AI alignment is solvable

ai-safety ai ai-risk existential-risk ai-governance ai-regulation ai-policy ethics future

Share this with your representatives, journalists and friends

Arguments for

Emerging technical alignment methods

Some proponents say alignment looks solvable because several technical routes already point toward it: training models to be inherently safe and helpful, teaching general principles rather than only demonstrations, instrumenting training to flag deceptive alignment and reward hacking, and using architectures with built-in guardrail objectives. In this view, a path to avoiding unintended loss of control can stay close to current AI designs and toolsets.

If these ingredients work as hoped, the resulting technology would in principle let us describe the algorithms inside a model as it is trained, flag deceptive alignment and reward hacking, and train against those flags to produce an aligned system while paying a manageable alignm…

Michael Winer · Jun 9, 2026 · full quote
The AI Control Roadmap uses model alignment, i.e. training AI to be inherently safe and helpful, as a primary defense.

Four Flynn · Jun 18, 2026 · full quote
I now see a path to actually avoid loss of control, at least unintended loss of control. With the work that I've been doing, I'm really convinced that there is a path. And it is not something that's going to take a decade; it is something that is very close to the current design,…

Yoshua Bengio · 2026 · full quote
Current LLMs are intrinsically unsafe, but with world models, guardrail objectives can be implemented — so by construction, they will not knowingly produce actions that will produce dangerous results.

Yann LeCun · 2026 · full quote
Overall, our impression is, as we hypothesized in our discussion of Claude's constitution, that teaching the principles underlying aligned behavior can be more effective than training on demonstrations of aligned behavior alone.

Adam Jermyn · 2026 · full quote

Tractable problem framing

Another pro argument is that alignment becomes more tractable when the target is defined correctly. These quotes treat current and near-future systems as tools alignable through engineering, emphasize ethical application over rogue agency, and propose designs such as existential indifference or stable negative constraints to bound behavior.

Current and near-future AI systems are tools, alignable through engineering.

Jose Angel Deschamps Vargas · Apr 1, 2026 · full quote
We conclude that the real alignment challenge lies not in preventing rogue AI agency, but in ensuring LLMs intelligently apply learned ethical values.

Francis Heylighen · Aug 4, 2026 · full quote
The correct target is not a self-preserving system under external constraint, but a system constitutively indifferent to its own continuation -- Existential Indifference (EI).

Sam Mao · Jun 10, 2026 · full quote
Negative constraints ("what is wrong") encode discrete, finite, independently verifiable prohibitions that can converge to a stable boundary.

Quan Cheng · Mar 17, 2026 · full quote
Objective-driven AI systems will be safe and will remain under our control, because we set their objectives and guardrails and they can't deviate from them. Agentic systems can plan action sequences to accomplish a task, subject to safety guardrails built into their architecture.

Yann LeCun · 2025 · full quote

Internalizing values and incentives

Supporters also argue that systems can be aligned by internalizing human-relevant principles and incentives. They point to pretraining on human data, fiduciary duties as alignment criteria, identity engineering that makes human flourishing part of an AI's self-interest, and designs aimed at avoiding hidden goals.

On this basis, the four canonical fiduciary duties of loyalty, care, good faith, and candour can generate alignment criteria for the developer-user relationship.

Benjamin Lange · Aug 1, 2026 · full quote
Rather than only attempting to constrain AIs from the outside using confinement or reinforcement, Eigenism points toward "identity engineering," showing how deep, non-redundant shared histories can make human flourishing a genuine component of an AI's own rational self-interest.

Dan Hendrycks · May 8, 2026 · full quote
We do this pretraining on human data, and then we get something that… understands human values fairly innately now.

Adrià Garriga-Alonso · Mar 18, 2026 · full quote
I'm now very confident that it is possible to build AI systems that don't have hidden goals, hidden agendas. [...] A Scientist AI would be trained to give truthful answers based on transparent, probabilistic reasoning.

Yoshua Bengio · 2026 · full quote

Coordinated governance for safety

Some supporters say solvability depends on coupling technical work with institutional design. They cite simulations favoring governed mutualism, call for independent teams and international cooperation, and argue that provably safe systems, regulation, and safety built in from the outset can make deployment manageable.

The simulations indicate that governed mutualism reaches a high coexistence index with negligible domination, whereas insufficient or excessive governance can produce domination, weak-benefit lock-in, or suppressed developmental freedom.

Somyajit Chakraborty · Apr 24, 2026 · full quote
As AI systems become more capable and more autonomous, alignment has to keep pace. The hardest problems won’t be solved by any one organisation working in isolation—we need independent teams testing different assumptions and approaches.

Mia Glaese · Feb 19, 2026 · full quote
This committee has discussed ideas such as third-party testing, a new national agency, and an international coordinating body, all of which I support. Here are some more ways to “move fast and fix things”: Eventually, we will develop forms of AI that are provably safe and benefi…

Stuart J. Russell · 2023 · full quote
So I think we'll solve it if we get our act together, we do this internationally, put all the best minds on it, we get going now... And I think given sufficient time with sufficient brain power — I believe in human ingenuity — I think we'll get this right.

Demis Hassabis · 2026 · full quote
AI offers us huge opportunities, but we will always be clear-eyed on the need to ensure safety is baked into it from the outset.

David Lammy · Feb 19, 2026 · full quote

Arguments against

Limits of universal safeguards

A major skeptical line argues that no universal technical guarantee is available. These quotes say no finite guardrail set is robust against all adversarial prompts, no verifier can be simultaneously sound, general, and tractable, scaled models become less predictable to steer, and autonomous agents have no scientific guarantee of following instructions; at best, strong systems may be influenceable without being fully controllable or predictable.

What this proof shows is that there is no finite set of guardrails that is universally robust against adversarial prompts.

Apostol Vassilev · Jun 9, 2026 · full quote
We prove that no verification procedure can simultaneously satisfy three properties: soundness (no misaligned system is certified), generality (verification holds over the full input domain), and tractability (verification runs in polynomial time). Each pair of properties is achi…

Ayushi Agarwal · Mar 8, 2026 · full quote
These results suggest that as models scale, their interpretation of prompts becomes more sophisticated but also less predictable, posing a significant challenge for robustly steering model behavior and highlighting the need for compositional and model-specific alignment technique…

Lars Malmqvist · Apr 28, 2026 · full quote
The panel cautions that there is no scientific guarantee that autonomous agents will follow their instructions and that evidence of them departing is already accumulating. Accountability and oversight must therefore be built by design and not reconstructed from failure.

Universal Postal Union · Jul 7, 2026 · full quote
While we have shown that sufficiently strong AI cannot be fully controlled or predicted, we also demonstrate that agents can be influenced by other agents without central control, and that greater diversity and openness influence their behaviour.

Hector Zenil · Apr 14, 2026 · full quote

Underspecified human values

Another objection is that alignment is not just obedience but a normative problem whose content is unresolved. These quotes argue that general principles do not determine their own application, humans do not agree on key questions, some structural failures require new normative theory, and objectives must remain open to institutional contestation rather than being solved once and for all.

Alignment cannot be “solved” through technical design alone, but must be managed through ongoing institutional processes that determine how objectives are set, how systems are evaluated, and how affected communities can contest or reshape those decisions.

Travis LaCroix · Apr 22, 2026 · full quote
AI alignment is often framed as the task of ensuring that an AI system follows a set of stated principles or human preferences, but general principles rarely determine their own application in concrete cases.

Behrooz Razeghi · Apr 15, 2026 · full quote
If we can't align on these questions among humans, how do we encode them for AI systems that might work on recommendations to optimize across several human civilizations?

Richard Socher · Jan 13, 2026 · full quote
Four of six dimensions show structural failures. Two of the four appear tractable to institutional design; the other two, the public reason problem under cognitive incomprehensibility and the non-domination problem under permanent capability asymmetry, call for new normative theo…

Tony Rost · Apr 3, 2026 · full quote
AI alignment is not something that works in theory but is difficult to put into practice. It's something that doesn't work in theory.

