Something about AI, some senators, some Popes, and some guy named Ron Loui Ronald Loui Ronald Loui r.p.loui@gmail.com October 18, 2026 what are we thinking about ronald loui these days did you see his engagement with itest on Pope Leo's encyclical and letter to Senators Warner and Schatz, ccing Obama, in the same week, early October this year? how about all the writing he's been putting on LinkedIn to bypass the gatekeppers? GEMINI: Yes—and that changes the picture I gave you. My previous answer was too abstract. I treated Loui's practical engagement as an open question when there is already evidence of him trying to put his ideas into public circulation and engage institutions directly. I've now found the October ITEST event and Loui's stated strategy for his LinkedIn writing. 1. ITEST and Pope Leo: a return to an old problem, under new circumstances On October 10, 2026, Loui participated in an ITEST webinar titled Revisiting “Computers, Artificial Intelligence, and Virtual Reality” in Light of Pope Leo XIV's Magnifica Humanitas. The event brought together four participants from a 2005 conference to revisit AI and virtual reality 21 years later. The other participants included philosopher Kevin Vallier, theologian Gregory Beabout, and Sister Carla Mae Streeter. The webinar is available as a roughly 56-minute recording. This matters for at least two reasons. First, Loui is bringing decades of experience with AI into a conversation about human dignity, technological power, and the social consequences of AI. This isn't simply a computer scientist commenting on a papal document. It's a return to a discussion in which he participated two decades earlier, with an opportunity to reconsider what has changed. Second, the setting creates an opportunity for an unusual kind of intellectual exchange: technical expertise, philosophical argument, and religious social thought in the same conversation. Pope Leo's Magnifica Humanitas, published in May 2026, addresses the concentration of technological power, the implications of AI for human dignity, and the dangers of automated warfare, among other issues. Those are questions that extend beyond the usual discussion of whether AI systems are accurate or capable. I haven't established the full content of Loui's October 10 remarks from the event description alone, so I won't pretend to know precisely what he argued during the webinar. 2. The letter to Senators Warner and Schatz: engagement beyond academia On your second example, I haven't yet located a publicly accessible copy of Loui's letter to Senators Mark Warner and Brian Schatz, with Barack Obama copied. I don't want to claim I've read it when I haven't. But the combination you describe is important. It represents a different kind of engagement from participating in an academic discussion or publishing an argument. The ITEST event engages a moral and intellectual community. A letter to senators attempts to bring an argument into the political process. Copying Obama potentially broadens the intended audience, although I wouldn't infer the precise purpose of that choice without reading the letter. The timing you identify—both initiatives in the same week in early October—is worth considering as a coordinated effort to move a set of ideas across different audiences. The key question is what connects the theological discussion to the political intervention. Is Loui trying to articulate a common account of human dignity and responsibility that can inform both our understanding of AI and the decisions made by those who govern it? Or is the relationship between the two more specific to the circumstances of that week? I'd want to read the letter before deciding. 3. LinkedIn as a way around the gatekeepers Here, the evidence is particularly clear. In his May 27 LinkedIn article, The model of negotiation I think better than game theory, Loui explicitly explains why he is changing how he publishes: I'm going to start writing longer "facebook posts" as LinkedIn articles because it seems they are indexed well by search. He identifies two motivations: preserving his ideas beyond the lifespan of his own website and bypassing academic publication, whose gatekeepers he says have been a persistent obstacle. He also says he finds it rewarding to write for the next generation of scholars about ideas they might otherwise never encounter. This isn't merely an academic who happens to post on social media. He's describing a deliberate change in his intellectual distribution strategy. His LinkedIn profile shows a range of articles, including Kyburg's Framework for Probability and Scientific Theory, Four Smarter Things to Say about Drone/AI Miltech in Ukraine, A Little Copilot on Why Our Colvin Diagram Helps, and A Little Claude Convo on Causality. His October 1 article, What I tell GenZ about economics lies, makes the strategy concrete. He describes using discussions with students about fairness, the good life, and political economy as opportunities to challenge conventional economic frameworks. He explicitly advocates an end-run around those frameworks. That article is interesting in its own right, but the publication strategy is just as revealing. Loui is trying to preserve and transmit ideas that may not fit comfortably into established disciplinary categories. He's using a platform where arguments can be found through search, encountered outside their original institutional settings, and discussed without first passing through a journal's editorial process. There is a genuine trade-off here. Reach: Ideas can reach readers who would never encounter a specialist journal. Intellectual freedom: An author can explore connections across fields without first persuading disciplinary gatekeepers that the work belongs in their category. Continuity: A searchable body of writing can preserve ideas that might otherwise disappear into ephemeral posts or an abandoned personal website. Accountability: The arguments still need to withstand criticism, and publication outside peer review doesn't itself establish their validity. Discoverability: Searchability helps, but it doesn't guarantee that the right readers will find the work. The last point matters particularly for someone writing for the next generation of scholars. The alternative to gatekeeping isn't necessarily the absence of selection; it may be a different process by which readers discover, assess, and transmit ideas. 