Reid and AIs riff on what could go right with AI

This week, we're bringing you the introduction of Reid’s new book, Superagency: What Could Possibly Go Right with Our AI Future, which he co-wrote with last week’s Possible guest, Greg Beato. The Superagency audiobook is narrated by voice actor Scott Wallace and converted with Respeecher to Reid’s AI voice. Plus, hear what happens when Reid and Aria load Superagency’s introduction into Google’s NotebookLM and turn the final minutes of the episode over to two AI co-hosts. For more info on the podcast and transcripts of all the episodes, visit https://www.possible.fm/podcast/ For more on Superagency, check out superagency.ai.

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:04Aria Finger?

  Hey there, possible listeners. So last week, as you know, I interviewed Greg Beato and Reid about their new book, Superagency. If you enjoyed that episode and want more, you are in the right place. You can listen to the audiobook introduction of Superagency right after this. And fun fact, so while the voice you are about to hear sounds like Reid, it's actually not the real Reid. We got a recording from a professional voice actor, which with the help of some cool voice generation technology from Respeacher, we converted into a clone of Reid's voice. And honestly, I work with Reid every day, and so I think I know him pretty well, but sometimes I can't even tell the difference. So let's see if you can.

• 0:45Aria Finger?

  And then if you like what you hear, please go to superagency.ai and you can buy the book, the audiobook, et cetera. And Reid's actually also giving away custom Superagency book covers for free for a limited time. So check out superagency.ai for all of that information. And then one last note, after the audiobook introduction that you're about to hear, we did a little AI experiment. We uploaded the Superagency intro to Google's AI tool Notebook LM, and in a matter of seconds, it generated a short podcast to discuss the content. So you'll hear two AI podcast hosts discussing what you just listened to. Hopefully, they're not better than Reid and I, but you be the judge. So sorry for the long preamble.

• 1:30Aria Finger?

  Without further ado, here is the intro to Superagency: What Could Possibly Go Right With Our AI Future by Reid Hoffman and Greg Beato.

• 1:41Reid Hoffman?

  Introduction. Throughout history, new technologies have regularly sparked visions of impending dehumanization and societal collapse. The printing press, the power loom, the telephone, the camera, and the automobile all face significant skepticism and sometimes even violent opposition on their way to becoming mainstays of modern living.

• 2:08Reid Hoffman?

  Fifteenth-century doom-mongers argued that the printing press would dramatically destabilize society by enabling heresy and misinformation, and by undermining the authority of the clergy and scholars. The telephone was characterized as a device that could displace the intimacy of in-person visits, and also make friends too transparent to one another. In the early decades of the cars' ascent, critics claimed it was destroying family life, with unmarried men choosing to save up for Model Ts instead of getting married and having kids, and married men resorting to divorce to escape the pressures of consumption that cars helped create.

• 2:51Reid Hoffman?

  The same kind of doom and gloom was applied to society-wide automation in the 1950s, when increasingly sophisticated machines were dramatically impacting factories and office buildings alike, with everyone from bakers, meat cutters, auto workers, and US Census Bureau statisticians seeing their overall numbers dwindle. In 1961, Time Magazine reported that labor experts believed that without intervention from business interests, unions, and the government, automation would continue to grow the permanently unemployed. By the mid-1960s, congressional subcommittees were regularly holding hearings regarding

• 3:36Reid Hoffman?

  the mainframe computer's potential threat to privacy, free will, and the average citizen's capacity to make a life of their own choosing. Today, US unemployment rates are lower than they were in 1961. The average US citizen lives in a world where PCs, the internet, and smartphones have ushered in a new age of individualism and self-determination, rather than crushing authoritarian compliance or the end of humanity. But with the emergence and ongoing evolution of highly capable AIs, it's not just that familiar fears about technology persist, they're growing.

• 4:19Reid Hoffman?

  Even among AI developers, some believe that future instances of superintelligent AIs could represent an extinction-level threat to humanity. Others point out that, at the very least, humans acting with malicious intent will be able to use AIs to create catastrophic damage well before the machines themselves wage unilateral war against humanity. Additional concerns include massive job displacement, total human obsolescence, and a world where a tiny cabal of techno-elites capture whatever benefits, if any, AI enables. "The doomsday warnings are different this time," these observers insist, "because the

• 5:04Reid Hoffman?

  technology itself is different this time." AI can already simulate core aspects of human intelligence. Many researchers believe it will soon attain the capacity to act with complete and extremely capable autonomy in ways that aren't aligned with human values or intentions.

• 5:24Reid Hoffman?

  Robots and other kinds of highly intelligent systems have long existed in sci-fi novels, comic books, and movies as our dark doppelgangers and adversaries. So as today's state-of-the-art AIs hold forth, like benevolent but coolly rational grad students, it's only natural to see foreshadowing of how, from 2001: A Space Odyssey, or in a less self-aware and more overtly menacing form, The Terminator's relentless killer robot. These narratives have shaped our worst visions of the future for a long, long time. But are they the right narratives? The future is notoriously hard to foresee accurately for pessimists and optimists alike.

• 6:09Reid Hoffman?

  We didn't get the permanent mass unemployment that labor experts in the early 1960s anticipated, nor did we get the Jetsons and its flying cars, at least not yet.

• 6:21Reid Hoffman?

  As hard as it may be to accurately predict the future, it's even harder to stop it. The world keeps changing. Simply trying to stop history by entrenching the status quo through prohibitions, pauses, and other efforts to micromanage who gets to do what is not going to help us humans meet either the challenges or the opportunities that AI presents. That's because as much as collaboration defines us, competition does too. We form groups of all kinds at all levels to amplify our efforts, often deploying our collective power against other teams, other companies, other countries.

• 7:01Reid Hoffman?

  Even within our own groups of like-minded allies, competition emerges because of variations in values and goals. And each group and subgroup is generally adept at rationalizing self-interest in the name of the greater good. Coordinating at a group level to ban, constrain, or even just contain a new technology is hard. Doing so at a state or national level is even harder. Coordinating globally is like herding cats. If cats were armed, tribal, and had different languages, different gods, and different dreams for the future that went beyond their next meal. Meanwhile, the more powerful the technology, the harder the coordination problem, and

• 7:43Reid Hoffman?

  that means you'll never get the future you want simply by prohibiting the future you don't want. Refusing to actively shape the future never works, and that's especially true now that the other side of the world is only just a few clicks away. Other actors have other futures in mind. What should we do? Fundamentally, the surest way to prevent a bad future is to steer toward a better one that by its existence makes significantly worse outcomes harder to achieve. At this point, we know from thousands of years of experience that if a technology can be created, humans will create it.

• 8:24Reid Hoffman?

  As I've written elsewhere, including in my previous book, Impromptu, we're homo techne, at least as much as we are homo sapiens. We continuously create new tools to amplify our capabilities and shape the world to our liking. In turn, these tools end up shaping us as well. What this suggests is that humanism and technology, so often presented as oppositional forces, are in fact integrative ones. Every new technology we've invented, from language to books to the mobile phone, has defined, redefined, deepened, and expanded what it means to be human.

• 9:05Reid Hoffman?

  We're the initiators of this process, but we can't fully control it. Once set in motion, new technologies exert a gravity of their own. A world where steam power exists works differently than the world that preceded it. This is precisely why prohibition or constraint alone is never enough. They offer stasis and resistance at the very moment we should be pushing forward in pursuit of the brightest possible future.

• 9:32Reid Hoffman?

  Some might describe this as technological determinism, but we think of it as navigating with a kind of techno-humanist compass. A compass helps us to choose a course of action, but unlike a blueprint or some immutable manifesto, it's dynamic rather than determinative. It helps us orient, reorient, and find our way.

• 9:55Reid Hoffman?

  It's also crucial that this compass be explicitly humanist, because ultimately every major technological innovation impacts human agency, our ability to make choices and exert influences on our lives. A techno-humanist compass actively aims to point us toward paths in which the technologies we create broadly augment and amplify individual and collective agency.

• 10:21Reid Hoffman?

