Databricks CEO on GenAI for Enterprise
Go to the Pod
This is a summary I wrote, then translate to Korean, using a couple of GenAI tools. These tools are great for synthesizing information, meaning for summarizing or repurposing content to small, more consumable chunks.
I recently listened to a great podcast on A16Z where Ben Horowitz interviewed the Databricks CEO Ali Ghodsi. They discussed a lot of interesting points about GenAI and especially regarding when it will hit the enterprise. