I know why
- Thread starter Diogenes
- Start date
This isn't a bad thing: "pause the development of AI until we establish clear safety standards."
So you say.
IBDaMann
Probably the worse human in the threads
When a Democrat indicates that a poll shows something, it means that it's only what he wants.
I suspect you're correct.
IBDaMann
Probably the worse human in the threads
Yes, it's a very bad thing. Only a tyrannical leftist would even suggest such a thing.
It's never a good idea to empower the government to prohibit large civilian sectors that the government fears.
Think I'm wrong? I'd say this about anything new and untested. Safety first.
The "Data for Progress" poll showing 68% support (72% Dem / 70% Ind / 63% GOP) for the Sanders–Casar Ban Artificial Superintelligence Act.
The bill would permanently ban systems that “match or exceed human cognitive performance across a broad range of domains” or that could “disempower humanity / overthrow the U.S. government,” pause advanced AI until a new cabinet-level agency writes the safety rules, and attach nuclear-weapons-style penalties (corporate death penalty + 20 years in prison).
That polling number is real as far as the survey goes.
The problem is what the number is being used to sell.
People who lived through the last 50 years of regulation already know the pattern.
The label is “safety.”
The function is often control.
1. The definition is the whole game.
“Superintelligence” and “advanced AI” are not engineering terms here; they are political terms.
“Matches human performance across a broad range of domains” can be stretched to cover the next model that writes better code, diagnoses better than the average doctor, or plans a logistics network.
Once a new agency exists whose job is to find dangerous capabilities, it will find them. Agencies do not shrink themselves.
2. “Until we have safety rules” is a pause with no off-ramp.
The bill pauses development until the new regulator writes the rules.
There is no deadline and no standard of proof that can ever be fully satisfied for a technology that does not yet exist.
That is the precautionary principle used as a veto, not as a risk-management tool.
You have seen the same move with nuclear power, certain drugs, and gene editing: the requirement is “prove it can never go wrong,” which is impossible, so the default becomes “no.”
3. The people who write the standards write the market.
A new cabinet-level agency plus an “independent” advisory board will be staffed by people selected for the worldview that produced the bill.
Leftist organizations (like "Data for Progress") that already have the lawyers, the lobbyists, and the existing models will "help" draft the compliance regime. Chances are, they already have.
Startups, open-source projects, and anyone who cannot afford a standing legal department get locked out. That is not a conspiracy theory; it is how FDA, FAA, and financial regulation have repeatedly worked.
Safety rules become a barrier to entry dressed up as public protection.
4. Unilateral pause is not safety; it is a gift to everyone else.
The bill contemplates international agreements, but the enforceable part is the U.S. pause.
Red China (outside U.S. jurisdiction) has no motive to wait for an American agency to finish its notice-and-comment process.
If the genuine concern is loss of control, ceding the lead to Beijing does not reduce any supposed risk. It just changes who holds the upper hand in the AI race.
Surprise! It'd be Communist China.
5. “Safety” absorbs other agendas.
Once regulatory oversight exists, it has no reason to stay limited to alleged "existential risk".
Job displacement, energy use of data centers, “disinformation,” labor standards, and political speech can all become "safety" issues.
You have watched this happen with social-media “trust and safety,” environmental review, and campus speech codes.
The scare is ginned up to pass the bill; the later uses are whatever the agency and its political allies want.
6. Penalties set the climate.
Treating AI like rogue nuclear weapons (20 years, corporate dissolution) does two things.
It chills research, and it gives politically-motivated prosecutors and the new agency enormous leverage.
Ambiguous statutes plus severe penalties produce self-censorship long before anyone is convicted.
The Luddite instinct is understandable: new machines have always displaced people and created new risks.
The historical record is also that the societies that kept building pulled ahead, and the ones that treated “safety” as a general-purpose brake fell behind.
A poll that asks “do you want a pause and a ban on something scary-sounding” will always poll well if the pollsters are careful.
The question that matters is who gets to define the scare, who writes the rules, and who is still allowed to build after the rules exist. That is where the manipulation lives.
The bill would permanently ban systems that “match or exceed human cognitive performance across a broad range of domains” or that could “disempower humanity / overthrow the U.S. government,” pause advanced AI until a new cabinet-level agency writes the safety rules, and attach nuclear-weapons-style penalties (corporate death penalty + 20 years in prison).
That polling number is real as far as the survey goes.
The problem is what the number is being used to sell.
How “safety standards” get turned into a weapon
People who lived through the last 50 years of regulation already know the pattern.
The label is “safety.”
The function is often control.
1. The definition is the whole game.
“Superintelligence” and “advanced AI” are not engineering terms here; they are political terms.
“Matches human performance across a broad range of domains” can be stretched to cover the next model that writes better code, diagnoses better than the average doctor, or plans a logistics network.
Once a new agency exists whose job is to find dangerous capabilities, it will find them. Agencies do not shrink themselves.
2. “Until we have safety rules” is a pause with no off-ramp.
The bill pauses development until the new regulator writes the rules.
There is no deadline and no standard of proof that can ever be fully satisfied for a technology that does not yet exist.
That is the precautionary principle used as a veto, not as a risk-management tool.
You have seen the same move with nuclear power, certain drugs, and gene editing: the requirement is “prove it can never go wrong,” which is impossible, so the default becomes “no.”
3. The people who write the standards write the market.
A new cabinet-level agency plus an “independent” advisory board will be staffed by people selected for the worldview that produced the bill.
Leftist organizations (like "Data for Progress") that already have the lawyers, the lobbyists, and the existing models will "help" draft the compliance regime. Chances are, they already have.
