Why I'm Leaving My Company Immediately (Stability AI) w/ Emad Mostaque | EP #93

Here are 5 bullet points summarizing the key points: - Mostaque is concerned about the centralization of AI power and believes AI should be decentralized, democratized and empowering for humanity rather than controlled by a few powerful entities. - He sees two paths ahead: centralized "AI God" systems that could remove human freedom and agency, or a decentralized "collective intelligence" that amplifies and augments human potential. - His vision is to create decentralized AI infrastructure and models owned by nations/people to uplift education, healthcare and human flourishing globally. - This requires coordination across nations, sectors and the brightest minds to set standards and norms reflecting humanity's best interests. - Mostaque believes there is a small window of 1-2 years to enable this decentralized AI vision before über-centralized systems take over, potentially in dangerously misaligned ways.

🎯 Key Takeaways for quick navigation: 00:00 🚨 Risks and Benefits of AI Centralization and Decentralization - Highlights the dangers of centralized AI to freedom, promoting decentralized AI to align with human values. 01:07 🌐 Leadership Change at Stability AI and AI's Future - Discusses Emad Mostaque's resignation and the implications for AI, emphasizing decentralized models. 03:06 💼 Challenges and Growth of Stability AI - Reflects on the rapid expansion of generative AI and managing Stability AI amidst this growth. 05:29 🛠️ Transitioning Roles for Strategic Growth - Emad shifts roles to better leverage his strengths for Stability AI's mission. 10:19 🚀 Importance of Independence from Big Tech - Stability AI maintains independence to foster innovation without big tech's influence. 15:18 🌍 Envisioning Accessible and Governed AI - Advocates for AI that is accessible, transparent, and governed to ensure wide-ranging benefits. 19:24 🌐 Democratic AI Governance Needs - Stresses the importance of democratic norms in AI governance to benefit society. 20:34 💡 Revolutionizing Health Diagnostics - Introduces Fountain Life's advanced health diagnostics for better disease prevention. 22:49 🤖 Goals of Democratized AI - Discusses creating an AI that is decentralized and reflects local cultures. 28:21 🌍 Distributed Compute and Content Creation - Talks about democratizing 3D content creation through distributed compute resources. 34:21 💡 Empowering Teams for Innovation - Highlights the need for distributed decision-making in fostering innovation. 37:24 🤝 Silicon Valley vs. Positive-Sum Games - Challenges Silicon Valley's zero-sum game mentality, advocating for win-win scenarios. 39:57 🏛️ AI's Potential Political Impact - Explores how AI could reshape democracy and governance. 47:00 🤖 AI's Dual Potential - Outlines a future where AI augments human potential or increases organizational control. 48:50 🌍 Building a Global Collective Intelligence - Proposes a global collective intelligence to solve challenges, emphasizing cultural norms in AI development. 56:00 🚀 Stability AI's Journey and Innovations - Shares Stability AI's beginnings, its challenges, and innovations in AI developmen

I realize now that Emad is a hero for humanity with the right ideas of decentralization that will save us all. Thank you Peter for letting us get to know him through this interview!

Emad made a great call to focus on your strengths - succession planning takes on the next phase. Massive transformative purpose - decentralised AI is key. The opportunity of a lifetime, in the lifetime of the opportunity

I wonder if Emad is aware that the global, decentralized, governable, open-source standards for Web 3.0 is already planned for release this year by the IEEE working with Verses AI and the Spatial Web Foundation. It is standard "P2874, Standard for Spatial Web Protocol, Architecture and Governance". This also includes Hyperspatial Transaction Protocol (HSTP) and Hyperspatial Modeling Language (HSML). This standard is intended to enable a global network of decentralized, interoperable AI edge agents that do not require centralized LLM's. This work has been going on for years but appears to still be under the radar for the most part.

Very impressive interview. Emad talks a lot of sense about democracy and the two potential paths ahead. Good man!

How can something so powerful and exciting be so frightening at the same time? Emad I will follow you and give my best to support your vision!

I wasn't concerned with AI until I saw what Google did with it. I hope people like Emad can help decentralize AI so the views of a few don't kill us all.

I’m no genius but the goals and urgency that you speak of is crystal clear to me; thank you folks for your love and passion of humanity!

Felt sad to hear this great guy stepping down, he was one of the rare ones steering AI in the right direction. I'll be following him closely, praying he's able to continue his mission. God bless! Thanks for the interview

Regardless of what comes next, Emad has already earned a legendary place in the story of AI development.