  With AI, this orientation is especially important. Because what happens to human agency when these systems and devices, often described as agents themselves, do become capable of replacing us entirely? Shouldn't we slow down that eventuality as much as possible? A techno-humanist perspective sees it the other way around. Our sense of urgency needs to match the current speed of change. We can only succeed in prioritizing human agency by actively participating in how these technologies are defined and developed. First and foremost, that means pursuing a future where billions of people around the world get equitable, hands-on access

• 11:04Reid Hoffman?

  to experiment with these technologies themselves in ways of their own choosing. It also means pursuing a future where the growing capabilities of AI help us reduce the threats of nuclear war, climate change, pandemics, resource depletion, and more. In addition, it means pursuing this future even though we know we won't be able to predict or control every development or consequence that awaits us. No one can presume to know the exact final destination of the journey we're on or the specific contours of the terrain that exists there. The future isn't something that experts and regulators can meticulously design. It's something that society explores and discovers collectively.

• 11:48Reid Hoffman?

  That's why it makes the most sense to learn as we go and to use our techno-humanist compass to course correct along the way. In a nutshell, that's iterative deployment, the term that OpenAI, ChatGPT's developer, uses to describe its own method in bringing its products into the world. It's a concept my co-author Greg Beato and I will explore and emphasize throughout this book. As a longtime founder and investor in technology companies, my perspective is inevitably shaped by the technology-driven progress and positive outcomes I participated in over the course of my career. I was a founding board member at PayPal

• 12:29Reid Hoffman?

  and part of its executive team when eBay purchased it in 2002. I co-founded LinkedIn and have sat on Microsoft's board since 2017 following its purchase of LinkedIn.

• 12:42Reid Hoffman?

  I was also one of the first philanthropic supporters of OpenAI when it launched as a nonprofit research lab in 2015. I led the first round of investment in 2019 when OpenAI established a for-profit limited partnership in order to support its ongoing development efforts. I served on its board from 2019 to early 2023. Along with Mustafa Suleyman, who co-founded DeepMind, I co-founded a public benefit corporation called Inflection AI in 2022 that has developed its own conversational agent, Pi. In my role at the venture capital firm Greylock, I've invested in other AI companies. On my podcast, Possible, I regularly

• 13:27Reid Hoffman?

  talk with a wide range of innovators about the impacts AI will have on their fields, with a techno-humanist compass guiding our conversations. I also provide philanthropic support to Stanford University's Institute for Human-Centered Artificial Intelligence, HI, and to the Alan Turing Institute, the United Kingdom's National Institute for Data Science and Artificial Intelligence.

• 13:52Reid Hoffman?

  I recognize that some might say such qualifications actually disqualify my perspective on AI, that my optimism is merely hype, that my idealism about how we might use AI to create broad new benefits for society is just an effort to generate economic return for myself, that my roles as founder, investor, advisor, and philanthropic supporter of many AI-focused companies and institutions create an ongoing incentive for me to over-promote the upsides and downplay the dangers and downsides. I argue that the opposite is true. I'm deeply involved in this technology, and I want to see it succeed exactly because I believe

• 14:34Reid Hoffman?

  it can have profoundly positive impacts on humanity. My engagement in this domain has meant that I've seen firsthand the progress being made. That has strengthened my commitment, and thus I've continued to invest in and support a widening range of companies and organizations. I stay alert to potential dangers and downsides, and am ready to adapt, if necessary, precisely because I want this technology to succeed in ways that broadly benefit society.

• 15:03Reid Hoffman?

  One reason iterative deployment makes so much sense in the case of pioneering technologies like AI is that it favors flexibility over some grand master plan. It makes it easier to change pace, direction, and even strategy when new evidence signals the need for that. Meanwhile, here we are, presenting our argument to you in a book. Roughly 2,400 years ago, Socrates critiqued the written word for its lack of dynamism in Plato's Phaedrus, and for the way it made knowledge accessible to anyone. You know, Phaedrus, writing shares a strange feature with painting. The offsprings of painting stand there as if they are alive, but if anyone asks them

• 15:48Reid Hoffman?

  anything, they remain most solemnly silent. The same is true of written words. You think they were speaking as if they had some understanding, but if you question anything that has been said because you want to learn more, it continues to signify just that very same thing forever. When it has once been written down, every discourse rolls about everywhere, reaching indiscriminately those with understanding no less than those who have no business with it, and it doesn't know to whom it should speak and to whom it should not. For Socrates, apparently, fixing his thoughts into written text represented a loss of agency.

