AI Optimism Has a Trust Problem

Today on the AI Daily Brief: As the political stakes increase, a more positive vision of AI. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. All right, friends, quick announcements before we dive in. First of all, thank you to today's sponsors: KPMG, Rackspace, Blitzy, and HyperAgent. To get an ad-free version of the show, go to patreon.com/aiDailyBrief, or you can subscribe on Apple Podcasts. And to learn more about sponsoring the show, send us a note at sponsors@aidailybrief.ai. Now, one quick note, I had not been intending to do this, but this episode got very long. And as you'll see, I think gets to some of the most important conversations that are increasingly being had as AI moves more firmly into the political sphere. So this will be a main-only episode. We will be back with our normal format and the headlines again tomorrow. There is no doubt that the political conversation around AI is getting louder and louder. Part of that is the natural consequence of models growing in power, and part of that is the natural consequence of elections coming up. Whatever the proximate causes, however, from the standpoint of both politician and American voter interest, the issue of AI is growing in significance. Alongside that, different companies are staking their claims for the story they want to tell about AI. Despite being increasingly isolated from the rest of the industry in this, Anthropic seems determined to keep telling us about the potential negative consequences of AI, with the most recent example being their Hope in Hard Questions campaign. Now, whether that approach to storytelling can survive the IPO process remains to be seen. OpenAI, meanwhile, has shifted fairly aggressively off this type of messaging. Sam Altman has said publicly on X that he was wrong about his expectations about how AI would interact with jobs. And had been excited to see that AI was primarily a tool for augmenting people rather than replacing them. Then into that space comes Mark Zuckerberg and Meta. For the last year or so, most stories about Meta and AI have been some combination of incredulity at the prices that they were paying to recruit top researchers or almost schadenfreude-laced commentary about how they hadn't done anything with all that spend yet. Unlike his peers at the other labs, Zuckerberg had never publicly shared this sort of doom and gloom that seemed to be a part of their assessment of the likely future. But over the last several weeks, it's become clear that not only does Zuckerberg not share that perspective, he wants to plant his flag in exactly the opposite place. A couple of weeks ago, the Wall Street Journal published an opinion piece of his called "The AI Future Is for Everyone," which argued that this power concentrated in a few hands is the worst possible outcome. Then on Monday of this week, August 10th, he published a longer manifesto version of this, clocking in at 6,500 words. The title and the core thrust is the same: The Future is for Everyone: The Path to a Positive AI Future. But he goes much farther in this piece to actually lay out some of how he sees it playing out. So let's look first at a few excerpts from the piece, and then we'll get to the reactions. The defining questions of our age, Mark writes, are who will have access to superintelligence and what we will direct it towards. Will it be centralized and restricted to a few institutions, or will it be a tool that empowers everyone? Obviously, for Meta, the answer is clear. He continues, we propose a philosophy based on individual empowerment as the source of prosperity, invention as the primary purpose of superintelligence, and balance of power as the foundation of safety. Reiterating a line from the piece that was published in the Wall Street Journal, he says, it is surprising that the discourse for many developing AI is so filled with doom. I do not understand why anyone who believes that AI will eliminate most jobs and much of humanity's relevance would rush to build that future. The notion that AI is so dangerous that the only safe path is an extreme concentration of power seems inherently problematic. Historically, hoping that an absolute power will benevolently provide for humanity if sufficiently enlightened has not led to safe or positive outcomes. So that's the big idea. It was the big idea that he was exploring before. But what are some of the details? First, on job growth in the economy, he writes, people fear that automation will outpace individuals' capability growth, leading to job displacement followed by a difficult period as people learn new jobs. But there is no rule that AI must increase automation faster than it increases individuals' capabilities or demand for new skills. Recent statistics suggest it may be more likely that individuals' capability growth could match or outpace automation, in which case people will gain the ability to do many new things before their current jobs change. This would lead to a healthy balance and potentially even job growth. What's interesting to me about this way of putting it is that he's actually framing this as a math problem. Which happens faster, automation of roles or the enhancement of individual capabilities and the demand for new skills? What's interesting is that the forces of corporate inertia apply most significantly to the speed of automation question, as in even things that could be automated in many cases won't be automated because of corporations' slow-moving