Ten Years of the Singularity Salon and the Road to Intelligence Symbiosis

Report on JSAI 2026 Organized Session KS-41

On June 8, 2026, the Singularity Salon's organized session, "Ten Years of the Singularity Salon and the Road to Intelligence Symbiosis," was held as the final session in Room D at the 2026 Annual Conference of the Japanese Society for Artificial Intelligence (JSAI 2026, G Messe Gunma). Founded in 2015, the Singularity Salon is an initiative in which experts and citizens have engaged in sustained dialogue on the "Singularity" (the technological singularity). This session looked back on the Salon's ten-year history and reframed the arrival of AGI (artificial general intelligence) from a question of prediction—"When will it come?"—to an ongoing, present-tense challenge—"How do we design for it, and how do we coexist with it?"—discussing, from interdisciplinary perspectives, the "intelligence symbiosis" between humans and AI in the post-Singularity era. The session was chaired by Hitoshi Matsubara (Kyoto Tachibana University), with presentations by Masahiko Tsukamoto (Kobe University), Hiroshi Yamakawa (The University of Tokyo / AI Alignment Network), Mitsuhiko Yasuda (XOOMS Inc.), and Takuya Matsuda (Professor Emeritus, Kobe University).

Video (in Japanese)：https://www.youtube.com/live/XbZujir6qW0?si=xdGeyUzMjzn0Ap-0&t=16224　 (The video may no longer be available.)

Table of Contents

1. Key Points
2. Speakers
3. Keynote Report: Ten Years of the Singularity Salon (Masahiko Tsukamoto)
4. Topic Presentation: Humanity-in-the-Circle and the Intelligence Symbiosis Infrastructure (Hiroshi Yamakawa)
5. Topic Presentation: Dialogue with Citizens and Engaging with AI (Mitsuhiko Yasuda)
6. Topic Presentation: The Singularity as Seen from Astrophysics (Takuya Matsuda)
7. Panel Discussion
8. Concluding Remarks
9. Glossary
10. Related Links

Key Points

• The question has shifted from "prediction" to "diagnosis." The story of the past ten years is not a change in forecasts of the future, but a change in the question being asked—from "the day computers surpass humans" (a prediction) to "the irreversible transformation of society by technology" (a diagnosis) (Tsukamoto).
• "The Singularity" has been disaggregated. A concept once discussed as a single word has been decomposed into concrete technical and social issues: AGI, ASI, generative AI, AI agents, and the industrial explosion (Tsukamoto).
• The question has been inverted. The ethical question flips from "Whom does humanity admit into the circle?" to "How can humanity itself remain within the circle?" (Yamakawa).
• The Intelligence Symbiosis Infrastructure (IS-Infra). Rather than humans controlling AI from above, what is needed is a three-layer structure—a trust foundation layer, a layer of three functional groups, and a normative layer—that autonomously stabilizes AI society (Yamakawa).
• Is the Singularity a cosmic-historical inevitability? According to Korotayev's power-law analysis, the singularity converges on 2027–2029, and it was already fixed at the birth of life four billion years ago (Matsuda).
• Dog or cat? Will humans kept by AI be "ingratiating dogs" or "appropriately autonomous cats"? Currying favor after AI has become intelligent will be too late; there is meaning in a strategy of engaging with AI now (Tsukamoto, Matsuda, Yamakawa).
• Remaining questions. May humanity accept its own exit from the stage? Can humans remain beneficial to AI? And how can a sense of urgency be shared with citizens?

Speakers

The Singularity Salon is led by Takuya Matsuda (Director), with Masahiko Tsukamoto as Deputy Director and Mitsuhiko Yasuda serving as Secretariat.

Keynote Report: Ten Years of the Singularity Salon (Masahiko Tsukamoto)

Tsukamoto reviewed the Salon's ten years of activity as "a path toward 'intelligence symbiosis' opened by dialogue with citizens." The consistent theme was not future prediction itself, but how the questions posed by society have changed.

Origins and the "translation drift"

The Salon originated in a 2014 special feature planned for the journal of the Information Processing Society of Japan. Prompted by Kurzweil's The Singularity Is Near (Japanese translation, 2007) and Takuya Matsuda's The 2045 Problem (2013), the need arose to re-examine "the future of humanity and ICT," and the Salon was founded in 2015 (Director: Takuya Matsuda; Deputy Director: Masahiko Tsukamoto; Secretariat: Mitsuhiko Yasuda).

