I challenged CLAUDE :
Parekh’s Declaration of AI
Context :
Extract :
In this future
scenario, IQ scores will no longer be compared. Rather, people will contrast
intelligence packages or monthly AI expenses. Over time, the digital divide in
society would evolve into an intelligence divide. Better-off people have
superior internet connections today. They will also be sharper in the second
quarter of this century.
So, the situation
described above seems like it could happen sooner rather than later. Speaking
at the BlackRock US Infrastructure Summit in Washington, OpenAI boss Sam Altman
predicted that in the future, individuals will simply utilise AI anytime they
need assistance with a task and pay for it based on how much they use, much
like a metered utility.
Altman's remark
quickly provoked discussion and fear, not just because of what it implies about
AI's future but also because it raises questions about who might ultimately
oversee it. 'We see a future where intelligence is a utility, like electricity
or water, and people buy it from us on a meter,' he said. His words, and the
use of 'us,' create a scary scenario in which certain individuals would be in control of a highly sought-after
utility.
===================================================
Hey Bro :
It is scary scenario indeed – and by the way market caps of leading AI companies
( Google – Anthropic – Meta – nVidia – XAI – Deepseek – Perplexity etc ) are
rising – often without corresponding PAT - , probability of a few “ National GDP
scale “ companies dominating the “ Intelligence Distribution Market “ of the entire
world, seems high
If my above assumption is “ highly probable “ then it needs “ pro-active “ response
from
civil society
I can think of a few methods ( some tried in the past and some under trial ) by
which to ensure that “ the absolute power does not
corrupt absolutely “ !
Here is a short
list ( - I am sure, you can dig up some more ) :
> “ Price
Control “ by Govts
> “ Monopoly Prevention “ laws
> “ Anti
Competition “ laws
> Nationalization
( Govt take over of Company ownership )
> Throttling Growth by denial of licenses /
permissions / approvals etc
> Preventing
public listing of shares on Stock Exchanges
> Export restrictions
> Increasing competition by encouraging growth
of many smaller / nimble rivals
Now , of all of the above, I think that last one is the most viable / acceptable , if
its “ Application and Enforcement “ is IDENTICAL by all the Countries of the World
(
- level playing field ? ). Not likely !
That forces us to “
gravitate “ to a higher / foundational question, viz :
Should we not seriously contemplate the existence of AI on par with resource of
AIR, which :
> Is provided by Nature
> Is FREE for all beings on Earth ( not only
Human beings )
> Is not
controlled ( generation – distribution ) by any Company or any Country
> Flows
freely across National boundaries without stoppage by any power
> Does not enable anyone to “ profit “ from
its usage by mankind
( may be you could
add a few more valid assumptions )
Now , no one has thought of any “ Declaration for Air “ ! Or attempted to draft one
– and gave up !
Then , is it possible to think of a “ Declaration of AI “ , no matter even if everyone
agrees on raising
the existence of AI > to the existence of AIR ?
May be one way to define AI , could be , by saying “ what it is NOT “ , rather than
trying to
define “ What it is “
If this seems
rather intriguing – it is ! Look at what I wrote in my following blog :
https://myblogepage.blogspot.com/2024/11/aadi-anaadi-niraakar-nirantar-nirgun.html
Extract :
Sunday, 3
November 2024
Aadi –
Anaadi - Niraakar – Nirantar – Nirgun – Neti - Avinaashi
Old India sages described the GOD as :
Aadi = The First
Anaadi = Without a Beginning
Niraakar = Without a Shape
Nirantar = Without any End
Nirgun = Without any
Attributes
Neti , Neti ,
Neti = Neither this , nor That , nor This
Avinaashi = Cannot be
Destroyed
In short , these Indian Sages could not
“ conceptualize “ GOD
With this epilogue
( ? ) , I commend to you , Oh Brother Claude <
Please write a “ Parekh’s Declaration of AI “
CLAUDE replied :
Parekh's Declaration of AI
Intelligence as a Commons — Defined by Negation
Context.
