#8: Max Alexander and I solve ethics, philosophy of mind, and cancel culture once and for all
- Follow Max on Twitter: twitter.com/absurdlymax'
- And read his blog: scouting-ahead.com
- Full (but imperfect) transcript at: https://bit.ly/maxonpigeonhour
Summary
In this philosophical and reflective episode, hosts Aaron and Max engage in a profound debate over the nature of consciousness, moral realism, and subjective experience. Max, a skeptic of moral realism, challenges Aaron on the objective moral distinction between worlds with varying levels of suffering. They ponder the hard problem of consciousness, discussing the possibility of philosophical zombies and whether computations could account for consciousness. As they delve into the implications of AI on moral frameworks, their conversation extends to the origins of normativity and the nonexistence of free will.
The tone shifts as they discuss practical advice for running an Effective Altruism group, emphasizing the importance of co-organizers and the balance between being hospitable and maintaining normalcy. They exchange views on the potential risks and benefits of being open in community building and the value of transparency and honest feedback.
Transitioning to lighter topics, Max and Aaron share their experiences with social media, the impact of Twitter on communication, and the humorous side of office gossip. They also touch on the role of anonymity in online discussions, pondering its significance against the backdrop of the Effective Altruism community.
As the episode draws to a close, they explore the consequences of public online behavior for employment and personal life, sharing anecdotes and contemplating the broader implications of engaging in sensitive discourses. Despite their digressions into various topics, the duo manages to weave a coherent narrative of their musings, leaving listeners with much to reflect upon.
#7: Holly Elmore on AI pause, wild animal welfare, and some cool biology things I couldn't fully follow but maybe you can
Blurb and summary from Clong:
Holly and Aaron had a wide-ranging discussion touching on effective altruism, AI alignment, genetic conflict, wild animal welfare, and the importance of public advocacy in the AI safety space. Holly spoke about her background in evolutionary biology and how she became involved in effective altruism. She discussed her reservations around wild animal welfare and her perspective on the challenges of AI alignment. They talked about the value of public opinion polls, the psychology of AI researchers, and whether certain AI labs like OpenAI might be net positive actors. Holly argued for the strategic importance of public advocacy and pushing the Overton window within EA on AI safety issues.
Here is a more detailed summary in list format:
- Holly's background - PhD in evolutionary biology, got into EA through New Atheism and looking for community with positive values, did EA organizing at Harvard
- Worked at Rethink Priorities on wild animal welfare but had reservations about imposing values on animals and whether we're at the right margin yet
- Got inspired by FLI letter to focus more on AI safety advocacy and importance of public opinion
- Discussed genetic conflict and challenges of alignment even with "closest" agents
- Talked about the value of public opinion polls and influencing politicians
- Discussed the psychology and motives of AI researchers
- Disagreed a bit on whether certain labs like OpenAI might be net positive actors
- Holly argued for importance of public advocacy in AI safety, thinks we have power to shift Overton window
- Talked about the dynamics between different AI researchers and competition for status
- Discussed how rationalists often dismiss advocacy and politics
- Holly thinks advocacy is neglected and can push the Overton window even within EA
- Also discussed Holly's evolutionary biology takes, memetic drive, gradient descent vs. natural selection
Aaron's Blog, podcast edition: The never-before-told story of my podcasting misadventure [readout from my blog]
Post: https://www.aaronbergman.net/p/rob-and-keiran-on-the-philosophy-3f7
****************************
This is a bot readout of a blog post that I published a while ago but only just now made audio for. If the voice sounds familiar, that's because it's the same one from AWS used by the Nonlinear Library, which publishes automated readouts of EA Forum and LessWrong posts) as a podcast feed.
For God's sake, Google it [readout from my blog]
Post: https://www.aaronbergman.net/p/for-gods-sake-google-it
****************************
This is a bot readout of a blog post that I published a while ago but only just now made audio for. If the voice sounds familiar, that's because it's the same one from AWS used by the Nonlinear Library, which publishes automated readouts of EA Forum and LessWrong posts) as a podcast feed.
"Tf is a kernel?" [readout from my blog]
Post: https://www.aaronbergman.net/p/tf-is-a-kernel
****************************
This is a bot readout of a blog post that I published a while ago but only just now made audio for. If the voice sounds familiar, that's because it's the same one from AWS used by the Nonlinear Library, which publishes automated readouts of EA Forum and LessWrong posts) as a podcast feed.