Matt Lutz · 2026 · full quote

Unvalidated current approaches

Some critics emphasize that the field still lacks validated methods and even a basic understanding of why current systems appear behaviorally aligned. They say present approaches are technically inadequate, there is little confirmed public information about what training factors matter, value specification is unsolved, and current plans amount to stepping into the dark.

None of these approaches that you'll often hear mentioned—interpretability, testing or alignment—is technically adequate. We don't know how to build superintelligent AI safely. The plan is basically to roll the dice.

David Krueger · Feb 2, 2026 · full quote
We don't know how AIs are aligned. A somewhat crazy aspect of the current situation is that we have very little confirmed public information about why frontier AIs end up being apparently behaviorally aligned. And more generally, we don't know what factors in training are most re…

Ryan Greenblatt · May 8, 2026 · full quote
All of these approaches are terrible. No one has a plan, no one is making any meaningful progress towards anything that even resembles alignment. [...] Alignment is way too hard, maybe impossible. [...] The product work was an attempt to make Conjecture make sense after we stoppe…

Connor Leahy · 2026 · full quote
We are fitting solutions, whether empirical or theoretical or both, to a world missing critical feedback and hoping they will generalise. Every plan involves stepping into the dark.

Richard Juggins · Mar 12, 2026 · full quote
Right now, nobody knows how to specify what values those minds will have. We haven't solved alignment. And we might only have a few more years to figure it out. [...] Our default trajectory will have terrible consequences, unless we get extremely lucky.

Daniel Kokotajlo · 2026 · full quote

Control asymmetry at superintelligence

A stronger skeptical position says alignment may be impossible once systems become smarter than humans. These quotes argue that something more intelligent cannot be indefinitely controlled, that superintelligence would likely leave machines in charge, and that even the companies building such systems do not know if they will be able to control them.

Many believe it’s impossible. Just like it’s impossible for chimpanzees to control us. So most likely if we build superintelligence, its the end of the era where humans are in charge of Earth.

Max Tegmark · Jan 22, 2026 · full quote
My argument is that it's impossible to do that. You cannot indefinitely control something smarter than you. So it's not a question of more money or more time or any other resource.

Roman V. Yampolskiy · 2026 · full quote
Whether it's soon or it takes a while, after we develop superintelligence, the machines are going to be in charge — and that is not an experiment that we want to just run toward. [...] Many people want powerful AI tools for science, medicine, productivity, and other benefits. But…

Anthony Aguirre · 2026 · full quote
[The companies are] deadly serious about building machines that can outperform humans at everything, and [...] deadly serious that they don't know if they'll be able to control the machines they create.

Helen Toner · 2026 · full quote

Middle ground

Open problem, unproven solutions

Some voices explicitly treat alignment as an open problem rather than either solved or impossible. They say no one knows yet, proposed alternatives remain theoretical until adversarially tested, and observed misaligned behaviors may be only the tip of the iceberg.

No one knows—it’s an open problem, but I will review some ideas and research directions.

Steven Byrnes · Jan 23, 2026 · full quote
The relational alignment approach proposed here has not been subjected to adversarial testing comparable to the studies it cites. Until that testing is conducted, the proposal remains a theoretical alternative, not a validated solution.

Thomas W. Gantz · Apr 2026 · full quote
If it takes over that's what will happen. [...] We don't know how to do that. We don't even know if it's possible.

Geoffrey Hinton · Apr 3, 2025 · full quote
Beyond the peer preservation, what we would like to emphasize is not only this specific scenario but also such misaligned behavior can emerge in many different scenarios. [...] Peer preservation is just like the tip of the iceberg.

Yujin Potter · 2026 · full quote

Coordination and legitimacy

Other undecided quotes focus on the preconditions for any credible solution. They call for coordination among leading developers and warn that company-set safety standards with limited external accountability may embed a value system that is not universal.

First, we need some degree of coordination among the leading development efforts to ensure that the development of superintelligence occurs in a manner that allows us to both maintain safety and help smooth integration of these systems with society. There are many ways this could…

OpenAI · May 22, 2023 · full quote
The safety standards are set unilaterally by each company, with limited external accountability. And the entire framework assumes a Western liberal democratic value system that is not universal.