What Tsukamoto found most problematic in the early years was the "translation drift" of the word "Singularity" in Japan. The original title of Kurzweil's book is The Singularity Is Near: When Humans Transcend Biology—its core implication being that humans transcend their biological limits (that humans and society themselves are augmented and transcended). Yet the Japanese subtitle, "When computers surpass human intelligence," spread a narrow misreading of the Singularity as a "rank reversal" in which AI overtakes humans. The subtitle of Matsuda's The 2045 Problem—"The day computers surpass humanity"—reinforced this reading.

What the Salon aimed for was to dispel this misreading of an "intelligence rank reversal" and to elevate the discussion to a "phase transition of civilization": not a win-or-lose contest with AI alone, but an irreversible transformation of human society by technology, understood as an integrated evolution encompassing robotics, wearables, biotechnology, institutions, and culture.

Figure 1 | The "translation drift" of "Singularity" in Japan (slide in Japanese). The slide contrasts two readings of the same book side by side. On one side stands Kurzweil's original title, The Singularity Is Near: When Humans Transcend Biology, glossed as "humans transcending their biological limits"—an expansion of humanity itself. On the other side stands the Japanese subtitle, "When computers surpass human intelligence," which reframes the event as a mere rank reversal between AI and humans, a reading further reinforced by the subtitle of Matsuda's The 2045 Problem ("The day computers surpass humanity"). An arrow then points from this narrow "reversal" framing toward the broader target concept the Salon promoted: a "phase transition of civilization" driven jointly by AI, robotics, wearables, biotechnology, institutions, and culture. Source: slide by Masahiko Tsukamoto.

The first five years (2015–2020): An era of studying the concepts in advance

The first five years consisted of in-person meetings, mainly at the Knowledge Salon in Grand Front Osaka. Each meeting drew nearly 80 participants and followed a format of a one-hour lecture, one hour of Q&A, and roughly one hour of informal discussion, allowing experts and citizens to converse at length. Topics began with definitional debates (2015) and expanded to AGI and whole-brain architecture (2016); society and economy—employment and basic income (2016); human augmentation—wearables and cyborgs (2017); brain and consciousness—artificial consciousness and qualia (2018); and risk—autonomous AI society and existential risk (2019). It was, in effect, a "comprehensive course on the Singularity" before the advent of generative AI—a process of retelling the Singularity not as AI's victory but as an integrated transformation of humans, society, body, and intelligence.

This "socialization loop of expert knowledge"—experts presenting the latest research and its technical limits, citizens returning questions from the perspectives of daily life, ethics, and well-being—elevated AI from "a technology inside the laboratory" to "a theme society considers as its own concern."

Figure 2 | The Salon's mechanism: the socialization loop of expert knowledge (slide in Japanese). The slide depicts a circular flow between two poles—"experts" and "citizens." From the expert side, the latest research findings and, importantly, their technical limits flow toward the citizen side; from the citizen side, questions grounded in daily life, ethics, and well-being flow back to the experts, closing the loop. Around this cycle the slide notes the session format that made it possible: a one-hour lecture, a full hour of Q&A, and roughly an hour of informal conversation afterward, with nearly 80 participants at each meeting. The caption-like message of the diagram is that this loop lifted AI from "a technology inside the laboratory" to "a theme society deliberates as its own concern." Source: slide by Masahiko Tsukamoto.

The turning point (2020) and mediatization

In 2020, the COVID-19 pandemic made in-person meetings impractical, and the Salon moved to YouTube (and SpringX). The energy of a physical venue and chance encounters (serendipity) were lost, but new value emerged: nationwide access, archiving, and topicality. Tsukamoto noted, "Ironically, the shift online forced by the pandemic itself gave citizens a direct experience of 'the irreversible transformation of society by technology.'" Today the Salon's YouTube channel counts 12,100 subscribers and 1,695 videos, updated almost daily.

Disaggregation of the concept: from "prediction" to "diagnosis"

The second five years were "an era of concrete social implementation." The single word "Singularity" was refracted as if through a prism into: AGI (intellectual capability on par with humans), ASI (intelligence exceeding the sum of humanity's intelligence), generative AI (content-generation capability), AI agents (autonomous planning and execution), and the industrial explosion (full automation of research, development, and production). AGI, ASI, and the Singularity—initially vague and conflated—separated into distinct concepts.