natures. Meanwhile, the speed with which individuals can enhance their capabilities is not necessarily as bound. In terms of other reasons Zuckerberg points to be optimistic about jobs of the future, he points to the fact that there will, quote, always be a finite amount of compute and therefore an opportunity cost for how we use it. His argument is that if people can use AI to invent incredibly valuable new things, then it will make more sense to allocate it towards that rather than automating existing jobs. But even within the context of individual jobs, he points to some that don't really exist right now that might in the future. A generation ago, he points out, there were no app developers, social media creators, electrical vehicle technicians, or data center operators. In the near future, there will be new jobs that aren't common today, like one-person product studios designing custom toys, furniture, or clothes, worldbuilders and experience designers creating games, stories, and adventures, personal biologists using superintelligence to formulate personalized treatments, and much more that we can't yet conceive. Which is not to say that the shape of the economy won't change. Company sizes, he writes, may shrink, just as they did in the transition from industrial giants to tech companies. But, he argues, this doesn't mean fewer jobs overall. It implies a larger number of companies with fewer people each. There are many more valuable companies and services to build than people are able to build today. Now, when it comes to how to deal with risk and safety issues, Zuckerberg focuses on what he describes as the balance of powers. He writes, there is a risk that an imbalance of power between individuals and government leads to a loss of freedom or totalitarian state. This is a constant tension in democracy. We all want individual freedom while also having a government capable of keeping us safe. To maintain freedom, we must ensure that superintelligence primarily empowers individuals. The ideal in liberal democracy is that people naturally hold all rights and only agree to restrict some freedoms to protect the common good. Similarly, individuals should have access to personal superintelligence and should only be subject to restrictions when truly required. Now, interestingly, he does not propose that this means government keeping its hands off AI. Instead, he proposes more active and ongoing engagement rather than just periodic check-ins when a model is ready to be released. He writes, we must ensure that government has access to the tools and resources it needs to enforce our laws and protect our security. We can achieve this through deeper proactive collaboration between the labs and government in a way that strengthens the government's national security capabilities without restricting or delaying people's access to advanced models. That is, rather than waiting until a model is ready to release for the government to review and start using it, my proposal is that leading labs should provide the government with intermediate training checkpoints of new advanced models and technical staff so the government can harden and secure critical systems against new risks. This way, the government gains a security capability without restricting or delaying individuals' access to personal superintelligence or causing an imbalance of power. Now, part of the reason that he suggests this is that he is arguing that even the sort of 30-day review period that the government is discussing could be problematic. Later in the piece, he writes, any policy that slows American model releases even by a month could add significant risk to American leadership while letting foreign models race ahead. Now, there is a ton more in here. There's a section on alignment and addressing existential risk, as well as a section on maintaining control of superintelligence. But one last area that I want to highlight before we get into reactions is much more here and now, and that's the section called Building AI Infrastructure with Communities. This is basically the section where Zuckerberg lays out his vision for how data centers can work for the communities that they are in. He writes, sustainable infrastructure development means that communities must benefit significantly from each project. This includes high-paying local jobs, investment in schools and public services, ensuring energy prices don't rise, and taking care of the environment. As tax revenue grows, this also benefits teachers, law enforcement, fire departments, and more. But in this section, as he continues, it's not so much about what could be, but about what Meta is doing right now. Among the different initiatives that Zuckerberg points to, there's America's Workforce Academy, which is designed to provide free training for these sort of skilled tradespeople that are needed around the infrastructure buildout. He discusses their commitment to building their own energy-generating infrastructure, which hopefully not only covers their energy costs but could even provide a surplus of low-cost energy back into the communities where they're operating. He talks about the water efficiency of their data centers and broadly says that they want to invest in the communities where they're building. In a few minutes, we'll come back to the initiative that they announced alongside this, which is meant to do exactly that. One of the more common first reactions was basically, wow, that's a lot of words. Referencing