| Score |
Entry 1 |
Entry 2 |
| (1.00) |
What are finite factored sets? |
What are finite factored sets? |
| (1.00) |
What is relaxed adversarial training? |
What is relaxed adversarial training? |
| (1.00) |
How could an intelligence explosion be useful? |
How could an intelligence explosion be useful? |
| (1.00) |
What beneficial things would an aligned superintelligence be able to do? |
What beneficial things would an aligned superintelligence be able to do? |
| (0.98) |
What is Conjecture's main research agenda? |
What is Conjecture's research agenda? |
| (0.98) |
Why can't we build an AI that is programmed to shut off after some time? |
Why can’t we build an AI that is programmed to turn off after some time? |
| (0.97) |
What exactly does “AI alignment” mean? |
What is AI alignment? |
| (0.96) |
What are some introductory videos about AI safety? |
Where can I find videos about AI safety? |
| (0.96) |
What are Responsible Scaling Policies (RSPs)? |
What is a responsible scaling policy (RSP)? |
| (0.96) |
Which organizations are working on AI alignment? |
What organizations are working on technical AI alignment? |
| (0.93) |
What is the weak scaling hypothesis? |
What is the strong scaling hypothesis? |
| (0.93) |
What is the general nature of the concern about AI alignment? |
What are some objections to the importance of AI alignment? |
| (0.92) |
What is offline reinforcement learning (RL)? |
What is online reinforcement learning (RL)? |
| (0.92) |
What is AI-assisted alignment? |
What is AI alignment? |
| (0.92) |
What is online reinforcement learning (RL)? |
What is reinforcement learning (RL)? |
| (0.92) |
What are some proposed training techniques to solve outer misalignment? |
What are some proposed training techniques to solve inner misalignment? |
| (0.92) |
The alignment problem |
How can we solve the alignment problem? |
| (0.91) |
Should selfish people care about AI safety? |
Are there “selfish” reasons for caring about AI safety? |
| (0.91) |
Other resources |
Resources elsewhere |
| (0.91) |
Is AI alignment easy? |
Aren't there easy solutions to AI alignment? |
| (0.91) |
How can I help AI alignment researchers be more effective? |
Are there promising ways to make AI alignment researchers smarter? |
| (0.91) |
What are the capabilities of GPT-4? |
What is GPT-4 and what is it capable of? |
| (0.91) |
Why is AI alignment a hard problem? |
At a high level, what is the challenge of AI alignment? |
| (0.90) |
We have dealt with dangerous technologies before, why is AGI different? |
Other technologies have been deemed potentially world-ending, why is AGI different? |
| (0.90) |
What is AI: Futures and Responsibility (AI:FAR)'s research agenda? |
What is FAR AI's research agenda? |
| (0.90) |
What is this site about? |
What is this website about? |
| (0.90) |
Do we have an example/evidence of outer misalignment? |
Do we have an example/evidence of inner misalignment? |
| (0.90) |
What should I do with my machine learning research idea for AI alignment? |
What should I do with my idea for helping with AI alignment? |
| (0.89) |
What is Contrast Consistent Search (CCS)? |
What is Cross-Contrast Search (CCS)? |
| (0.89) |
What is the strong scaling hypothesis? |
What is the scaling hypothesis? |
| (0.89) |
What is out of context learning? |
What is in-context learning? |
| (0.89) |
Isn't the real concern AI-enabled authoritarianism? |
Isn't the real concern AI-enabled totalitarianism? |
| (0.89) |
Is AI alignment easy? |
Is AI alignment possible? |
| (0.89) |
What can't AI do yet? |
What will an AI never be able to do? |
| (0.89) |
What is the UK's AI Security Institute? |
What is the UK’s AI Safety Institute? |
| (0.89) |
What is a Task AI? |
What is a Task-directed AI? |
| (0.89) |
What is GPT-4? |
What is GPT-4 and what is it capable of? |
| (0.89) |
What are "reasoning" AI models? |
What are “Simulated reasoning” AI models? |
| (0.89) |
Would a misaligned superintelligence kill literally everyone? |
Why would a misaligned superintelligence kill everyone? |
| (0.89) |
Why do people disagree on the likelihood of existential risks from AI? |
What are some arguments against AI being an existential risk? |
| (0.88) |
What are the different AI Alignment / Safety organizations and academics researching? |
Briefly, what are the major AI safety organizations and academics working on? |
| (0.88) |
Aren't there easy solutions to AI alignment? |
What would a good solution to AI alignment look like? |
| (0.88) |
Which alignment strategies can scale to superintelligence? |
What concrete work is being done on alignment strategies which won’t scale to superintelligence? |
| (0.88) |
What other options are there for pursuing a technical career in AI alignment? |
How can I build a career in AI alignment? |
| (0.88) |
Aren't there easy solutions to AI alignment? |
Is AI alignment possible? |
| (0.88) |
How powerful could a superintelligence become? |
Isn't it impossible for a superintelligence to become very powerful? |
| (0.88) |
Why do people disagree on the likelihood of existential risks from AI? |
Do people seriously worry about existential risk from AI? |
| (0.88) |
Future AI |
Predictions about future AI |
| (0.88) |
What is the general nature of the concern about AI alignment? |
At a high level, what is the challenge of AI alignment? |
| (0.88) |