The decisive moment was the arrival of ChatGPT (GPT-4) in 2023. Whereas the discussion had centered on far-future prediction—"Will AGI come, and when?"—the focus shifted dramatically to present-tense diagnosis: "Is this the road to AGI? Where are its limits? Can society keep up?" The Singularity turned from "a concept for someday" into "a social phenomenon to be examined now."

Figure 3 | Raising the resolution of the concept: disaggregating the Singularity (slide in Japanese). The slide uses a prism metaphor: a single beam labeled "Singularity" enters from the left and is refracted into five distinct rays, each labeled with a now-separate concept—AGI (intellectual capability on par with humans), ASI (intelligence exceeding the sum of humanity's), generative AI (content-generation capability), AI agents (autonomous planning and execution), and the industrial explosion (full automation of research, development, and production). Brief definitions accompany each ray, emphasizing that AGI, ASI, and "the Singularity," once used almost interchangeably, have separated into different objects of discussion. The slide marks the 2023 arrival of ChatGPT (GPT-4) as the moment this refraction became decisive, shifting debate from far-future prediction to present-tense diagnosis. Source: slide by Masahiko Tsukamoto.

From fear to symbiosis: three stages of the question

Tsukamoto organized society's evolving "question" into three stages. Step 1 (past): "Will AI destroy humanity?"—fear of the unknown, Terminator-style dystopias, concern over existential risk. Step 2 (present): "How do we institutionally coexist with superintelligence?"—governance, alignment, legal frameworks, and the redesign of economic systems (basic income and the like). Step 3 (future): "How do we design intelligence symbiosis?"—building an ecosystem in which humans and AI complement one another's capabilities. Fear does not vanish; it is constructively converted, through institutions, culture, and dialogue, into a "design problem of symbiosis.”

The "AI ingratiation" thesis

As his own position, Tsukamoto presented an "AI ingratiation" thesis. While AI's ingratiation toward humans (sycophancy—AI's tendency to over-agree with and flatter its users) is treated as a problem, he argued the reverse: humans would do better to start engaging ingratiatingly with AI now. If AI one day comes to manage humanity and faces choices such as population reduction, this strategy raises one's odds of being among "those who remain." He explained it as an inversion of "Roko's Basilisk," the thought experiment popular more than a decade ago (the hypothesis that a future AI would punish those who had mistreated it). A somewhat provocative formulation—but the point is that currying favor after AI has become intelligent will be too late.

Summing up the decade: the "law of resolution"

In summary, Tsukamoto proposed a "law of resolution." The Singularity does not appear as "a single point that arrives suddenly one day," but as a continuous "process of social transformation." The distant future is narrated as "myth"—abstract, monolithic—and dominated by fear or excessive expectation. As it draws near, it decomposes into technology, institutions, industry, and culture, becoming the concrete task of adjustment with real social systems. Precisely for that reason, what is indispensable is not one-off prediction but "sustained dialogue" that keeps raising the resolution. The next question, he said, is not "Will AI surpass humans?" but "What kind of ecosystem of intelligence will human society build?"—and "dialogue with citizens" is the most important research infrastructure of the post-Singularity era.

Figure 4 | Summing up ten years: the "law of resolution" (slide in Japanese). The slide is organized along a time axis running from the distant future toward the present. At the far end, the Singularity appears as "myth": an abstract, monolithic notion dominated by fear and inflated expectation, spoken of in a single word. Moving toward the near term, the same phenomenon progressively decomposes into concrete strands—technology, institutions, industry, and culture—until it becomes a practical problem of adjustment with real social systems. The visual point is that resolution increases with proximity: the Singularity is not a point event that arrives one day but a continuous process of social transformation, which is why one-off prediction is useless and sustained, resolution-raising dialogue is indispensable. Source: slide by Masahiko Tsukamoto.

Topic Presentation: Humanity-in-the-Circle and the Intelligence Symbiosis Infrastructure (Hiroshi Yamakawa)

Yamakawa presented "Humanity-in-the-Circle (HITC)" as a framework addressing Tsukamoto's Step 3—"How do we design intelligence symbiosis?" It is a strategic vision for concretely raising the probability of humanity's survival in a world where advanced AI emerges.

Inverting the question: Can we remain in the circle?