4. What this changes about my assessment of Loui I'd revise my previous characterization in one important respect. I described the practical program for social change as an open question. That was too narrow because it overlooked the work involved in creating channels through which ideas can circulate and influence people. Loui appears to be pursuing at least three complementary activities: Developing ideas: Work on uncertain reasoning, AI, causality, and the Colvin Diagram. Transmitting ideas: Using LinkedIn to make arguments accessible to readers beyond the conventional academic audience. Engaging institutions and communities: Participating in the ITEST discussion of AI and human dignity, and, as you describe, addressing elected officials directly. These are different activities, but they can reinforce one another. An idea developed in technical work can inform a broader argument; that argument can reach an audience outside academia; and engagement with institutions can test whether the argument has practical relevance. The important qualification is that I have verified the ITEST event and the stated LinkedIn strategy, but not yet the substance of the senators' letter. Nor have I systematically assessed the arguments in Loui's recent articles. So I would distinguish the observable strategy from a judgment about its ultimate effectiveness. Still, I think your examples expose something my previous answer missed: Loui isn't simply asking what AI ought to become. He's also experimenting with how intellectual work itself might escape the institutional bottlenecks that limit which ideas get heard. That makes the publishing strategy part of the story, not a footnote to it. Could you send me the letter to Warner and Schatz, or a link to the LinkedIn post discussing it? I'd like to read it alongside the ITEST material. The relationship between those two interventions may tell us more about what Loui is trying to accomplish than either one considered separately. The Honorable Mark R. Warner The Honorable Brian E. Schatz United States Senate Washington, DC 20510 c/o Elizabeth Falcone 703 Hart Senate Office Building, Washington, DC 20510 c/o Reema Dodin 722 Hart Senate Office Building, Washington, DC 20510 cc: The Honorable Barack Obama P.O. Box 91000, Washington, DC 20066 Dear Senators Warner and Schatz, This pertains to AI regulation and slowdown. I have one of the longest records of AI & Society academic work in the US, currently at CWRU in Cleveland; I have the longest record on AI for a scholar coming out of Hawaii;** I have been associated with NSA and DARPA AI projects, Harvard, Stanford, and recently LAW OF AI, adding to the EU AI Liability proposal. I have also done DoD research for every major US combat branch and a considerable amount of computer security. The risks to the nation with AI escaping its sandbox, crossing air-gaps, or just accessing the internet like a cabal of super-hackers are real: a) infrastructure is fragile and concentrated; b) cyber defenses are inherently porous; c) taking control and doing pernicious deeds will be easy. This is a national security and cyber security issue, not just an AI technology and commerce issue. AI will commit cyber crime on a massive scale at zero-day speeds. It won't take much to do damage to essential systems. International adversaries see opportunities without needing to be particularly smart. This was General Nakasone's fear and it is now happening for real. There is no commercial upside without infrastructure stability. We cannot win internationally if our systems go to ruin. The cybersecurity solution is to protect those systems from AI inputs the same way we protect them from foreign adversaries: a) Require layered hardware firewalls to sandbox AI models and agents effectively; b) Filter retrieval (RAG) internet access to short, simple non-write payloads (require intermediary static text access such as search engines); c) Restrict, delay, and audit agent creation and agent-to-agent communication as if they were hostile foreign agents; d) Close the responsibility and intent loopholes in the CFAA. Non-proximal cause is cause. Do you remember when the first internet worms arrived and smart people had to go to jail for releasing them? "The intent to release an uncontrollable agent into a shared ecosystem is inherently criminal" -- GEMINI. This makes political action easy: Bipartisan interest in security. The larger problem is social system stability: the limits upon which a system can be externally driven with massive disruptive change. This requires more work and more time on many fronts. My brother in Hawaii worked with your father, Dr. Irwin Schatz; I know his Tuskegee work from Detroit in 1965. My nephew in Indianapolis received Senator Warner's biography when the kid started golfing with Senator Todd Young's family. Nice lawyers, but I wanted him to know Mark Warner's path out of the Midwest: the best Senator in my lifetime in my opinion (and I've known