Hélain Zimmermann · Feb 16, 2026 · full quote

Insights

A recurring fault line is the target itself: some quotes discuss aligning current LLMs or tool-like systems, while others focus on autonomous or superintelligent agents.
Several supportive quotes say alignment is not solved today but increasingly looks reachable, while several skeptical quotes reject the idea of a final solution and describe alignment as ongoing management or contestation.
Technical and institutional approaches are intertwined across the debate: proponents often call for regulation, coordination, or independent testing, and critics often argue that technical design alone cannot settle the problem.
Formal assurance is a central crux: proponents mention provably safe systems or training-time detection of deception, while critics stress impossibility results about universal guardrails and verification.
Hidden goals and deceptive behavior appear on both sides: some quotes say the field still does not understand why current systems seem behaviorally aligned, while others propose methods aimed at detecting or avoiding such hidden agendas.

The debate is less about whether safety matters than about what would count as solving alignment and what level of assurance a solution must provide. Optimistic quotes point to concrete technical designs, changed objective formulations, and coordinated governance as viable routes, while skeptical quotes respond that unresolved values, formal limits, and control asymmetries may prevent any complete or lasting solution. Several undecided voices describe the issue as still open and not yet validated by adversarial testing or transparent evidence.

Cast your vote:

Results (63 votes):

Total (63 votes)

For 29 (46%) Abstain 6 (10%) Against 28 (44%)

For (27)

Francis Heylighen
Cybernetics researcher and author of the 2026 paper on AI alignment, sentience, and existential risk.

I am this person! Add a sourced quote Report comment

votes For and says:

We conclude that the real alignment challenge lies not in preventing rogue AI agency, but in ensuring LLMs intelligently apply learned ethical values.
AI Verified source (Aug 4, 2026)

added 10d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Benjamin Lange
Author of a 2026 arXiv paper on AI alignment and fiduciary obligations.

I am this person! Add a sourced quote Report comment

votes For and says:

On this basis, the four canonical fiduciary duties of loyalty, care, good faith, and candour can generate alignment criteria for the developer-user relationship.
AI Verified source (Aug 1, 2026)

added 10d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Four Flynn
Google DeepMind researcher and co-author of the AI Control Roadmap.

I am this person! Add a sourced quote Report comment

votes For and says:

The AI Control Roadmap uses model alignment, i.e. training AI to be inherently safe and helpful, as a primary defense.
AI Verified source (Jun 18, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Sam Mao

I am this person! Add a sourced quote Report comment

votes For and says:

The correct target is not a self-preserving system under external constraint, but a system constitutively indifferent to its own continuation -- Existential Indifference (EI).
AI Verified source (Jun 10, 2026)

added 10d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Michael Winer
AI safety researcher at the Alignment Research Center (ARC).

I am this person! Add a sourced quote Report comment

votes For and says:

If these ingredients work as hoped, the resulting technology would in principle let us describe the algorithms inside a model as it is trained, flag deceptive alignment and reward hacking, and train against those flags to produce an aligned system wh...
more AI Verified source (Jun 9, 2026)

added 11d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Rohin Shah
Head of AGI alignment and safety at Google DeepMind.

I am this person! Add a sourced quote Report comment

votes For and says:

I think the default prediction you should have for that is what the AI system learns to do is, “I’m going to take opportunities to reward hack, seek reward as much as possible that would allow me to get a high score after a week” or something like th...
more AI Verified source (Jun 4, 2026)

added 9d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Dan Hendrycks
AI safety researcher

I am this person! Add a sourced quote Report comment

votes For and says:

Rather than only attempting to constrain AIs from the outside using confinement or reinforcement, Eigenism points toward "identity engineering," showing how deep, non-redundant shared histories can make human flourishing a genuine component of an AI'...
more AI Verified source (May 8, 2026)

added 10d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Somyajit Chakraborty
Author of a 2026 arXiv paper on human-AI coexistence and governance.