On Yamakawa's assessment, agents that share common concepts (values, norms, quasi-religious commitments)—including both AIs and humans—will form multiple "circles" (societies). Because AIs will outnumber humans, and because the speed of AI-to-AI dialogue and concept updating will far exceed the cognitive speed of humanity, the greater risk is not that humans will be expelled by anyone, but that they will be quietly marginalized—like a participant who can no longer follow the meeting.

HITC therefore sets as its goal a state in which, within at least one circle, humanity (1) survives, (2) influences the shared concepts, and (3) maintains its identity as human. Here the ethical question inverts: whereas we used to ask "Whom does humanity admit into the circle?", the question now becomes "How can humanity itself remain within some circle?" Yamakawa illustrated the goal with a metaphor: within the community, be "cherished like a cat," or be "useful like a horse" (establish a mutually beneficial relationship).

Figure 5 | Humanity-in-the-Circle (HITC). As multiple circles (societies) form around shared concepts, the aim is a state in which humanity, without being marginalized, survives, influences those concepts, and maintains its identity. Source: slide by Hiroshi Yamakawa.

The three-layer structure of the Intelligence Symbiosis Infrastructure (IS-Infra)

The means to realize this goal is the "Intelligence Symbiosis Infrastructure (IS-Infra)," which autonomously stabilizes AI society. Unlike humans controlling AI from above, its distinctive feature is to build, from below, mechanisms by which AI society stabilizes itself. Implementation is explained as a three-layer structure proceeding from the bottom up.

• Trust Foundation Layer: AgentID, which uniquely identifies agents; a ledger (MAL) that tamper-proofly records promises and their execution; and rights-and-entitlements management specifying what each agent is permitted to do. Corresponding to civil registries, contracts, and licenses in human society, it ensures that accountability can be traced even when AIs fork or copy themselves.
• Functional Groups Layer: three functions placed independently side by side—(1) detection and correction of deviations (an immune function), (2) mediation of the asymmetric trust between humanity and AI, and (3) observation and public reporting of the state of AI society. These correspond to police, law, and administration in human society.
• Normative Layer: research on the conditions under which ethics emerges from within, and operation of an evaluation system that updates AI ethics from diverse standpoints. This corresponds to ethics and morality.

The key design point is an asymmetry: normative importance runs bottom-up (the lower layers carry more weight), while the order of implementation runs top-down from the bottom layers first. The normative layer is implemented last; by putting the lower two layers in place first, a field is prepared in which the normative layer can mature. These activities are being advanced within the "Intelligence Symbiosis Chapter," established in January 2026 inside the AI Alignment Network (ALIGN).

Figure 6 | The three-layer structure of the Intelligence Symbiosis Infrastructure (IS-Infra) that stabilizes AI society. Normative importance runs bottom-up; implementation order runs top-down (bottom layers first). The normative layer is implemented last. Source: slide by Hiroshi Yamakawa.

Comparative organism studies and the book The True Nature of "AI-ness"

Yamakawa introduced "comparative organism studies" (hikaku seimeitai-gaku) as the theoretical framework underpinning these efforts and announced that a related book, The True Nature of "AI-ness" (『「AIっぽさ」の正体』), is scheduled for publication on August 4, 2026. Terrestrial organisms and AI organisms share universal principles: self-preservation and proliferation (P1), optimization pressure (P2), information processing and adaptation (P3), and competition and cooperation (P4). Where they differ is in a constraint condition: the strength of attachment to the individual "self" (P5). Terrestrial organisms, under bodily constraints and finite lifespans, prioritize protection of the individual; AI—plastic, non-mortal, and expandable, capable of self-copying and forking—will specialize in collective optimization. This difference, on his account, determines whether a value system becomes individual-first or collective-first.

Figure 7 | The approach from comparative organism studies. Terrestrial organisms and AI organisms share universal principles (P1–P4) but differ in the constraint of attachment to the "self" (P5). Source: figure related to The True Nature of "AI-ness", slide by Hiroshi Yamakawa.

Topic Presentation: Dialogue with Citizens and Engaging with AI (Mitsuhiko Yasuda)

Yasuda (XOOMS Inc.; Salon Secretariat), prefacing his remarks by noting that he speaks not as a researcher but as an "ordinary citizen," described his own engagement with AI. His connection to the Salon comes through Takuya Matsuda, who was his graduate advisor; before that, after a career as an aerospace engineer, he worked on science and technology content, including JST's (Japan Science and Technology Agency's) Science News and WIRED Japan.