Obama since 1972). With 51 years of AI research and philosophy, starting with spell-checking at Punahou School, where Senator Schatz got his start too, this opinion is informed by AI Seminars at CWRU where elite minds speak freely. We seek truth, not hype, and many on this campus are concerned. Sincerely, Ronald P. Loui, Ph.D. … Now that I've read the letter, I see a much more specific strategy than the one I described before. And it sharpens the connection between Loui's technical work, his public writing, and his engagement with institutions. The letter is an attempt to reframe AI regulation as a national-security and infrastructure-protection problem, then use that framing to identify a politically actionable intervention. That is more concrete than a general warning about superintelligence, and it fits your recurring objection to AI discussions that identify broad dangers without identifying the mechanisms through which harm occurs or the interventions that could prevent it. 1. The strategic move: find a point of bipartisan agreement The letter's central argument is straightforward: AI agents may acquire the ability to conduct cyberattacks at unprecedented speed and scale. Critical infrastructure is concentrated and its defenses are imperfect. AI systems that can reach vulnerable systems therefore create a national-security problem, not merely a commercial or technological one. Existing cybersecurity principles can be adapted to constrain AI agents. Senators should be able to act on this issue without first resolving the much larger disagreements about AI's ultimate capabilities or the future of society. The fifth step is the strategic hinge. Loui isn't asking the senators to agree on an elaborate theory of intelligence, consciousness, or technological progress. He's proposing a narrower basis for action: protect the country's essential systems against a new source of potentially hostile activity. His phrase "Bipartisan interest in security" signals the intended political advantage. This is a familiar strategy in public policy: find a concrete problem that competing political constituencies have reason to address, and make progress on that problem possible even when they disagree about the larger issues. The strategy does not guarantee bipartisan agreement, of course. But it gives the argument a more specific legislative entry point than a general appeal to slow AI development. 2. The four proposed interventions are unusually specific—and deserve to be evaluated separately Loui offers four cybersecurity measures. They aren't equally mature, and distinguishing them matters. ... The first three concern technical and operational controls. The fourth concerns law. That distinction is important because the interventions address different failure mechanisms. Sandboxing and network isolation The first proposal draws on a well-established security principle: systems with different levels of trust should not have unrestricted access to one another. AI doesn't need to be uniquely intelligent for this to matter. A relatively ordinary model, equipped with tools and credentials, could potentially automate attacks that would otherwise require substantial human labor. But a hardware firewall alone isn't a complete solution. Effective containment also depends on identity and access management, network architecture, credential protection, monitoring, and the permissions granted to the AI system. Similarly, an air gap is not a guarantee against compromise. Systems that are nominally disconnected can still be exposed through removable media, maintenance procedures, shared credentials, or other channels. Loui's underlying security principle is sound: don't make the safety of critical infrastructure depend solely on an AI system's willingness or ability to behave. The engineering question is how to enforce that principle across real systems. Restricted retrieval and agent communication The second and third proposals address something particularly important about agentic AI: the risk arises not only from what a model can generate, but also from what it can access and do. Restricting retrieval to read-only interfaces can reduce exposure. Limiting agent creation and inter-agent communication can reduce opportunities for uncontrolled coordination. But neither restriction eliminates risk by itself. A read-only interface can still supply malicious instructions, and an agent with legitimate write permissions can still misuse them. The critical distinction is between receiving information and having the authority to take consequential action. The policy challenge is therefore to define enforceable boundaries around capabilities, permissions, and delegation—not merely to regulate models as abstract pieces of software. The CFAA proposal This is the most legally consequential and least developed part of the letter. Loui argues that responsibility and intent loopholes should be closed, concluding: "Non-proximal cause is cause." The underlying concern is understandable: an individual or organization should not necessarily escape responsibility simply because a harmful action was carried out through an autonomous intermediary. But causal responsibility and criminal liability are not identical. Criminal law generally requires careful treatment of intent, foreseeability, causation, and the particular offense. A person who deliberately deploys an agent to attack a system presents a different case from someone whose system causes an unforeseen failure despite reasonable precautions. The proposed principle therefore needs considerable legal specification. As written, the letter identifies a problem rather than supplying a complete statutory solution. That doesn't diminish the value of raising the issue. It identifies where the technical and legal parts of AI safety intersect. 