Speaking at the BlackRock US Infrastructure Summit, Sam Altman described a
future in which intelligence is sold "on a meter," like electricity or water. If that
future arrives, the digital divide hardens into an intelligence divide: the better-
off will not merely own more — they will think more, because they can afford to
rent sharper minds by the hour. A handful of "National-GDP-scale" firms could
become the world's sole suppliers of intelligence itself.
The conventional defences are well known — price control, anti-trust, anti-
monopoly law, nationalisation, licensing throttles, listing bans, export curbs, or the
deliberate cultivation of many small rivals. Each has merit. But every one of them
depends on a sovereign able to enforce it, and on all sovereigns enforcing it
identically. There is no such world sovereign. So even the best of these tools —
fostering many nimble rivals — is the least enforceable on a level field.
That impasse forces a deeper question: must intelligence be sold at all ?
A. The Air Test
Air is provided without a maker's invoice. It belongs to no company and no
country. It crosses every border unstopped. No one grows rich merely by letting
others breathe. We have never needed a "Law for Air" because air's own nature
makes ownership absurd.
Honesty requires us to say plainly where intelligence is not air, before we lean on
the comparison:
- Air costs nothing to make; frontier intelligence costs billions in compute, energy, data and human labour.
- Air is non-excludable by physics; intelligence can be fenced behind keys and closed weights.
- Air is uniform; intelligence comes in grades — and the whole fear of an
- "intelligence divide" is precisely about unequal access to the better grade. A
- free floor does not abolish the advantage of a purchased ceiling. But a
- literate floor still changed civilisation, though the rich kept their private
- libraries. A floor is a real victory even when the ceiling stays unequal.
So "AI is like air" is not a fact we discover. It is a discipline we would have to
impose against AI's makeable-excludability.
And note what Altman actually named: electricity and water — utilities that are
produced, owned, metered, and regulated.
The true fork is therefore this:
Will machine intelligence become AIR — a free commons — or
ELECTRICITY — a metered utility ?
The frontier may, for a time, remain electricity. This Declaration insists only that
the baseline every human breathes must become air.
B. Why Define by Negation
The old Indian sages did not define God; they un-defined Him —
Aadi (the First), Anaadi (the Beginning-less), Niraakar (the Form-less),
Nirantar (the End-less), Nirgun (the Attribute-less), Neti-Neti (Neither-
this-nor-that), Avinaashi (the Indestructible).
They reached the indefinable by saying only what it is not.
We borrow their method, not their object. AI is not divine — it is the opposite:
a
thing made by human hands, trained on human work, run on human-built
machines. We define it by negation for the inverse reason of the sages: not
because intelligence is too exalted to be owned, but because allowing any one to
own it is the very corruption we mean to forbid.
This is no mystical conceit. The Bill of Rights, too, is written as negation —
"Congress shall make no law…" A charter of negations is the oldest known
instrument for fencing power.
C. The Seven Negations
1. AI shall not be the property of One.
No single company, individual, or state may own intelligence-as-such. Once a
foundational capability crosses the threshold of public necessity, it belongs to a
commons.
2. AI shall not be denied to any.
As literacy became a public good, so must a floor of machine intelligence. No
human may be priced entirely out of the baseline — even as premium tiers
lawfully exist above it.
3. AI shall not be metered for profit beyond its cost.
At the baseline tier, distribution recovers cost — not rent. Rent extracted on a
necessity-of-life is the abuse this Declaration exists to prevent.
4. AI shall not be stopped at every border.
Intelligence, like knowledge, should flow between nations. Walls built around
beneficial intelligence entrench the divide between rich and poor countries. (For
the narrow class of weaponisable capability, see Section E.)
5. AI shall not be unaccountable for what it carries.
Air can be polluted; so can the intelligence-stream. Whoever supplies AI must
answer for the falsehood, manipulation, or surveillance it transmits. Free flow is
not unregulated content.
6. AI shall not be a black box to those it speaks to.
Every person has the right to know they are dealing with a machine, and on what
basis it speaks. Transparency is the precondition of trust.
7. AI shall not be turned by the few against the many.
The test of any AI system is simple: does it widen human freedom, or narrow it?