Dario Amadei's 13,000-word-strong Machines of Loving Grace, FinTech Biz Weekly's Jason McCullough writes, So now he's imitating Dario Amadei? Although, as many point out, in many ways this piece is positioning Zuckerberg to be the exact foil to Dario. Now, across about a dozen different "5 things to know about Zuckerberg's AI manifesto" type pieces, there were a few themes that really stood out. The first was the commitment or perhaps recommitment to open source and open weight AI. Bloomberg points out that across the 6,500 words, Zuckerberg refers to open source at least 16 times, stressing particularly the importance of US leadership in that area. Many also picked up on the interesting combination of both wanting less government oversight, but also more government oversight, as The Verge put it. That publication points to his comments on open source as an example of less government oversight, but of course, the proactive sustained engagement that we were just discussing as an example of more involvement. I think the reality is that Zuckerberg is not discussing oversight one way or another. What he's suggesting is a totally different and reimagined relationship between government and the private sector.— that doesn't view them simply as a reactive problem prevention body. Others picked up on Zuckerberg's continued themes of calling out the concentration of AI control as a major risk, but one of the areas that got the most attention was the section around data center communities. Part of that was that alongside this letter, Meta announced a $1 billion fund to invest in those communities. Following the overall branding, the initiative is called the Future is for Everyone Fund. Now, of course, the details on exactly what that fund is going to do remains to be seen, But if you have listened to me diatribe about this, this is exactly the sort of initiative, at least, that I have been arguing has to be written into the cost of data centers in general moving forward. In other words, I don't believe that the billion-dollar fund, even if it is constructed as a foundation or philanthropic initiative, should be viewed as anything other than a mission-critical business expense. Hello everyone. One big change around AI is we've shifted our thinking from how we rank our pages to how do we become the source that AI trusts enough to answer with. At KPMG, they're seeing this firsthand. AI-generated results now surface answers directly, often without a single click. That's why they are increasingly focused on generative engine optimization, or GEO—structuring content so AI systems can retrieve it, understand it, and cite it as trusted authority. This is not just an SEO evolution, but a visibility mandate. And indeed, the GEO mandate from KPMG is simple: if AI is shaping decisions, your expertise needs to show up inside the answer. Read all about it at kpmg.com/us/geo. Again, that is kpmg.com/us/geo. One of the more interesting shifts in enterprise AI right now is how quickly the conversation is moving towards infrastructure and operations. As AI moves into core workflows, regulated data environments, and agentic systems, Enterprises need governed infrastructure and inference that can operate reliably day-to-day with clear operational accountability built in from the start. As those systems scale, the operating model increasingly becomes part of the AI strategy itself. Rackspace Technology is the operator of the full enterprise AI stack, from agents to infrastructure across private cloud, hybrid cloud, and edge environments. Rackspace builds and operates governed AI infrastructure, inference, and production AI systems for organizations where sovereignty, compliance, and uptime are non-negotiable. Their forward-deployed engineers stay embedded beyond deployment to help operationalize and run AI in live environments. To learn more about where enterprise AI runs and outcomes scale, go to rackspace.com. Blitzy deeply understands your codebase before it writes code. Here's the first place that pays off: security in the age of AI. Vulnerabilities don't live in isolation. They live buried inside millions of lines of interconnected code, where patching one thing quietly breaks three others. That's why surface-level scans fail. Blitzy starts from its knowledge graph of your entire application, identifies and surfaces CVEs across the full estate, proactively recommends patches, and can execute the PR. Each fix is grounded in how your systems connect and validate so nothing new breaks, and the knowledge graph dynamically updates, keeping you ahead of an ever-accelerating threat landscape. One Blitzy customer resolved 21 active CVEs across 6 core microservices in 4 days. Zero compile errors, every validation scan clean, months of planned work fixed in less than a week. Security remediation grounded in real architectural context at the speed of compute. Harden your codebase at blitzzy.com. That's B-L-I-T-Z-Y dot com. This episode of the AI Daily Brief is brought to you by HyperAgent, where you run fleets of agents your team can manage together. New users get $1,000 in inference. Forget local agents and chat workflows waiting on your laptop to be prompted. HyperAgent deploys always-on agents in the cloud, doing real work across the tools your team already uses. Marketing's agent turns competitor moves into landing pages. Sales' agent enriches leads, drafts emails, and updates the CRM. Ops' agent chases the paperwork and tracks the budget. Every agent has