• 16:32Reid Hoffman?

  Had he turned his teachings into books himself, or rather scrolls, the reigning technology of his day, he would not have been able to control who read them. He would not have always been on hand to provide updates on his thinking, elaborate on nuances in the text, or correct misreadings. Consequently, face-to-face dialogic inquiry was his preferred technology for transmitting ideas. But clearly, generations of authors and readers thought differently. Why? Because ultimately, written works increase the agency of authors and readers, enabling the latter to engage with, learn from, modify, expand upon, and yes, perhaps even misinterpret

• 17:16Reid Hoffman?

  or appropriate ideas from authors with which they might never have otherwise crossed paths.

• 17:22Reid Hoffman?

  As printing technologies improved, books evolved into a transformative global resource. Rolling about everywhere, indiscriminately reaching everyone, they functioned as early mobility machines, decoupling human cognition from human brains, democratizing knowledge, accelerating human progress, and providing a way for individuals and whole societies to benefit from the most profound and impactful human insights and innovations across time and space. Of course, there are myriad other ways to share information now, and we'll be using many of them to convey the ideas in Super Agency too. Along with the usual podcasts and social media, we'll be experimenting with AI-generated

• 18:07Reid Hoffman?

  video, audio, and music to augment and amplify the key themes we're exploring here. To see how, check our website, superagency.ai.

• 18:20Reid Hoffman?

  But we're starting with a book, in part as homage to the essential truth that technologies that often seem decidedly flawed and even dehumanizing at first, usually end up being exactly the opposite.

• 18:38Speaker 02

  All right, so that was the audiobook intro to Super Agency, and then up next is the AI-generated podcast recap from Notebook LM.

• 18:49Speaker 03

  Welcome to our deep dive into the intro of Super Agency, a book that really tackles like all of our AI anxieties head on.

• 18:57Speaker 02

  Yeah.

• 18:57Speaker 03

  You know, losing our jobs, rogue AI, maybe even like a Terminator situation.

• 19:02Speaker 02

  Right.

• 19:03Speaker 03

  But they jump right into that conversation, but with a fascinating historical twist.

• 19:08Speaker 02

  Yeah, before we, you know, freak out about AI, they want us to look back at how people reacted to, wait for it, the printing press.

• 19:16Speaker 03

  The printing press.

• 19:17Speaker 02

  Yeah.

• 19:17Speaker 03

  Seriously.

• 19:18Speaker 02

  Yeah.

• 19:19Speaker 03

  I mean, that's, that feels like ancient history.

• 19:21Speaker 02

  It does seem that way, but the book makes this, this really compelling point. The printing press, along with other groundbreaking inventions like the telephone, the automobile.

• 19:32Speaker 03

  Okay.

• 19:32Speaker 02

  These were all initially met with like widespread fear and resistance.

• 19:35Speaker 03

  So instead of like, you know, embracing these new technologies, people were like running for the hills.

• 19:41Speaker 02

  Mm-hmm.

• 19:41Speaker 03

  I can't quite picture like people protesting against the Model T.

• 19:45Speaker 02

  Well, so imagine this, right? People worried that the automobile would lead to like a decline in marriage. Because couples could elope more easily.

• 19:55Speaker 03

  Oh, interesting.

• 19:56Speaker 02

  It sounds kind of silly now, but it just shows how people feared that technology would disrupt, you know, societal norms and values.

• 20:04Speaker 03

  OK, I see what you mean. People were afraid of the unknown.

• 20:08Speaker 02

  Right.

• 20:08Speaker 03

  Of how these technologies might upend their lives. But is that really the same as our anxieties about AI? I mean, it feels like a whole different ballgame.

• 20:17Speaker 02

  It's true. AI feels different, and that's because it simulates human intelligence in ways that no other technology has before.

• 20:23Speaker 03

  Yeah.

• 20:23Speaker 02

  Like, the printing press could reproduce words, sure. But it couldn't write sonnets.