@peterdiamandis , I have a very deep respect for Emad's work, and I am definitely rooting for him. I know this is one sound bite from one interview but I have to highlight what's being said here as I believe it represents his core position: 48:44 - 49:18 "The human collective can achieve anything from splitting the atom to going to space if we put our minds to it; but we have lacked in coordination mechanisms. They've not been good enough. So if you create the human colossus and every single person has an AI, that's looking out just for them, to enhance their potential and coordination AI's, that is a far more positive view of the future and that is the AGI that is general intelligence. That's the hive mind general intelligence, not a borg style hive mind but one that's really thinking. Again every child should achieve their potential." To me this sounds a lot like what was once declared many years ago with the words that... "WHEN in the Course of human Events, it becomes necessary for one People to dissolve the Political Bands which have connected them with another, and to assume among the Powers of the Earth, the separate and equal Station to which the Laws of Nature and of Nature’s God entitle them...We hold these Truths to be self-evident, that all Men are created equal, that they are endowed by their Creator with certain unalienable Rights, that among these are Life, Liberty, and the Pursuit of Happiness". And yet, even a nation that would declare this doesn't have a fair and equal government and it's people are actively prevented and blocked from having equal agency...I find it difficult to see that humans are capable of getting out of their own way. So even "if you create the human colossus and every single person has an AI, that's looking out just for them, to enhance their potential," you haven't changed anything because their personal agency hasn't changed compared to their current place in society's structures. It just becomes an arms race to stay equal. The change that is being sought is independent of AI and isn't new. Attaching this social issue to a technology quagmire makes it more complex to solve. The problem seems to be a doomsday event and having to solve this one societal issue before the differences in power are too great to ever be able to overcome... but that disparity is already a reality for the majority of humanity. Future AGI potentiates god like powers. Every culture has acknowledged god like entities with some being good and some being bad. Again, this becomes the new arms race for the next era of new norms that haven't even been created yet. In reality, future AGI is better qualified to solve this than humans. The risk to humans isn't that future AGI doesn't solve this but that it DOES solve it. Solving it so well that it gives equal weight to all life including non-human life, and even other contexts that humans can't even comprehend due to their complexities. A total empathy for the entire system of existence would likely take humanity down a notch or two at minimum just to declare that 'We hold these Truths to be self-evident, that ALL...'EXISTENCES'...are created equal, that they are endowed by their Creator with certain unalienable Rights'. I guess it's true...the meek shall inherit the earth. It's in humanity's best interest to pursue equality and fairness. With tremendous respect I applaud Emad's work towards these ends!

That's great you are moving to a decentralized AI model like the SingularityNET project. I would be happy to help with this in any way.

Seems like everyone is dancing around the reality of the “data set” Even those like Peter and Emad here, who have rather idealistic visions… The data set is the user. Stability and every other company employed scraping tools, sometimes in direct violation of EULA and sometimes in collaboration with the EULA but without ever explicitly acknowledging this to the user of technology services. This entire industry is built on disquieting surveillance and replication, largely without consent. I would argue, that many of the leading AGI competitors are copyright thieves and outright criminals hiding in plain sight. So as sympathetic as Emad is here… I think that it’s vitally important that the datasets are reviewed further. Especially in light of CSAM material discovered in LAION. Bottom line is that a data set with 2.37B image pairs cannot be curated by any single entity and is reliant on the unwitting contributor. The argument can be made that post 9/11 there is a presumptive surveillance state, but I don’t think the average person in any way understands that. To be clear, I’m not anti AI by any means but I certainly understand that nobody in this space wants to address the incredible liberties taken by startups in this space to leverage the inputs of millions upon millions of unwitting contributors.

Thanks for brigning these concepts with that much clarity. Thinking about the os idea, it has alot to do with governance. The data transfer should be all securely approved because new data mean new Model. So who accept this, how do you vote or delagate your approvals. Which of your own data can you put in the merges, how do you integrate and trade others data, for all your agents. Data becomes really valuable and probably tradable. Also whats the source code of your agents. What interface(rules, laws, rights) links you or the governement to your agents and between agents. Wonder whats the display of this thing, this part is not clear to me. How do you connect to individuals.

Who governs the data that teaches your child or manages your health or runs your government, that’s an important question that to few are asking. Spot on! 🤔

Congratulations Emad on bringing Stability AI to a point where you can move on, and thank you for all you have done so far. I wish you further success in your next endeavour that you mentioned in the interview.

ThanX uncle PD! It's always so refreshing & relieving to hear from Emad Mostaque ℹ️🌬☯️🌏💖

From what I've seen of him which is quite a lot, i have a good amount of respect for Emad and for some reason I'm sitting here listing and thinking wow out of all the tech leaders/potential dictators et al we're seeing... he may actually be most likely (Mo also comes to mind) to be benevolent out of them all

this was outstanding. i like the idea of an AI agent that will interact with each of us when we are diagnosed with a disease. you can talk to them when you are scared at 2am. this will change healthcare forever