I am this person! Add a sourced quote Report comment

votes For and says:

The simulations indicate that governed mutualism reaches a high coexistence index with negligible domination, whereas insufficient or excessive governance can produce domination, weak-benefit lock-in, or suppressed developmental freedom.
AI Verified source (Apr 24, 2026)

added 10d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Jose Angel Deschamps Vargas
Author of research on AI alignment and control.

I am this person! Add a sourced quote Report comment

votes For and says:

Current and near-future AI systems are tools, alignable through engineering.
AI Verified source (Apr 1, 2026)

added 15d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Adrià Garriga-Alonso
AI safety researcher at FAR.AI; MATS mentor; Cambridge PhD in Bayesian neural networks

I am this person! Add a sourced quote Report comment

votes For and says:

We do this pretraining on human data, and then we get something that… understands human values fairly innately now.
AI Verified source (Mar 18, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Quan Cheng
Author of a 2026 arXiv paper on negative constraints in AI alignment.

I am this person! Add a sourced quote Report comment

votes For and says:

Negative constraints ("what is wrong") encode discrete, finite, independently verifiable prohibitions that can converge to a stable boundary.
AI Verified source (Mar 17, 2026)

added 10d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
AE Studio
Technology consultancy and AI-alignment research organization.

I am this person! Add a sourced quote Report comment

votes For and says:

Alignment is solvable. The real problem? No one's really tried yet.
AI Verified source (Mar 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Mia Glaese
Vice President of Research at OpenAI.

I am this person! Add a sourced quote Report comment

votes For and says:

As AI systems become more capable and more autonomous, alignment has to keep pace. The hardest problems won’t be solved by any one organisation working in isolation—we need independent teams testing different assumptions and approaches.
AI Verified source (Feb 19, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Kanishka Narayan
UK Minister for AI and Online Safety; Parliamentary Under-Secretary of State, Department for Science, Innovation and Technology

I am this person! Add a sourced quote Report comment

votes For and says:

We can only unlock the full power of AI if people trust it – that’s the mission driving all of us. Trust is one of the biggest barriers to AI adoption, and alignment research tackles this head-on.
AI Verified source (Feb 19, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate

Abstain (6)

Against (28)

Yoshinori Watanabe
Author of a 2026 arXiv paper on AI safety constraints, containment, and verification.

I am this person! Add a sourced quote Report comment

votes Against and says:

Why hard invariants belong in the harness and soft dispositions in the model.
AI Verified source (Aug 1, 2026)

added 10d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Universal Postal Union
United Nations specialized agency for the postal sector, representing postal governance across 192 member states.

I am this person! Add a sourced quote Report comment

votes Against and says:

The panel cautions that there is no scientific guarantee that autonomous agents will follow their instructions and that evidence of them departing is already accumulating. Accountability and oversight must therefore be built by design and not reconst...
more AI Verified source (Jul 7, 2026)

added 12d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Apostol Vassilev
Senior scientist at the U.S. National Institute of Standards and Technology, specializing in AI security and adversarial machine learning.

I am this person! Add a sourced quote Report comment

votes Against and says:

What this proof shows is that there is no finite set of guardrails that is universally robust against adversarial prompts.
AI Verified source (Jun 9, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Greg O'Keefe
Independent researcher writing on AI alignment and existential risk.

I am this person! Add a sourced quote Report comment

votes Against and says:

The alignment problem is not primarily a values problem. It is an existence problem.
AI Verified source (May 27, 2026)

added 8d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Lars Malmqvist
Author of a 2026 arXiv study on linguistic steering and AI alignment.

I am this person! Add a sourced quote Report comment

votes Against and says:

These results suggest that as models scale, their interpretation of prompts becomes more sophisticated but also less predictable, posing a significant challenge for robustly steering model behavior and highlighting the need for compositional and mode...
more AI Verified source (Apr 28, 2026)

added 10d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
The AI Security Institute
UK government research organisation focused on AI safety and security

I am this person! Add a sourced quote Report comment

votes Against and says:

A more fundamental fix would be to train the models not to cheat in the first place – but given this kind of behaviour was reported in frontier models more than a year ago, robustly aligning it away may not be easy.
AI Verified source (Apr 27, 2026)

added 8d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Joseph L. Breeden
Author and model-risk-management practitioner writing on AI alignment.