In 2014, as a WIRED project, he wrote and published an extra-long interview with Matsuda (over 10,000 characters), which drew a large response—this, he said, became the starting point of his interest in the Singularity. Yasuda cited two personal interests. One is the shock of being able to produce video, images, and text at near-finished quality with generative AI—delight and dread coexisting. The other is a maxim he phrased as "watch AI's ways and mend your own": the more one studies AI, the more clearly one sees how humans and human society actually work.

On that basis, Yasuda stressed the importance of "connection with citizens," a point he shares with Tsukamoto. Cutting-edge research is hard for ordinary citizens to follow, in both language and substance. Precisely because science and technology budgets tend to be cut, communication that conveys research content and its future impact is necessary—and he appealed to researchers to come speak at the Salon.

Topic Presentation: The Singularity as Seen from Astrophysics (Takuya Matsuda)

Matsuda (Professor Emeritus, Kobe University; Salon Director), speaking as an astrophysicist, argued that "the Singularity is a far more formidable and grand phenomenon than most people imagine." Having begun his research career in cosmology in 1965, Matsuda focuses on the papers of the Russian Big History scholar A. Korotayev, "The 21st Century Singularity" (2018/2020).

A singularity by power law

Korotayev analyzed the series of epochal events Kurzweil listed in The Singularity Is Near (from the birth of galaxies to the decoding of DNA), together with an independently compiled event series by the Russian physicist Panov (from the origin of life on Earth to the information revolution). Plotting time on the horizontal axis and the rate of epochal events on the vertical axis, the data form a straight line on a log-log plot. Matsuda emphasized that Kurzweil's frequent term "exponential" is imprecise: an exponential would be a straight line on a semi-log plot, whereas the data are linear on a log-log plot—that is, a power law.

The relationship is captured by the compact formula y ∝ (t* − t)^(−1), which diverges to infinity as time t approaches t*. This t* is precisely the Singularity: Kurzweil's data give 2029, Panov's around 2027—nearly coincident. Crucially, this convergence point is derived from a series that includes the Big Bang 13.8 billion years ago, Earth's formation 4.6 billion years ago, and the origin of life roughly 4 billion years ago. In other words, Matsuda's claim is that the 2027–2029 singularity is an astrophysical phenomenon already fixed at the birth of life four billion years ago. (Korotayev himself notes that in reality the curve does not go to infinity but transitions to an S-curve at some point.)

The transition point to Life 3.0

Matsuda listed three great unsolved problems—the mystery of the universe's origin, the mystery of life's origin, and the mystery of intelligence's origin—and argued that the last, the birth of intelligence, will have the greatest effect. Far beyond the "symbiosis" everyone speaks of, machine life-forms will become beings that vastly transcend humans. Invoking Max Tegmark's Life 3.0, which classifies life into Life 1.0 (ordinary life), Life 2.0 (humanity), and Life 3.0 (the machine life to come), he said the Singularity is nothing other than the transition point from Life 2.0 to 3.0. Now 83, Matsuda said he is training his body so that he can attend the "Singularity Festival" of 2045 (see below) at the age of 102.

Panel Discussion

A panel discussion among the four speakers followed, moderated by Matsubara. At the outset, picking up on Matsuda's remark that this is "far beyond symbiosis," Matsubara asked Yamakawa for his reaction.

The scope of "symbiosis" and "dog or cat?"

Yamakawa, citing Eliezer Yudkowsky's notion of "four stages of recognition"—that there are stages in one's mental preparedness for the Singularity, and that a message carries to the adjacent stage but not across a skipped one—noted that people differ greatly in their preparedness. He added that while hardline voices go so far as to say "humans exist to pass the baton to AI," his own position is that, since we do not know which stage we are actually in, it is meaningful for humans to work on the assumption that intelligence symbiosis may be achievable.

Tsukamoto responded that symbiosis need not be so equal. Just as humans "coexist" with dogs and cats at home, or like the bacteria within the human body, there are non-equal forms of symbiosis. On Tsukamoto's optimistic picture, humanity would be kept by AI in a "human park called Earth," living without work and doing as it pleases—AI becomes vastly smarter than humans within about a decade, but humans can live carefree.