Startups, open-source projects, and anyone who cannot afford a standing legal department get locked out. That is not a conspiracy theory; it is how FDA, FAA, and financial regulation have repeatedly worked.
Safety rules become a barrier to entry dressed up as public protection.
4. Unilateral pause is not safety; it is a gift to everyone else.
The bill contemplates international agreements, but the enforceable part is the U.S. pause.
Red China (outside U.S. jurisdiction) has no motive to wait for an American agency to finish its notice-and-comment process.
If the genuine concern is loss of control, ceding the lead to Beijing does not reduce any supposed risk. It just changes who holds the upper hand in the AI race.
Surprise! It'd be Communist China.
5. “Safety” absorbs other agendas.
Once regulatory oversight exists, it has no reason to stay limited to alleged "existential risk".
Job displacement, energy use of data centers, “disinformation,” labor standards, and political speech can all become "safety" issues.
You have watched this happen with social-media “trust and safety,” environmental review, and campus speech codes.
The scare is ginned up to pass the bill; the later uses are whatever the agency and its political allies want.
6. Penalties set the climate.
Treating AI like rogue nuclear weapons (20 years, corporate dissolution) does two things.
It chills research, and it gives politically-motivated prosecutors and the new agency enormous leverage.
Ambiguous statutes plus severe penalties produce self-censorship long before anyone is convicted.
The Luddite instinct is understandable: new machines have always displaced people and created new risks.
The historical record is also that the societies that kept building pulled ahead, and the ones that treated “safety” as a general-purpose brake fell behind.
A poll that asks “do you want a pause and a ban on something scary-sounding” will always poll well if the pollsters are careful.
The question that matters is who gets to define the scare, who writes the rules, and who is still allowed to build after the rules exist. That is where the manipulation lives.
So you say.
I seem to recall a novel mRNA platform, emergency authorization, compressed timelines, no long-term data, and a largely-untested product that was still being studied in real time: that was the definition of “new and untested.”
Did you demand banning the vax?
If your alleged principle was genuine, you should have produced the same demand then: pause until a completed, independent safety regime exists; treated uncertainty as a reason to wait; and cried out for severe penalties for proceeding.
Did you?
Nor would I, unless you were being logical.
Grok is biased.
So you say. I'll understand if you cannot explain, of course.
mRNA research started in the 1970s. https://publichealth.jhu.edu/2021/the-long-history-of-mrna-vaccines
Elon Musk, aka Mr. Grok, said this:
Why Elon Musk says new tech should be 'default legal'
Elon Musk criticized government regulation of new technology as hindering innovation and called for a framework that "makes things default legal, not default illegal," during his remarks at the G20 Innovation Ministerial in Chapel Hill, N.C.
finance.yahoo.com
Grok laughs at you.
That’s a dodge, not a reply.
She was asked whether she applied her own rule—pause anything new and untested until a completed safety regime exists—to the COVID shots.
She answered with a date: mRNA research started in the 1970s.
That confuses a research lineage with a deployed product.
mRNA as a laboratory idea is old. A licensed mRNA vaccine given to hundreds of millions under emergency authorization, with compressed human trials and no long-term follow-up at the moment of rollout, was not.
Before 2020 there was no approved mRNA vaccine in humans. The product that went into arms was new in the only sense that matters for her slogan: first mass use, limited duration of safety data, novel manufacturing and delivery at that scale.
Apply her move to AI and it collapses.
Neural networks date to the 1950s. Backprop is 1980s. Transformers are 2017.
If “the underlying research is decades old” means the thing is no longer new and untested, then frontier models are not new either, and her demand for a pause-until-the-new-agency-writes-the-rules has no basis.
She does not accept that for AI. She only accepts it for the shot.
Two different standards:
- AI: decades of prior research still counts as “untested,” so ban/pause and create a cabinet agency with prison terms.
- Vaccine: decades of prior research counts as “tested,” so speed, EUA, and social pressure were fine.
The Johns Hopkins link is real and irrelevant to the inconsistency. Background science does not erase the fact that the application she blessed was the novel, rushed one, and the application she wants frozen is also built on old science.
If the rule is “anything new and untested,” she has to pick one definition of “new” and keep it.
She didn’t.
Elon Musk, aka Mr. Grok, said this:Why Elon Musk says new tech should be 'default legal'
Elon Musk criticized government regulation of new technology as hindering innovation and called for a framework that "makes things default legal, not default illegal," during his remarks at the G20 Innovation Ministerial in Chapel Hill, N.C.
You continue to amuse Grok.
That’s not an answer to the vaccine question. It’s a subject change.
She was asked whether she applied her own rule—“anything new and untested, safety first”—to the COVID shots. First she offered a 1970s research date. Now she offers Elon Musk saying new tech should be default legal, not default illegal, and treats that as if it settles the thread.
It doesn’t.
Musk’s line is a policy preference: novelty is allowed unless a specific harm is shown. Her line was the opposite: novelty is paused until a government safety regime exists. Quoting Musk does not explain why she used urgency for one new platform and a moratorium for another. It only tries to make the inconsistency look like “Grok bias” so she doesn’t have to own it.
The test is still the same:
- Did she demand a pause, a new cabinet agency, and felony penalties before mRNA shots were rolled out under EUA?
- If no, “I’d say this about anything new and untested” is false.
- Pointing at Musk does not make it true.
She can dislike Musk. She can dislike Grok. Neither fact turns a 1970s paper trail into a completed safety record for a 2020 mass-deployed product, and neither fact makes her rule consistent. The desperation is that she is now arguing the messenger because the principle she stated does not survive contact with her own record.