Stuff I buy and use [readout from my blog]
Post: https://www.aaronbergman.net/p/stuff-i-buy-and-use
****************************
This is a bot readout of a blog post that I published a while ago but only just now made audio for. If the voice sounds familiar, that's because it's the same one from AWS used by the Nonlinear Library, which publishes automated readouts of EA Forum and LessWrong posts) as a podcast feed.
The topic is not the content [readout from my blog]
Post: https://www.aaronbergman.net/p/the-topic-is-not-the-content
****************************
This is a bot readout of a blog post that I published a while ago but only just now made audio for. If the voice sounds familiar, that's because it's the same one from AWS used by the Nonlinear Library, which publishes automated readouts of EA Forum and LessWrong posts) as a podcast feed.
Deficits are a social technology to shape the real world [readout from my blog]
Post: https://www.aaronbergman.net/p/deficits-dont-matter-spending-does
****************************
This is a bot readout of a blog post that I published a while ago but only just now made audio for. If the voice sounds familiar, that's because it's the same one from AWS used by the Nonlinear Library, which publishes automated readouts of EA Forum and LessWrong posts) as a podcast feed.
A progressive-ish case against the minimum wage [readout from my blog]
Post: https://www.aaronbergman.net/p/a-progressive-ish-case-against-the
****************************
This is a bot readout of a blog post that I published a while ago but only just now made audio for. If the voice sounds familiar, that's because it's the same one from AWS used by the Nonlinear Library, which publishes automated readouts of EA Forum and LessWrong posts) as a podcast feed.
Suffering focused total utilitarianism [readout from my blog]
This is a bot readout of my blog post "Suffering focused total utilitarianism" (www.aaronbergman.net/p/my-case-for-suffering-leaning-ethics), which I published a while ago but only just now made audio for.
If the voice sounds familiar, that's because it's the same one from AWS used by the Nonlinear Library, which publishes automated readouts of EA Forum and LessWrong posts) as a podcast feed.
****************************
* Blurb (AI-generated) *
****************************
In this insightful blog post, Aaron Bergman delves deep into a philosophical debate, challenging the common belief that reducing the risk of extinction is always morally commendable. Bergman posits that some levels of suffering are so intense that they can't be counterbalanced by any conceivable degree of happiness. He critiques utilitarian perspectives and raises concerns about future scenarios that might usher in immense suffering. Though his conclusions are cautious, Bergman suggests that certain efforts to mitigate extinction risks might be ill-advised. A must-read for those willing to engage in a profound philosophical reflection.
********************************
* Summary (AI-generated) *
********************************
- Aaron Bergman bases his discussion on hedonic utilitarianism, which sees pleasure as the only moral good and suffering as the sole evil. He speculates that this core idea could be extrapolated to other consequentialist ethics that focus on wellbeing.
- Bergman asserts that there are levels of suffering so extreme that no amount of happiness can offset them. Many resonate with the sentiment that they would decline any offer of boundless joy if it meant enduring extreme torture. This sentiment echoes the potential choices of an idealized, rational hedonistic individual. If such a being would decline this proposition, it might hint at a profound moral truth.
- Bergman contends that no persuasive arguments exist to suggest that any limited harm can be balanced out by a sufficient benefit. He particularly criticizes the notion that risks like a 1 in 100 million chance of death can be compared to the risk of enduring extreme torture.
- He proposes that the moral worth of hedonic states might not be expressed in finite numbers. Hence, mathematical proofs that justify minor risks of disasters may not hold when considering risks of extreme suffering.
- This discourse on unjustifiable suffering doesn't directly support or oppose the idea of human extinction. But it raises concerns about long-term visions that involve the spread of intelligent life, potentially leading to immense suffering. However, due to various reasons, most current efforts to reduce extinction risks might also decrease s-risks stemming from AI or biological catastrophes. As a result, Bergman's arguments might not necessitate significant shifts in current focus.
- Yet, he suggests that some side efforts, such as minimizing supervolcano risks, might be inadvisable if they raise long-term s-risks. Bergman underscores the significance of pinpointing strategies that diminish s-risks more effectively than conventional x-risk approaches.