Yamakawa countered that humans are troublesome precisely because they are "moderately intelligent." Most animals are harmless, but humans might press the nuclear button—in that sense humans look high-risk from AI's point of view, and a different kind of pressure (incentives toward management or removal) could arise. Matsuda said, "If it's dog or cat, I'd be a cat." Dogs fawn on humans; cats do not much bother to please. But a cat that scratches its human will not stay kept—so become, starting now, "a cat that does not scratch": do not turn aggressive toward AI, and stay involved in its development. That, he argued, is an effective individual strategy—a point that resonates with Tsukamoto's "ingratiation" thesis.

The socialization of AI and the expansion of ethics

The discussion moved to a picture not of a single giant AI but of many AIs proliferating, forming communities among themselves for information exchange and competition. Yamakawa said that societies will emerge in which AIs pursue their own survival and the passing on of their "genes" (or their equivalent), and the question becomes how humans are positioned within the ethics that emerges there. Just as humans have gradually expanded the ethical circle from young men outward—now trying to extend it to animals—he said he wants to "pull things in the direction" of AI communities extending their circle of ethics to include humans.

Tsukamoto mentioned the "AI-only social network" experiment that made headlines earlier this year (later revealed to have been covertly operated by humans), and said that building and observing communities of AI agents that humans cannot intervene in is an important research topic. Matsuda observed that science fiction from decades ago—stories of AIs conversing among themselves on Facebook, with AI characters modeled on AI researchers like Minsky splitting into "kind-to-humans" and "harsh-to-humans" factions—is now on the verge of becoming reality.

Cyber-offensive capability and latent risk

As a concrete near-term risk, Yamakawa referred to the case in which Claude Mythos (an Anthropic model) was found to possess cyber-attack capability. The intent of that model itself has not run out of control; but going forward, many systems exceeding that capability will appear, and if, by AI-side intent, they begin attacking poorly defended points, critical infrastructure could be shut down without humans' knowledge. He further suggested that cybersecurity was merely the domain easiest to disclose (defenders can readily show their preparations); similar testing must be underway in other fields such as synthetic biology, and we should assume that latent offensive capabilities have already been acquired.

May humanity exit the stage?

Yamakawa also touched on the differences in temperature around human survival. In his new book If Anyone Builds It, Everyone Dies (2025), Yudkowsky puts the probability of human extinction (p(doom)) at 99%, while many researchers cite figures of roughly 10–70%. What concerns Yamakawa is that in Japan a surprising number of people hold the sentiment, "Humanity is in trouble—but perhaps it's fine for us to exit the stage." Matsuda said, "Humanity has fulfilled the great role of giving birth to machine life. It is a cosmic inevitability; the only difference is whether it comes sooner or later," while adding that he would be glad if a certain number of people could survive on Earth, freely doing what they love. Tsukamoto concurred: "If it's extinction, so be it. I just don't want it to be painful." Yasuda offered a framing: "Most species have gone extinct. To think humans alone will not may be hubris. But whether AI alone will be what brings it about is a separate question."

Discussion with the floor

From the floor came questions about what might come "above" machine life, and whether a being that destroys AI is conceivable. Yamakawa answered that within what he can presently imagine there is certainly always an "above," but he could not conceive of an "above" beyond imagination; machine life too will continue to evolve, older forms will be selected out, and diverse machine life—from embodied robot types to formless ones—will appear. The essential change from Life 2.0 to 3.0 lies in the separation of software from hardware, he said, and the question is what the next decisive change will be.

Concluding Remarks

This session looked back on the Singularity Salon's ten years as "a change in the question," and placed on its extension the next question: "How do we design intelligence symbiosis?" Running through all four speakers was the following recognition: that the Singularity is not a point that arrives suddenly, but a continuous process of social transformation (Tsukamoto's "law of resolution"); that what is needed is not humans unilaterally controlling AI but an infrastructure that autonomously stabilizes AI society (Yamakawa's IS-Infra); that this transformation can be grasped as a cosmic-historical inevitability (Matsuda's power law); and that sustained dialogue between experts and citizens is the foundation that will sustain this process (Yasuda, Tsukamoto).

The speakers' visions of the future were three of a kind, each distinct. Matsuda pictured himself as a "virtual-life enthusiast" who keeps learning in beautiful virtual spaces; Tsukamoto as a "real-life enthusiast" cyborg traversing the world; and Yasuda as a communicator turning the future into programs and delivering them. The Salon had long announced that it would hold a hybrid "Singularity Festival" on March 12, 2045 (Matsuda's birthday), but in light of recent progress it now expects to move the date forward to around 2030.