• 20:28Speaker 03

  Right.

• 20:28Speaker 02

  AI can do that now, and that raises some genuinely new and complex questions.

• 20:33Speaker 03

  Exactly. AI can write poems, compose music, even hold conversations that feel like eerily human.

• 20:40Speaker 02

  Yeah.

• 20:41Speaker 03

  So the question is, are fears justified this time? Or are we just repeating the same old pattern of, like, techophobia?

• 20:47Speaker 02

  That's the big question, right? And Hoffman and Beto acknowledged those fears, particularly the ones around, you know, job displacement and the potential for AI to be misused.

• 20:59Speaker 03

  Right.

• 20:59Speaker 02

  However, they argue that trying to ban or severely restrict AI is just not a viable solution.

• 21:05Speaker 03

  So they're not advocating for, like, a world where we try to outlaw AI?

• 21:10Speaker 02

  No, not at all.

• 21:11Speaker 03

  OK.

• 21:11Speaker 02

  They use the analogy of herding cats to describe how difficult, if not impossible, it would be to coordinate a global effort to control AI development.

• 21:22Speaker 03

  Herding cats. Yeah, that sounds about right. Especially when you consider how many different companies and governments and individuals are working on AI.

• 21:29Speaker 02

  Exactly. It's everywhere.

• 21:30Speaker 03

  All over the world.

• 21:31Speaker 02

  It's not just the practical challenges either.

• 21:33Speaker 03

  OK.

• 21:34Speaker 02

  They also argue that trying to stifle innovation in AI would be counterproductive.

• 21:38Speaker 03

  OK.

• 21:38Speaker 02

  Because, you know, someone, somewhere is going to push those boundaries.

• 21:41Speaker 03

  Right.

• 21:41Speaker 02

  Regardless of regulations.

• 21:43Speaker 03

  OK, so if banning AI isn't the answer.

• 21:46Speaker 02

  Right.

• 21:46Speaker 03

  What do they suggest we do? Just, like, cower in fear as the robots take over?

• 21:51Speaker 02

  Yeah, not quite. Hoffman and Beto believe we need a more proactive approach.

• 21:56Speaker 03

  Okay.

• 21:56Speaker 02

  One where we actively shape AIs development with a focus on what they call human agency.

• 22:02Speaker 03

  Okay.

• 22:02Speaker 02

  This idea forms the core of their book.

• 22:05Speaker 03

  Gotcha.

• 22:05Speaker 02

  And it's what we'll be exploring in more detail as we dive deeper.

• 22:09Speaker 03

  Awesome, I'm ready.

• 22:09Speaker 02

  Yeah, you know, I'm sure a lot of you are thinking like, how do we even control something as powerful and potentially unpredictable as AI?

• 22:18Speaker 03

  You took the words right out of my mouth.

• 22:19Speaker 02

  Yeah.

• 22:20Speaker 03

  It's like a tidal wave, you know? How do we even begin?

• 22:23Speaker 02

  Right, they argue that instead of trying to control every aspect of AI, we need to focus more on guiding its development

• 22:30Speaker 03

  Okay.

• 22:31Speaker 02

  in a direction that benefits humanity.

• 22:33Speaker 03

  Okay.

• 22:33Speaker 02

  And this is where the whole human agency concept comes into play.

• 22:36Speaker 03

  Okay, but human agency, I mean, that sounds a little abstract. What does that actually look like in the real world?

• 22:42Speaker 02

  Well, think of it this way, human agency is all about empowering individuals to make choices and take actions.

• 22:47Speaker 03

  Okay.

• 22:48Speaker 02

  That shape their own lives. Even as AI becomes more and more prevalent, it's about ensuring the AI tools enhance our capabilities rather than like replace or diminish our autonomy.

• 23:00Speaker 03

  So it's not about humans versus machines?

• 23:02Speaker 02

  Right.

• 23:03Speaker 03

  It's about figuring out how we can partner with AI.

• 23:07Speaker 02

  Exactly.

• 23:07Speaker 03

  To create a better future.

• 23:08Speaker 02

  Exactly, and and they propose a pretty compelling approach to actually achieving this called iterative deployment.