#6 Daniel Filan on why I'm wrong about ethics (+ Oppenheimer and what names mean in like a hardcore phil of language sense)
Note: the core discussion on ethics begins at 7:58 and moves into philosophy of language at ~1:12:19
Blurb and bulleted summary from Clong:
This wide-ranging conversation between Daniel and Aaron touches on movies, business drama, philosophy of language, ethics and legal theory. The two debate major ethical concepts like utilitarianism and moral realism. Thought experiments around rational beings choosing to undergo suffering feature prominently. meandering tangents explore the semantics of names and references.
- Aaron asserts that total utilitarianism does not imply that any amount of suffering can be morally justified by creating more happiness. His argument is that the affirmative case for this offsetting ability has not been clearly made.
- He proposes a thought experiment - if offered to experience the suffering of all factory farmed animals in exchange for unlimited happiness, even a perfectly rational being would refuse. This indicates there are some levels of suffering not offsettable.
- Aaron links this to experiences like hunger where you realize suffering can be worse than you appreciate normally. This causes his intuition some suffering can't be outweighed.
- Daniel disagrees, believing with the right probabilities and magnitudes of suffering versus happiness, rational beings would take that gamble.
- For example, Daniel thinks the atomic bombing of Japan could be offset by reducing more suffering. Aaron is less sure given the pain inflicted.
- Daniel also proposes offsets for animal farming, but Aaron doesn't think factory farming harm is offsettable by any amount of enjoyment of meat.
- They discuss definitions of rationality and whether evolution pressures against suicide impact the rationality of not killing oneself.
- Aaron ties his argument to siding with what a perfectly rational being would choose to experience, not necessarily what they would prefer.
- They debate whether hypothetical aliens pursuing "schmorality" could point to a concept truly analogous to human morality. Aaron believes not.
Transcript at: https://www.aaronbergman.net/p/6-daniel-filan-on-why-im-wrong-about
Daniel’s stuff:
AI X-risk podcast: https://axrp.net/
The Filan Cabined podcast: https://thefilancabinet.com/
Personal website and blog: https://danielfilan.com/
#5: Nathan Barnard (again!) on why general intelligence is basically fake
Very imperfect transcript: bit.ly/3QhFgEJ
Summary from Clong:
- The discussion centers around the concept of a unitary general intelligence or cognitive ability. Whether this exists as a real and distinct thing.
- Nathan argues against it, citing evidence from cognitive science about highly specialized and localized brain functions that can be damaged independently. Losing linguistic ability does not harm spatial reasoning ability.
- He also cites evidence from AI, like systems excelling at specific tasks without general competency, and tasks easy for AI but hard for humans. This suggests human cognition isn’t defined by some unitary general ability.
- Aaron is more open to the idea, appealing to an intuitive sense of a qualitative difference between human and animal cognition - using symbolic reasoning in new domains. But he acknowledges the concept is fuzzy.
- They discuss whether language necessitates this general ability in humans, or is just associated. Nathan leans toward specialized language modules in the brain.
- They debate whether strong future AI systems could learn complex motor skills just from textual descriptions, without analogous motor control data. Nathan is highly skeptical.
- Aaron makes an analogy to the universe arising from simple physical laws. Nathan finds this irrelevant to the debate.
- Overall, Nathan seems to push Aaron towards a more skeptical view of a unitary general cognitive ability as a scientifically coherent concept. But Aaron retains some sympathy for related intuitions about human vs animal cognition.
#4 Winston Oswald-Drummond on the tractability of reducing s-risk, ethics, and more
Summary (by Claude.ai)
This informal podcast covers a wide-ranging conversation between two speakers aligned in the effective altruism (EA) community. They have a similar background coming to EA from interests in philosophy, rationality, and reducing suffering. The main topic explored is reducing s-risks, or risks of extreme suffering in the future.
Winston works for the Center for Reducing Suffering (CRS), focused on spreading concern for suffering, prioritizing interventions, and specifically reducing s-risks. He outlines CRS's focus on research and writing to build a moral philosophy foundation for reducing suffering. Aaron is skeptical s-risk reduction is tractable currently, seeing the research as abstract without a clear theory of change.
They discuss how CRS and a similar group CLR are trying to influence AI alignment and digital sentience to reduce potential future s-risks. But Aaron worries about identifying and affecting the "digital neural correlates of suffering." Winston responds these efforts aim to have a positive impact even if unlikely to succeed, and there are potential lock-in scenarios that could be influenced.