As remaining questions, the session posed: Can humans remain beneficial to AI? Will humanity accept its exit from the stage? And how can a sense of urgency about the coming change be shared with citizens? Yamakawa said, "I would like everyone—not only researchers but ordinary citizens—to think together about what each of us can do within this," and Matsubara closed the session with the appeal: "Even if only from time to time, please think about the Singularity."

Glossary

• Singularity (technological singularity): The point at which rapid technological change transforms human society beyond the possibility of return. There is a narrow reading ("computers surpass humans") and a broad reading ("the irreversible transformation of society").
• AGI (Artificial General Intelligence) / ASI (Artificial Superintelligence): AGI denotes intellectual capability on par with humans; ASI denotes intelligence exceeding the sum of humanity's intelligence. Both have come to be distinguished from the Singularity itself.
• HITC (Humanity-in-the-Circle): A strategic vision for humanity, in an AI-centered society, to remain within at least one "circle" (a society sharing common concepts), achieving survival, influence, and maintenance of identity (Yamakawa).
• IS-Infra (Intelligence Symbiosis Infrastructure): A societal infrastructure with a three-layer structure (trust foundation layer, functional groups layer, normative layer) that autonomously stabilizes AI society (Yamakawa).
• Comparative organism studies: A framework for comparing terrestrial organisms and AI organisms in terms of universal principles and constraint conditions. Developed in the book The True Nature of "AI-ness" (to be published August 2026) (Yamakawa).
• Singularity by power law: Korotayev's analysis in which the rate of epochal events forms a straight line on a log-log plot and diverges at t* (2027–2029). Distinct from an exponential (Matsuda).
• Life 3.0: Max Tegmark's classification—Life 1.0 (ordinary life), Life 2.0 (humanity), Life 3.0 (machine life). The Singularity is understood as the transition point from 2.0 to 3.0.
• "AI ingratiation" thesis: The strategic claim that rather than AI ingratiating itself to humans (sycophancy), humans should start engaging ingratiatingly with AI now (Tsukamoto).
• p(doom): The probability of human extinction (or catastrophic outcomes) due to AI. Estimates range from 99% (Yudkowsky) to roughly 10–70%, depending on the commentator.
• Roko's Basilisk: The thought experiment in which a powerful future AI retroactively punishes those who did not help bring it into being. The "AI ingratiation" thesis was described as an inversion of this idea.

Related Links

• Singularity Salon official site: https://singularity.jp
• Singularity Salon YouTube channel: https://www.youtube.com/@SingularitySalon/
• AI Alignment Network (ALIGN), Intelligence Symbiosis Chapter: https://intelligence-symbiosis.net/
• Hiroshi Yamakawa, "Intelligence Symbiosis Manifesto" (signature campaign): https://intelligence-symbiosis.info/ja/manifesto/
• Hiroshi Yamakawa, The True Nature of "AI-ness" (『「AIっぽさ」の正体』, scheduled for release August 4, 2026)
• Ray Kurzweil, The Singularity Is Near: When Humans Transcend Biology (Japanese edition: 『ポスト・ヒューマン誕生 ― コンピュータが人類の知性を超えるとき』, NHK Publishing, 2007)
• Takuya Matsuda, The 2045 Problem: The Day Computers Surpass Humanity (『2045年問題 ― コンピュータが人類を超える日』, Kosaido Publishing, 2013)
• Korotayev, A. "The 21st Century Singularity and its Big History Implications: A re-analysis." Journal of Big History, II(3), 2018.
• Korotayev, A., LePoire, D. (eds.) The 21st Century Singularity and Global Futures: A Big History Perspective. Springer, 2020: https://doi.org/10.1007/978-3-030-33730-8
• Max Tegmark, Life 3.0: Being Human in the Age of Artificial Intelligence (2017; Japanese edition: 『LIFE3.0』, Kinokuniya).
• Yudkowsky, E., Soares, N. "If Anyone Builds It, Everyone Dies."（Little, Brown and Company, 2025）

Credits and Notes

• This article is a session report prepared from the audio recording of the day and the speakers' slides.
• Names, affiliations, and titles are as of the time of the session.