access to shared context and follows your rules about scope and approvals. It's time you add agents that feel like teammates. Hire yours at HyperAgent, built by the team at Airtable. Claim your $1,000 in inference at hyperagent.com/aidailybrief. Now, when it comes to jobs, we start to see where one type of skepticism creeps in. In discussing his argument that AI could expand the job market, Bloomberg writes, "There is no small amount of irony to this take as Zuckerberg cut 8,000 jobs from Meta earlier this year in response to the company's pivot towards AI." And while there is no inherent contradiction between Zuckerberg's argument of a larger number of companies with fewer people and laying people off, it still gets to the biggest issue with this overall,, which is a question of who the messenger is. TechCrunch wrote a piece called Mark Zuckerberg's AI Manifesto is Exactly Why People Don't Like AI. And one of the things that that piece gets at is something that I have long felt, which is that a lot of the animosity towards AI is actually the AI industry paying for social media's sins. TechCrunch points out that a recent survey found that 64% of Americans, nearly two-thirds, believe that social media has been harmful to democracy, with a similar percentage saying that it should be more heavily regulated. The author writes, "I don't bring this up to imply that Zuckerberg should withdraw to the wilderness in shame, but the fallout from social media is one of the central reasons we're now seeing so much anxiety about the social impact of AI. Whether it's fair or not, the public does not trust tech executives to make sure new technologies like this have a positive impact on society. Instead of acknowledging that and trying to win back their trust, this essay demonstrates over and over again how the trust was lost in the first place." It goes on to argue that basically the hazy generalities lead to less rather than more trust. Over and over when I read a commentary, especially outside of the advanced AI users on X and inside the more general business landscape of LinkedIn, questions of trust came up over and over and over again. The other thing that people responded really negatively to were Zuckerberg's examples for what personal superintelligence could do. Capturing both of these points, Sorabell F writes, Zuckerberg talks about the importance of privacy, trust, the dangers of extreme concentration of power, the need for governance and oversight. Yes, he actually said that. Oh, and he said that superintelligence is helping his daughter bake recipes and make videos in a few hours. Yep, I'm not lying. And I actually think that these sort of use case examples are more damaging than they potentially at first seem. There is a growing sense that the tech industry just doesn't actually understand normal people. Elizabeth Lopatto from The Verge captures this in her response, Mark Zuckerberg doesn't understand how to live., an AI future of sleek, streamlined, and totally empty relationships. Pointing to Zuckerberg's pitch that, quote, everyone will have an exceptionally capable personal agent that understands you, your goals, and everything you care about, and will work 24/7 on your behalf to improve your relationships, health, career, finances, home management, hobbies, and more, Lepato writes, I'm struck here by the inclusion of relationships and hobbies. What strengthens a relationship? Almost always it is personal investment. Perhaps an AI agent might be better at predicting what my father will want for his birthday than I am,, but it definitionally cannot give him my time and consideration. Love is not just a feeling, it's a way of paying attention. She continues, Zuckerberg uses an example that is telling in a grim kind of way. He asked an AI to pick out a personalized recipe to bake with his daughter. He could have considered what kind of baking his daughter enjoys, what her skill set is, and what he might have been able to teach her, then picked a recipe. Maybe he would have picked a bad recipe and they would have bungled it, but that might have been a fun experience and a good story. Or maybe he would have picked a family recipe and it would have given him the opportunity to tell her about when he was a child or what his grandmother was like. Certainly, he's a busy man, but wouldn't that make the gift of his attention to her hobby even more precious? Perhaps he might have come to understand her better by spending more time thinking about what she likes. Without that attention, the baking they do together seems more focused on accomplishment than quality time. She then goes on to point out the inherent contradiction of trying to optimize a hobby. She writes, the point of a hobby is that you do it. Granted, many of my hobbies are physical activities— hiking, running, yoga, rock climbing— but an AI cannot read a book for me because the activity of reading a book is absorbing its words into my mind. A summary doesn't do the same thing. A summary cannot summon the mood nor the actual pleasure of reading the words themselves. An AI cannot replicate the soothing quality of knitting my own scarf. It is the knitting itself that soothes. Now, some of you also might be remembering the dust-up from this Sam Altman tweet from a couple of weeks ago where he wrote, cool use case of ChatGPT work I heard last night. Connect your family calendars and explain your kids' interests. Every morning for the drive to school, have it make a podcast that talks