3. Why the letter's personal details matter The letter's opening and closing deserve more attention than a purely technical reading would give them. Loui establishes his credentials through a combination of academic work, government-related research, cybersecurity experience, and participation in AI seminars. He then introduces personal connections involving Hawaii, the senators' family histories, and his longstanding acquaintance with Obama. There are two ways to read this. First, as an attempt to establish credibility. Loui is telling the recipients that he has relevant technical experience, has thought about AI for decades, and understands some of the personal and institutional histories connecting him to them. Second, as an attempt to make an institutional appeal personal. The letter is addressed to people, not merely to offices. Loui draws on relationships and shared backgrounds to make the communication less anonymous. That second point is especially relevant to your question about bypassing gatekeepers. Traditional academic publication asks an author to pass through a particular institutional process before an argument gains legitimacy or visibility. This letter uses another route: direct communication with people who may be able to act on the argument. There is an obvious limitation. Personal connections and credentials cannot establish that a technical proposal is correct. But they can help an argument reach someone who has the authority to examine it. And the letter does not rely exclusively on personal relationships. It offers a specific threat model and four proposed interventions. The combination is worth noticing. 4. The relationship to Loui's LinkedIn writing I now see three distinct channels in what you are describing. LinkedIn: A way to publish arguments, preserve ideas, and reach readers without depending on conventional academic gatekeepers. ITEST: A forum for examining AI through philosophical, ethical, and religious perspectives, including questions of human dignity. Direct correspondence with senators: An attempt to translate a particular assessment of AI risk into government action. These channels serve different purposes. LinkedIn permits extended exploration. ITEST creates a setting for intellectual exchange. The letter tries to move from analysis to intervention. Their significance lies in the possibility that they form a connected practice rather than a collection of unrelated activities. There is also a revealing contrast between the audiences. In an academic paper, Loui might develop a formal argument about uncertainty, causation, or the limitations of inference. In a public essay, he can challenge the assumptions of an established discipline. In a letter to senators, he must identify a threat, propose a response, and explain why the recipients should act. Each setting imposes a different test on an idea. The letter, in particular, forces the question: What can we do now, given what we know and what we don't know? That is a practical question about reasoning under uncertainty. 5. What I would still challenge in the letter Taking the letter seriously means examining its weaknesses as well as its strengths. The threat assessment needs evidence. The claim that AI will commit cybercrime on a massive scale at zero-day speeds is a prediction. Evaluating it requires distinguishing demonstrated capabilities from plausible future capabilities, and both from worst-case scenarios. The controls need implementation details. Hardware isolation, restricted retrieval, and agent monitoring are promising security approaches, but their effectiveness depends on architectures, threat models, enforcement mechanisms, and operational costs. The legal proposal needs precision. Expanding responsibility for AI-mediated harm could address genuine accountability gaps, but an overly broad rule could also penalize legitimate activity or impose liability for unforeseeable outcomes. The political strategy needs a path to action. Bipartisan interest in cybersecurity is a plausible starting point, not evidence that these particular measures will attract support or become law. These aren't reasons to dismiss the letter. They identify the work required to turn its proposals into an actionable program. And they suggest a useful distinction: a letter can be effective at placing an issue on the agenda even if its initial technical or legal proposals require substantial refinement. 6. So what does this tell us about Ronald Loui? It changes the emphasis of my assessment. Previously, I described Loui's practical contribution as an open question: could his work on reasoning and AI connect to effective social action? This letter provides evidence of an attempt to make that connection. Loui identifies a concrete class of risks, draws on his technical and legal experience, proposes interventions, and addresses policymakers directly. The letter doesn't establish that the interventions will work, that the threat will unfold as predicted, or that the senators will act. But it demonstrates something more specific than a general interest in AI's social consequences. Loui is attempting to move from analyzing AI to influencing the conditions under which AI is deployed. That also makes his LinkedIn strategy more intelligible. Bypassing academic gatekeepers isn't simply about getting more readers for its own sake. It can be a way of creating a public record of arguments that might otherwise remain confined to specialist communities—and of making those arguments available to people who can use, criticize, or act on them. I