• 23:16Speaker 03

  Iterative deployment, so are we like beta testing AI on all of society?

• 23:21Speaker 02

  In a way, yeah.

• 23:23Speaker 03

  It's a later.

• 23:24Speaker 02

  Acknowledging that we don't have all the answers upfront, right?

• 23:26Speaker 03

  Right.

• 23:26Speaker 02

  We need to be flexible and adaptable as AI technology evolves.

• 23:30Speaker 03

  Right.

• 23:30Speaker 02

  So imagine like AI being used to, let's say, optimize traffic flow in a city.

• 23:35Speaker 03

  Okay.

• 23:36Speaker 02

  As the system gathers data and learns, it can constantly adjust traffic patterns to improve efficiency and safety.

• 23:42Speaker 03

  So we're constantly learning and adapting alongside AI.

• 23:45Speaker 02

  Yeah.

• 23:46Speaker 03

  Making adjustments to make sure it's aligning with our goals and values.

• 23:49Speaker 02

  Exactly, and to guide this whole process.

• 23:52Speaker 03

  Okay.

• 23:52Speaker 02

  They emphasize the importance of developing what they call a techno-humanist compass.

• 23:56Speaker 03

  Okay.

• 23:57Speaker 02

  It's like a, almost like a set of guiding principles.

• 23:59Speaker 03

  Okay, like a North Star.

• 24:00Speaker 02

  Yeah, exactly. That helps us make decisions about AI development that prioritize, you know, human well-being and empowerment.

• 24:08Speaker 03

  So how do we make sure that this techno-humanist compass actually guides our decisions? It can't just be like a. A nice sounding idea.

• 24:18Speaker 02

  You're absolutely right. It requires a conscious effort to embed these values into the design and development of AI systems. For example, like ensuring equitable access to AI technology.

• 24:28Speaker 03

  Right, equity is important.

• 24:29Speaker 02

  Hugely important, yeah. We need to make sure that people from all walks of life have the opportunity to shape and benefit from AI.

• 24:38Speaker 03

  Not just a select few.

• 24:39Speaker 02

  Not just a select few, right? We've seen what happens when technology only benefits a small segment of society.

• 24:44Speaker 03

  Right, it creates divides.

• 24:46Speaker 02

  Exactly, yeah. It just exacerbates existing inequalities.

• 24:49Speaker 03

  Yeah.

• 24:50Speaker 02

  Another key aspect of this whole techno-humanist approach

• 24:53Speaker 03

  Okay.

• 24:54Speaker 02

  is focusing on AI applications that address, you know, these pressing global challenges.

• 25:00Speaker 03

  Like what?

• 25:01Speaker 02

  Well, instead of primarily using AI to just like make our lives more convenient or entertaining.

• 25:07Speaker 03

  Okay.

• 25:08Speaker 02

  We should prioritize its use in areas like climate change mitigation.

• 25:12Speaker 03

  Okay.

• 25:12Speaker 02

  Pandemic prevention.

• 25:14Speaker 03

  Yeah.

• 25:14Speaker 02

  Sustainable energy development.

• 25:16Speaker 03

  So we need to be intentional

• 25:18Speaker 02

  Hmm.

• 25:18Speaker 03

  about directing AIs power towards solutions that benefit all of humanity.

• 25:24Speaker 02

  Exactly.

• 25:24Speaker 03

  Not just a privileged few.

• 25:25Speaker 02

  That's right. It sounds like a pretty tall order.

• 25:28Speaker 03

  It does.

• 25:29Speaker 02

  But it's also incredibly inspiring.

• 25:30Speaker 03

  It is.

• 25:31Speaker 02

  And while Hoffman and Beto acknowledged the very real risks associated with AI.

• 25:37Speaker 03

  Right.

• 25:37Speaker 02

  They're cautiously optimistic.

• 25:39Speaker 03

  Okay.

• 25:39Speaker 02

  That we can actually navigate these challenges if we approach this whole new era.

• 25:44Speaker 03

  Yeah.

• 25:45Speaker 02

  With a sense of shared responsibility.

• 25:46Speaker 03

  Okay.

• 25:47Speaker 02

  A commitment to placing human values at the forefront of our decisions.