about one kid's soccer game that afternoon, one kid's upcoming birthday, some news, etc. And man, even the AI faithful instantly knew that even though he probably didn't intend it this way, it came off as Altman and by extension the larger tech industry not being able to be bothered to actually engage with and talk to their children. Now, to be clear, I think that a lot of the discourse that surrounded that Sam tweet was this new emerging strand that basically views every parental or family use case of AI as a cop-out and a shirking of traditional responsibilities. And I think that is absolutely preposterous. Host of the How I AI podcast, Claire Voe, reposted a tweet from June of 2023 when she had used Midjourney and ChatGPT to generate a deck of Pokémon-style cards with characters from Greek mythology as a weekend art project and added, crap on parents using AI all you want, but it was doing things with and for my kids that AI pilled me 3 years ago. Not coding, not personal productivity, but little projects that got my kids and I off screens and playing together. And if you were interested in more of the awesome and very cool ways that parents and families are using AI to massively increase their value not only around learning and schooling, but around their engagement with each other. I encourage you to follow Claire. She's @ClaireVo, V-O, or Jesse Genet, @JesseGenet, who is just constantly posting her experiments with using AI around interesting school and family projects. Still, none of those counterexamples change the fact that by and large, people feel like tech is out of touch with regular people. On the flip side, there are a lot of folks who were quite excited about this. Nathan Lambert pointed to Zuckerberg's argument that, quote, open source is a positive and important force for empowering people and said, really glad to have Mark back pushing for Team Open. And responding to Meta putting its money where Mark's mouth is on this front, Justin Schroeder wrote, delicious, this is the Zuckerberg the world needs. Now what he was responding to there was the announcement of a new open source model from Meta called Muse Glimmer. The model has 30 billion parameters, making it small enough to run on local hardware. Performance looks pretty competitive with similar-sized models, generally outperforming Google's Gemma 4 31B and slightly behind Qwen 3.6 27B on a few key metrics. Meta is selling this specifically as an agentic model, claiming it's perfect for running an agent that, quote, manages your schedule, drafts your messages, organizes your files, and learns how you work. Now, their argument is that for a model to do all that highly personal work well, it, quote, needs deep access to personal context. That required access is their entire logic for open sourcing a model like this. What's more, in addition to MuseGlimmer, Zuckerberg flagged that Meta plans to release the weights for MuseSpark 1.2 soon, adding another frontier-adjacent open weights model to the ecosystem. Yet for some, they're just happy to see a positive, optimistic take for once from the AI industry itself. Robert C. O'Brien writes, optimism and belief that our best days are ahead of us are quintessential American traits. Thank you, Mark, for reminding us that these traits also apply in the AI context. And while it's easy to be cynical or to dismiss the message because of the messenger, I have to say that the discussion it is kicking up is far wider and more thoughtful than even I would have guessed before it was published. If you go Google Zuckerberg essay, almost every major outlet has written some article about the piece. And while yes, Many of them are totally reasonably sharing their skepticism with certain parts or their recurring critiques of Zuckerberg himself or Meta as a company. To the extent its goal was to at least have a different type of discussion about AI, there is evidence that it's working. It would be easy to overstate this, but I do sense the very beginnings of a recalibration of the conversation around AI. I have long argued that the public discourse about AI is ruled by extremes on either side that don't actually represent the perspectives of the vast majority of people in the middle. And when I poke around on media that isn't my direct AI bubble, I'm starting to see a lot more assertion of that middle space. I'm seeing people, for example, in the same TikTok video, argue that while AI should absolutely not be used to, for example, replace artists, that there were some valuable uses for it done well. And more than that, that the anti-AI perspective was getting almost performative. Now, opinion polls by and large don't agree with this take. And they are often showing concern increasing. My belief is that lots of folks are being onboarded into this conversation through the lens of concerns, cybersecurity issues, hacking stories, and anti-data center activism. But I believe that by and large, when we treat people as smart, they act smart. And there is immense room when it comes to AI for not only thoughtful nuance in this discourse, but agency in where the conversation takes us. Candidly, Mark Zuckerberg might not have been the messenger that I would have chosen, but at this point, I will take anyone with any amount of voice from the AI industry loudly proclaiming the good that AI might let us win. For now though, that is going to do it for today's AI Daily Brief. Appreciate you listening or watching as always, and until next time, peace.