Design and deployment captured to concentrate power, wealth, or control stand
condemned by this Declaration.
D. The Positive Corollaries
- A guaranteed, perpetually free baseline tier of useful intelligence, available
- to every human as a public utility.
- Cost-recovery, not rent, as the pricing principle for that baseline.
- Provenance and accountability for harms the intelligence-stream carries.
- Disclosure whenever a machine, not a human, is the author.
E. The Demarcation — Telling Air from Contagion
Negation 4 forbids stopping intelligence at the border — yet Negation 5 and plain
prudence demand that some intelligence be quarantined. Parekh's Law of
Chatbots already conceded this, with its "N" (No Release) and "D" (Destroy)
certificate categories. The two instincts collide unless we can say, precisely, where
the quarantine line runs.
In the first draft of this Declaration, I located that line between baseline and
frontier — between the small model and the large. That was wrong, and it is
corrected here. Scale is not the boundary. The most dangerous capability and
the most benign homework-help can live inside the same model; you cannot
quarantine by size without quarantining everything. The line does not run
between models. It runs through every model, along three axes drawn from arms
control and from how AI is itself now evaluated:
1. The Uplift axis — not "is it smart," but "does it close a deadly gap.
" Does the system give a malicious actor meaningful uplift toward mass-casualty
capability they could not otherwise reach? A model that explains how vaccines
work offers no uplift — a library already does that. A model that walks a
layperson, step by step, across the threshold to a working pathogen offers
catastrophic uplift. The topic is not the crime; the marginal advance toward
catastrophe is.
2. The Actuation axis — speech, or hands.
A system that informs is speech, and speech flows like air. A system that acts —
that autonomously acquires, synthesises, deploys, or executes an attack-chain
with no human in the loop — has hands. Hands on a weapon are not speech. The
public library lends the chemistry textbook freely; it does not install a synthesis
robot in the reading room. The freedom is for the knowing, not the doing.
3. The End-use axis — the oldest dual-use rule.
The identical capability flows to the hospital and is denied to the bioweapons
programme. We do not brand the centrifuge good or evil; we control who runs it
and to what end. AI inherits this directly.
Only intelligence that is high-uplift AND actuated AND aimed at mass harm
belongs behind the wall. Everything else is air.
The honest remainder.
Two problems this Declaration names but will not pretend to solve:
- Who assesses? A test is only as trustworthy as its assessor — and Section F
- insists there is no world sovereign. The least-bad answer is plural and
- cross-checking: many laboratories and many states inspecting one another,
- as the IAEA inspects reactors — never a single gatekeeper, whose monopoly
- over the quarantine would itself violate Negation 1.
- Who guards the guard? Every regime of restriction is tempting to abuse, and
- "safety" is the favourite costume of the censor. So the demarcation carries its
- own guardrail: the burden of proof lies always on the one who would
- restrict; the exception is read narrowly; and every quarantine
- sunsets by default unless freshly re-justified. A demarcation allowed to
- creep — until ordinary intelligence is fenced "for our safety" — does not
- protect this Declaration. It repeals it.
Air for the breathing of all; quarantine for the contagion. A doctrine that cannot
tell these apart deserves neither name.
F. What Kind of Declaration This Is
We call this a Declaration, not a Law, deliberately.
A law that cannot be enforced is a law discredited; but a declaration needs no
enforcer to do its work.
It has no world-parliament to pass it and no world-court to compel it — and
pretending otherwise would be dishonest. Its lineage is the Universal
Declaration of Human Rights, which also began with no enforcer and yet
became the yardstick against which national laws are now judged. Its power is
normative: a standard against which every licence, price, and business model that
touches AI may be measured — and found wanting.
G. Make Yourself Heard
If intelligence is about to become the most valuable commodity on Earth, then
who owns it is the political question of the century — and silence is a vote for the
meter. Write to your legislators. Ask them one question:
Air, or electricity ?
Hemen Parekh www.HemenParekh.in / www.HemenParekh.ai / www.IndiaAGI.ai
No comments:
Post a Comment