• 25:51Speaker 03

  I like that. We've been talking about this, uh, this idea of a techno-humanist compass guiding our approach to AI.

• 25:59Speaker 02

  Yeah.

• 25:59Speaker 03

  But how do we make sure it's not just like a lofty concept? But something that actually influences like, you know, real world decisions.

• 26:07Speaker 02

  Yeah, that's where things get really interesting. And where we all have a role to play. Actually, Hoffman and Beto emphasize that this isn't something we can just like leave to tech experts. We're policymakers.

• 26:19Speaker 03

  Yeah.

• 26:19Speaker 02

  We need a much broader societal conversation about AI, one that involves everyone.

• 26:25Speaker 03

  So it's less about like finding all the right answers right now and more about asking the right questions.

• 26:30Speaker 02

  Exactly.

• 26:31Speaker 03

  Collectively.

• 26:32Speaker 02

  Collectively, yeah. Questions like, you know, what are the ethical boundaries we need to set for AI development?

• 26:38Speaker 03

  Right.

• 26:38Speaker 02

  Or how do we ensure AI is used to empower people, not exploit them?

• 26:43Speaker 03

  Yeah.

• 26:43Speaker 02

  And these aren't just technical questions. They touch on, you know, our values, our social fabric, our vision for the future.

• 26:48Speaker 03

  Yeah, it makes you realize that the decisions we make about AI today are gonna have ripple effects for like generations to come.

• 26:56Speaker 02

  Oh, yeah.

• 26:57Speaker 03

  It's a bit daunting, to be honest.

• 26:58Speaker 02

  It is. It's a huge responsibility. But Hoffman and Beto remind us that we've navigated technological revolutions before.

• 27:05Speaker 03

  Right.

• 27:06Speaker 02

  Think about the printing press.

• 27:07Speaker 03

  Okay.

• 27:08Speaker 02

  It brought about massive changes, both positive and negative.

• 27:12Speaker 03

  Of course.

• 27:12Speaker 02

  But ultimately, you know, led to a more informed and connected world.

• 27:17Speaker 03

  So we need to learn from history.

• 27:18Speaker 02

  Yeah.

• 27:18Speaker 03

  And embrace the uncertainty and be prepared to adapt as we go.

• 27:23Speaker 02

  Precisely. And this iterative deployment model that they propose.

• 27:28Speaker 03

  Right.

• 27:28Speaker 02

  Allows for that.

• 27:29Speaker 03

  Ultimately, the message is one of cautious optimism, right?

• 27:32Speaker 02

  Yeah.

• 27:33Speaker 03

  That we can harness AIs power for good if we approach it thoughtfully.

• 27:38Speaker 02

  That's the key takeaway. Super agency doesn't offer like a simple roadmap, but it gives us a framework for navigating this, you know, complex landscape.

• 27:49Speaker 03

  Hmm.

• 27:49Speaker 02

  And it reminds us that the future of AI isn't predetermined.

• 27:52Speaker 03

  Right.

• 27:52Speaker 02

  It's something that we actively shape.

• 27:54Speaker 03

  Yeah.

• 27:55Speaker 02

  Through our choices and actions.

• 27:57Speaker 03

  That's a powerful thought to end on. So as we wrap up this deep dive into super agency, we want to leave you with a question. What aspect of your life would you want AI to augment or improve? And what safeguards would you want to put in place? Think about it, discuss it, and let's work together to ensure that the future of AI is one that benefits all of humanity. Thanks for joining us.

• 28:25Reid Hoffman?

  Possible is produced by Wondery Media Network. It's hosted by Aria Finger and me, Reid Hoffman. Our showrunner is Sean Young. Possible is produced by Katie Sanders, Edie Allard, Sarah Shleid, Vanessa Handy, Aliyah Yates, and Paloma Moreno Jimenez. Jenny Kaplan is our executive producer and editor.

• 28:45Aria Finger?

  Special thanks to Soria Yalamanchili, Sayida Sapieva, Ian Alice, Greg Beato, Parth Patel, and Ben Relles. And a big thanks to Malia Agudelo, Scott Wallace, and Fanasi Delos.