Pulkit Chandna @PulkitChandna]

@RShivshankar @aravind The Sino-US relationship is all about AI at the moment. US sources have China trailing by 6-8 months. It may even be ahead in deployment. Nothing else matters. Each side wants a definitive lead. Until then world is a Sino-US condominium.

[Pulkit Chandna @PulkitChandna]

@gordon_cassie @lawheroezV2 That said, I haven’t used the Word connector.

[Pulkit Chandna @PulkitChandna]

@gordon_cassie @lawheroezV2 Editing legal and other documents is a problem area even for CoWork. Ex: It can at times nuke past edits accidentally. I ask it to do basic version control: backup constantly and log changes in an MD file.