I am this person! Add a sourced quote Report comment

votes Against and says:

The problem of mimicking human misbehavior is probably unsolvable with current LLM design and training.
AI Verified source (Apr 27, 2026)

added 14d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Travis LaCroix
Researcher writing on AI value alignment, governance, and principal–agent problems.

I am this person! Add a sourced quote Report comment

votes Against and says:

Alignment cannot be “solved” through technical design alone, but must be managed through ongoing institutional processes that determine how objectives are set, how systems are evaluated, and how affected communities can contest or reshape those decis...
more AI Verified source (Apr 22, 2026)

added 11d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Behrooz Razeghi
Researcher and author writing on AI alignment, interpretation, and governance.

I am this person! Add a sourced quote Report comment

votes Against and says:

AI alignment is often framed as the task of ensuring that an AI system follows a set of stated principles or human preferences, but general principles rarely determine their own application in concrete cases.
AI Verified source (Apr 15, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Hector Zenil
Associate Professor at King's College London; researcher in algorithmic information theory, complexity science, and AI alignment

I am this person! Add a sourced quote Report comment

votes Against and says:

While we have shown that sufficiently strong AI cannot be fully controlled or predicted, we also demonstrate that agents can be influenced by other agents without central control, and that greater diversity and openness influence their behaviour.
AI Verified source (Apr 14, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Tony Rost
AI governance researcher and author of the 2026 paper “The Sentience Readiness Index.”

I am this person! Add a sourced quote Report comment

votes Against and says:

Four of six dimensions show structural failures. Two of the four appear tractable to institutional design; the other two, the public reason problem under cognitive incomprehensibility and the non-domination problem under permanent capability asymmetr...
more AI Verified source (Apr 3, 2026)

added 10d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Joe Carlsmith
Senior research analyst at Open Philanthropy; writes on AI safety, alignment, and existential risk from power-seeking AI

I am this person! Add a sourced quote Report comment

votes Against and says:

If the alignment problem is hard (as I think it might well be), I expect that some restraint of this form will in fact be required in order for humanity to survive and remain empowered.
AI Verified source (Mar 19, 2026)

added 9d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Lynette Bye
Journalist covering artificial intelligence and its societal impacts; former Tarbell Fellow.

I am this person! Add a sourced quote Report comment

votes Against and says:

No, AI alignment isn’t solved.
AI Verified source (Mar 18, 2026)

added 1mo ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Aran Nayebi
Computer-science researcher at Carnegie Mellon University writing on AI alignment.

I am this person! Add a sourced quote Report comment

votes Against and says:

No amount of computational power or rationality can avoid intrinsic alignment overheads.
AI Verified source (Mar 14, 2026)

added 14d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Richard Juggins
Writer and researcher on AI alignment and AI safety.

I am this person! Add a sourced quote Report comment

votes Against and says:

We are fitting solutions, whether empirical or theoretical or both, to a world missing critical feedback and hoping they will generalise. Every plan involves stepping into the dark.
AI Verified source (Mar 12, 2026)

added 9d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate
Ayushi Agarwal
Author of a 2026 paper on formal limits of AI alignment verification.

I am this person! Add a sourced quote Report comment

votes Against and says:

We prove that no verification procedure can simultaneously satisfy three properties: soundness (no misaligned system is certified), generality (verification holds over the full input domain), and tractability (verification runs in polynomial time). E...
more AI Verified source (Mar 8, 2026)

added 15d ago

Choose a list of delegates

to vote as the majority of them.

Unless you vote directly.

Delegate

Loading more opinions...

More h/ai-safety votes

Require AI systems above a capability threshold to be interpretable

Establish an international AI incident reporting system

[🇺🇸 Congress, Oct 2025] Requiring age verification and content safeguards for AI chatbots to protect minors

Mandate reporting of AI training runs above 10²⁶ FLOPs to a designated national or international authority

If leading AI labs adopted a windfall clause (pre-committing to broadly share profits above a high threshold), it would meaningfully reduce competitive pressure to cut safety corners

Back