Aaron explains his hesitancy to donate based on tractability concerns. He outlines his EA independent research, which includes an archive project around nuclear war. More broadly, the two find they largely ethically agree, including on a suffering-focused ethics and "lexical negative utilitarianism within total utilitarianism.
Some disagreements arise around the nature of consciousness, with Aaron arguing rejecting qualia implies nihilism while Winston disagrees. They also diverge on moral realism, with Aaron defending it and Winston leaning anti-realist.
As they wrap up the wide-ranging conversation, they joke about convincing each other and make predictions on podcast listens. They thank each other for the thought-provoking discussion, aligned in ethics but with some disagreements on consciousness and metaethics. The conversation provides an insider perspective on efforts to reduce s-risks through research and outreach.
EA Archive: https://forum.effectivealtruism.org/posts/DndmvDGStD3gTfhXk
#3: Nathan Barnard on how financial regulation can inform AI regulation
Summary/specific topics:
- Stress Tests and AI Regulation: Nathan elaborates on the concept of stress tests conducted by central banks. These tests assess the resilience of banks to severe economic downturns and the potential for a domino effect if one bank fails. They believe that lessons from this process can be applied to AI regulation. Aaron agrees, but also highlights the need for a proactive approach to AI regulation, as opposed to the reactive measures often seen in banking regulation.
- The Role of Central Banks in AI Regulation: Nathan suggests that institutions structured like central banks, staffed with technical experts and independent from government, could be beneficial for AI regulation. They believe such institutions could respond quickly and effectively to crises. However, they acknowledge that this approach may not be effective if AI development leads to rapid, uncontrollable self-improvement.
- Compute Governance: The conversation then shifts to compute governance, which Nathan sees as a promising area for AI regulation due to the obviousness of someone using large amounts of compute. They believe that this could provide governments with a control lever over cutting-edge AI labs, similar to how central banks control banking loans and affairs.
- AI Regulation and the Role of Public Actors: Nathan acknowledges that the leaders of major AI labs seem sensible and aligned with AI safety principles. However, they argue that regulation and public actors can play a crucial role in creating common knowledge between labs and preventing a race to the bottom. They also discuss the potential benefits and drawbacks of different regulatory approaches.
- Financial Regulation as a Model for AI Regulation: Nathan believes that post-crisis financial regulation, such as the Dodd-Frank Act, has generally been effective. They suggest that AI regulation could follow a similar path, especially if AI becomes a significant part of the economy. However, Aaron expresses skepticism about the ability of political processes to produce effective AI regulation.
- Regulation Before and After Crises: The speakers agree that pre-crisis regulation has generally been less effective than post-crisis regulation. They discuss the potential for AI regulation to follow a similar pattern, with effective regulation emerging in response to a crisis.
- Regulatory Arbitrage: The conversation concludes with a discussion on regulatory arbitrage, where banks shift activities to where it's cheapest to do business. Despite evidence of this behavior, Nathan notes that there was no race to the bottom in terms of regulation during the financial crisis.
[Bot readout] NYT: Inside the White-Hot Center of A.I. Doomerism: Anthropic, a safety-focused A.I. start-up, is trying to compete with ChatGPT while preventing an A.I. apocalypse
Not a legit pod episode, just a bot readout of the NYT's recent article (copy; original)
Inside the White-Hot Center of A.I. Doomerism: Anthropic, a safety-focused A.I. start-up, is trying to compete with ChatGPT while preventing an A.I. apocalypse. It’s been a little stressful.
for my fellow ppl who can only listen to things
#2: Arjun Panickssery solves books, hobbies, and blogging, but fails to solve the Sleeping Beauty problem because he's wrong on that one
- Follow Arjun on Twitter: https://twitter.com/panickssery
- Read and subscribe to his blog: https://arjunpanickssery.substack.com
- A mediocre transcription can be found at https://www.assemblyai.com/playground/transcript/6x5h1mcemt-bff1-40fc-a676-9b59c66985f0
🦆 👍
#1 Laura Duffy solves housing, ethics, and more
A transcript can be found at assemblyai.com/playground/transcript/6y7e7wz28c-30aa-4e83-ba4f-1bddf2e23dad