What is concept distribution shift ? |
What is a distributional shift? |
| (0.88) |
Can't we just tell an AI to do what we want? |
Can we tell an AI just to figure out what we want and then do that? |
| (0.88) |
Why not just control AI? |
Why not just let AI take over? |
| (0.87) |
How can I work on public AI safety outreach? |
How can I work on AI safety outreach in academia and among experts? |
| (0.87) |
Alignment techniques |
Other alignment approaches |
| (0.87) |
What philosophical approaches are used in AI alignment? |
How can I do conceptual, mathematical, or philosophical work on AI alignment? |
| (0.87) |
What is the objective based perspective on the alignment problem? |
What is the optimization based perspective on the alignment problem? |
| (0.87) |
Could AI alignment research be bad? How? |
Might AI alignment research lead to outcomes worse than extinction? |
| (0.87) |
What are soft optimizers? |
What is soft optimization? |
| (0.87) |
What are Francois Chollet’s criticisms of AI Alignment? |
What are Andrew Ng’s criticisms of AI Alignment? |
| (0.87) |
What is outer alignment? |
What is the difference between inner and outer alignment? |
| (0.87) |
What is scaffolding? |
What are scaffolds? |
| (0.87) |
What exactly does “AI alignment” mean? |
What is AI-assisted alignment? |
| (0.87) |
What is reward modeling? |
What is recursive reward modeling? |
| (0.87) |
What are some existing alignment strategies, and what are their pitfalls? |
What alignment strategies are scalably safe and competitive? |
| (0.87) |
Does talk of existential risk from AI detract from current harms? |
Do people seriously worry about existential risk from AI? |
| (0.86) |
We have dealt with dangerous technologies before, why is AGI different? |
Why don't we just not build AGI if it's so dangerous? |
| (0.86) |
What are some objections to the importance of AI alignment? |
At a high level, what is the challenge of AI alignment? |
| (0.86) |
What are some proposed training techniques to solve outer misalignment? |
What are some proposed training techniques to solve deceptive misalignment? |
| (0.86) |
What are soft optimizers? |
What are mild optimizers? |
| (0.86) |
How could an intelligence explosion be useful? |
How might an "intelligence explosion" be dangerous? |
| (0.86) |
How could an intelligence explosion be useful? |
How might an "intelligence explosion" be dangerous? |
| (0.86) |
What benchmarks exist for evaluating the safety of AI systems? |
What benchmarks exist for measuring the capabilities of AI systems? |
| (0.86) |
Why aren't more people worried if superintelligence is so dangerous? |
Why should I worry about superintelligence? |
| (0.86) |
What would a good solution to AI alignment look like? |
Is AI alignment possible? |
| (0.86) |
Might an "intelligence explosion" never occur? |
How likely is an intelligence explosion? |
| (0.86) |
What does it mean for an AI to think? |
Can an AI really think? |
| (0.86) |
Why is AI alignment a hard problem? |
Aren't there easy solutions to AI alignment? |
| (0.86) |
Why is AI alignment a hard problem? |
Why do some AI researchers not worry about alignment? |
| (0.86) |
Difficulty of alignment |
How difficult should we expect alignment to be? |
| (0.86) |
Why is AI alignment a hard problem? |
How does AI taking things literally contribute to alignment being hard? |
| (0.86) |
Could AI alignment research be bad? How? |
What is the general nature of the concern about AI alignment? |
| (0.86) |
A case for AI safety |
What is AI safety? |
| (0.86) |
We have dealt with dangerous technologies before, why is AGI different? |
Why is AGI more dangerous than nanotechnology or biology? |
| (0.85) |
Could AI alignment research be bad? How? |
What are some objections to the importance of AI alignment? |
| (0.85) |
What is reward design? |
What is reward modeling? |
| (0.85) |
What is decision theory? |
What is "logical decision theory"? |
| (0.85) |
Governance research organizations |
Governance research |
| (0.85) |
What is a distribution shift and how is it related to alignment? |
What is a distributional shift? |
| (0.85) |
How could an intelligence explosion be useful? |
What is an intelligence explosion? |
| (0.85) |
What is an intelligence explosion? |
How could an intelligence explosion be useful? |
| (0.85) |
What are Francois Chollet’s criticisms of AI Alignment? |
What are Yann LeCun’s criticisms of AI Alignment? |
| (0.85) |
What is an "AI doomer"? |
What is “AI doomerism”? |
| (0.85) |
Why do some AI researchers not worry about alignment? |
What is the general nature of the concern about AI alignment? |
| (0.85) |
Superintelligence |
Implications of superintelligence |
| (0.85) |
What are the arguments for a slow takeoff? |
Why might we expect a fast takeoff? |
| (0.85) |
How can I contribute to AI safety as a volunteer? |
What are some simple things I can do to contribute to AI safety? |
| (0.85) |
What is narrow reward modeling? |
What is reward modeling? |
| (0.85) |
Why do some AI researchers not worry about alignment? |
What are some objections to the importance of AI alignment? |
| (0.85) |
What are Andrew Ng’s criticisms of AI Alignment? |
What are Richard Ngo's views on AI alignment? |
| (0.85) |
Should selfish people care about AI safety? |
If I only care about helping people alive today, does AI safety still matter? |