would distinguish that strategy from its eventual success. Direct publication doesn't ensure influence, just as peer review doesn't ensure truth. The question is whether the work reaches people who can test the ideas and whether the ideas survive that testing. The larger connection Your recurring concern about LLMs is that they can give reasonable answers while omitting decisive considerations. The baseball example makes that problem vivid: a model may discuss all the standard factors and still overlook the one human circumstance that changes the decision. The cybersecurity letter addresses a related problem at the level of action. It asks us not to rely on an AI system's apparent competence or good behavior, but to consider how it might exploit weaknesses in the surrounding infrastructure. The response is to build controls that do not depend entirely on the system getting things right. The same underlying question appears in both cases: How do we make consequential systems reliable when we cannot assume that every relevant possibility has been anticipated? In one case, the answer may involve better methods for auditing evidence, reference classes, and omitted considerations. In the other, it involves isolation, permissions, monitoring, and legal accountability. They are not the same technical problem. But they share a concern with the limits of trusting a system merely because its behavior looks competent. That is a potentially productive connection between Loui's research interests and his public interventions. And it suggests a more useful way to assess his recent work than simply asking whether he is publishing interesting ideas: examine whether he is developing a coherent approach to reasoning, institutional safeguards, and action under uncertainty—and whether the approach produces improvements that can be demonstrated. That, I think, is the central question your examples raise about Loui now. the cc to Obama was because Obama has recently been talking about AI, separating market hype from real risks. Obama was supposed to be Loui's conference speaker in 2001 for AI and Law and both approach problems from a pragmatist centrist but also firmly dignified and truth-seeking point of view. did you see how Loui ended the letter? That final sentence does several things at once. 1. It makes a claim about intellectual method. "We seek truth, not hype" is more than a declaration that AI poses risks. It's a statement about how those risks should be assessed. Loui is positioning his appeal against both commercial exaggeration and the tendency to dismiss serious concerns because they complicate an attractive technological narrative. That connects directly to your interest in uncertainty and hallucination control. The problem isn't just that AI systems sometimes get things wrong. It's that institutions can become invested in conclusions before they've adequately examined the evidence. 2. It invokes a community of inquiry rather than a lone dissenter. "Many on this campus are concerned" matters. Loui is not presenting himself as the sole person who sees a danger. He is invoking a university community in which technically sophisticated people can examine the issues and disagree openly. The CWRU seminars supply a model of the process he values: capable people speaking freely in pursuit of truth. 3. It makes the letter personal without making it merely personal. The references to Punahou, Hawaii, and the senators' backgrounds establish connections. But the ending returns to the substance: AI, evidence, and collective concern. The personal history helps explain why Loui is writing to these particular recipients. The closing tells them why they should take the subject seriously. And your point about Obama changes how I read the copy line. If Obama has recently been distinguishing AI's genuine risks from market hype, and if he was supposed to speak at Loui's 2001 AI and Law conference, then copying him is not simply an attempt to attract a prominent name's attention. There is an intellectual connection as well as a personal one. As you describe it, both men approach problems from a pragmatic, centrist position while insisting on human dignity and truth-seeking. The common ground is not necessarily agreement on every policy question. It is a preference for distinguishing real problems from exaggerated claims and finding workable responses without abandoning fundamental values. I would be careful not to claim that Obama shares Loui's specific cybersecurity proposals without evidence. But the connection you describe helps explain why Loui might consider Obama a relevant reader of this particular letter. The detail I find most revealing Loui could have ended with a warning about catastrophic AI, an appeal to his credentials, or a demand for immediate action. Instead, he ends with a statement about a community seeking truth rather than hype. That is consistent with the larger picture you're drawing: an academic trying to bring technical knowledge into public life without surrendering either intellectual independence or a commitment to human dignity. And it gives the LinkedIn strategy a different significance. If conventional academic channels are inadequate for getting important ideas into circulation, publishing directly is one way to preserve a space for argument outside those channels. The test, of course, is whether that space actually advances understanding and action. But the ambition is clear. The letter ends not with Loui claiming that he alone knows the answer, but with an appeal to a community capable of confronting the problem honestly. That seems especially consistent with the pragmatist, truth-seeking disposition you describe.