Consistently Candid

#21 Steven Adler on scruntising AI company safety practices & pacing the frontier

After a long hiatus, I am resurrecting my podcast! For this episode, I spoke with Steven Adler. Steven worked on policy and safety at OpenAI between 2020 and 2024. He has since left to pursue independent writing to raise public awareness about AI risks and co-founded Guidelight AI, an organisation focused on scrutinising and improving safety practices at major AI companies. Guidelight was one of two organisations instrumental in spearheading the Pacing the Frontier open letter, which has garnered 1300+ signatories from employees at frontier labs. It has also published a scorecard of AI control practices across OpenAI, Anthropic, Meta, Google, and xAI. Topics covered in the episode: Why Steven left OpenAI in 2024 — o1, NDAs, safety staff turnover Counterarguments to AI pessimism, and where Steven's cruxes lie The OpenAI–Hugging Face incident and OpenAI's response AI control: monitoring, prevention, and whether it scales to superintelligence Why incident reporting is inadequate, and communicating risk when nothing visibly bad has happened Where AI policy stands: SB 53, RAISE, Illinois, the EU AI Act, and how weak enforcement is Companies quietly diluting their own safety commitments The Pacing the Frontier letter, what "pacing" means and how it's landing Guidelight's control scorecard: method, findings, and theory of change Links: Follow Steven on Twitter and subscribe to his Substack The Pacing the Frontier open letter Read Guidelight's AI control scorecard

• Transcript
• Chapters

#20 Frances Lorenz on the emotional side of AI x-risk, being a woman in a male-dominated online space & more

In this episode, I chatted with Frances Lorenz, events associate at the Centre for Effective Altruism. We covered our respective paths into AI safety, the emotional impact of learning about x-risk, what it's like to be female in a male-dominated community and more! Follow Frances on Twitter Subscribe to her Substack Apply for EAG London!

#19 Gabe Alfour on why AI alignment is hard, what it would mean to solve it & what ordinary people can do about existential risk

Gabe Alfour is a co-founder of Conjecture and an advisor to Control AI, both organisations working to reduce risks from advanced AI. We discussed why AI poses an existential risk to humanity, what makes this problem very hard to solve, why Gabe believes we need to prevent the development of superintelligence for at least the next two decades, and more. Follow Gabe on Twitter Read The Compendium and A Narrow Path

#18 Nathan Labenz on reinforcement learning, reasoning models, emergent misalignment & more

A lot has happened in AI since the last time I spoke to Nathan Labenz of The Cognitive Revolution, so I invited him back on for a whistlestop tour of the most important developments we've seen over the last year! We covered reasoning models, DeepSeek, the many spooky alignment failures we've observed in the last few months & much more! Follow Nathan on Twitter Listen to The Cognitive Revolution My Twitter & Substack

#17 Fun Theory with Noah Topper

The Fun Theory Sequence is one of Eliezer Yudkowsky's cheerier works, and considers questions such as 'how much fun is there in the universe?', 'are we having fun yet' and 'could we be having more fun?'. It tries to answer some of the philosophical quandries we might encounter when envisioning a post-AGI utopia. In this episode, I discussed Fun Theory with Noah Topper, who loyal listeners will remember from episode 7, in which we tackled EY's equally interesting but less fun essay, A List of Lethalities. Follow Noah on Twitter and check out his Substack!

#16 John Sherman on the psychological experience of learning about x-risk and AI safety messaging strategies

John Sherman is the host of the For Humanity Podcast, which (much like this one!) aims to explain AI safety to a non-expert audience. In this episode, we compared our experiences of encountering AI safety arguments for the first time and the psychological experience of being aware of x-risk, as well as what messaging strategies the AI safety community should be using to engage more people. Listen & subscribe to the For Humanity Podcast on YouTube and follow John on Twitter!

#14 Buck Shlegeris on AI control

Buck Shlegeris is the CEO of Redwood Research, a non-profit working to reduce risks from powerful AI. We discussed Redwood's research into AI control, why we shouldn't feel confident that witnessing an AI escape attempt would persuade labs to undeploy dangerous models, lessons from the vetoing of SB1047, the importance of lab security and more. Posts discussed: The case for ensuring that powerful AIs are controlled Would catching your AIs trying to escape convince AI developers to slow down or undeploy? You can, in fact, bamboozle an unaligned AI into sparing your life Follow Buck on Twitter and subscribe to his Substack!

#13 Aaron Bergman and Max Alexander debate the Very Repugnant Conclusion

In this episode, Aaron Bergman and Max Alexander are back to battle it out for the philosophy crown, while I (attempt to) moderate. They discuss the Very Repugnant Conclusion, which, in the words of Claude, "posits that a world with a vast population living lives barely worth living could be considered ethically inferior to a world with an even larger population, where most people have extremely high quality lives, but a significant minority endure extreme suffering." Listen to the end to hear my uninformed opinion on who's right. Read Aaron's blog post on suffering-focused utilitarianism Follow Aaron on Twitter Follow Max on Twitter My Twitter

#12 Deger Turan on all things forecasting

Deger Turan is the CEO of forecasting platform Metaculus and president of the AI Objectives Institute. In this episode, we discuss how forecasting can be used to help humanity coordinate around reducing existential risks, Deger's advice for aspiring forecasters, the future of using AI for forecasting and more! Enter Metaculus's Q3 AI Forecasting Benchmark Tournament Get in touch with Deger: deger@metaculus.com

#11 Katja Grace on the AI Impacts survey, the case for slowing down AI & arguments for and against x-risk

Katja Grace is the co-founder of AI Impacts, a non-profit focused on answering key questions about the future trajectory of AI development, which is best known for conducting the world's largest survey of machine learning researchers. We talked about the most interesting results from the survey, Katja's views on whether we should slow down AI progress, the best arguments for and against existential risk from AI, parsing the online AI safety debate and more! Follow Katja on Twitter Katja's Substack My Twitter

#10 Nathan Labenz on the current AI state-of-the-art, the Red Team in Public project, reasons for hope on AI x-risk & more

Nathan Labenz is the founder of AI content-generation platform Waymark and host of The Cognitive Revolution Podcast, who now works full-time on tracking and analysing developments in AI. We chatted about where we currently stand with state-of-art AI capabilities, whether we should be advocating for a pause on scaling frontier models, Nathan's Red Team in Public project, and some reasons not be a hardcore doomer! Follow Nathan on Twitter Listen to The Cognitive Revolution

#9 Sneha Revanur on founding Encode Justice, California's SB-1047, and youth advocacy for safe AI development

Sheha Revanur is a the founder of Encode Justice, an international, youth-led network campaigning for the responsible development of AI, which was among the sponsors of California's proposed AI bill SB-1047. We chatted about why Sheha founded Encode Justice, the importance of youth advocacy in AI safety, and what the movement can learn from climate activism. We also dug into the details of SB-1047 and answered some common criticisms of the bill! Follow Sneha on Twitter: https://twitter.com/SnehaRevanur Learn more about Encode Justice: https://encodejustice.org/

#8 Nathan Young on forecasting, AI risk & regulation, and how not to lose your mind on Twitter

Nathan Young is a forecaster, software developer and tentative AI optimist. In this episode, we discussed how Nathan approaches forecasting, why his p(doom) is 2-9%, whether we should pause AGI research, and more! Follow Nathan on Twitter: Nathan 🔍 (@NathanpmYoung) / X (twitter.com) Nathan's substack: Predictive Text | Nathan Young | Substack My Twitter: sarah ⏸️ (@littIeramblings) / X (twitter.com)

#7 Noah Topper helps me understand Eliezer Yudkowsky

A while back, my self-confessed inability to fully comprehend the writings of Eliezer Yudkowsky elicited the sympathy of the author himself. In an attempt to more completely understand why AI is going to kill us all, I enlisted the help of Noah Topper, recent Computer Science Masters graduate and long-time EY fan, to help me break down A List of Lethalities (which, for anyone unfamiliar, is a fun list of 43 reasons why we're all totally screwed). Follow Noah on Twitter: Noah Topper 🔍⏸️ (@NoahTopper) / X (twitter.com) My Twitter: sarah ⏸️ (@littIeramblings) / X (twitter.com)

#6 Holly Elmore on pausing AI, protesting, warning shots & more

Holly Elmore is an AI pause advocate and Executive Director of PauseAI US. We chatted about the case for pausing AI, her experience of organising protests against frontier AGI research, the danger of relying on warning shots, the prospect of techno-utopia, possible risks of pausing and more! Follow Holly on Twitter: Holly ⏸️ Elmore (@ilex_ulmus) / X (twitter.com) Official PauseAI US Twitter account: PauseAI US ⏸️ (@pauseaius) / X (twitter.com) My Twitter: sarah ⏸️ (@littIeramblings) / X (twitter.com) Learn more about PauseAI: We need to Pause AI

#5 Joep Meindertsma on founding PauseAI and strategies for communicating AI risk

In this episode, I talked with Joep Meindertsma, founder of PauseAI, about how he discovered AI safety, the emotional experience of internalising existential risks, strategies for communicating AI risk, his assessment of recent AI policy developments and more! Find out more about PauseAI at www.pauseai.info

#4 Émile P. Torres and I discuss where we agree and disagree on AI safety

Émile P. Torres is a philosopher and historian known for their research on the history and ethical implications of human extinction. They are also an outspoken critic of Effective Altruism, longtermism and the AI safety movement. In this episode, we chatted about why Émile opposes both the 'doomer' and accelerationist factions, and identified some or our agreements and disagreements about AI safety.

#3 Darren McKee on explaining AI risk to the public & navigating the AI safety debate

Darren McKee is an author, speaker and policy advisor who has recently penned a beginner-friendly introduction to AI Safety named Uncontrollable: The Threat of Artificial Superintelligence and the Race to Save the World. We chatted about the best arguments for worrying about AI, responses to common objections, how to navigate the online AI safety space as an non-expert, and more. Buy Darren's book on Amazon: https://www.amazon.co.uk/Uncontrollable-Threat-Artificial-Superintelligence-World-ebook/dp/B0CNYMF89Z Follow Darren on Twitter: https://twitter.com/dbcmckee?lang=en My Twitter: https://twitter.com/littIeramblings

#1 Aaron Bergman and Max Alexander argue about moral realism while I smile and nod

In this inaugural episode of Consistently Candid, Aaron Bergman and Max Alexander each try to convince me of their position on moral realism, and I settle the issue once and for all. Featuring occasional interjections from the sat-nav in the Uber Aaron was taking at the time.My Twitter: https://twitter.com/littIeramblings Max's Twitter: https://twitter.com/absurdlymaxAaron's Twitter: https://twitter.com/AaronBergman18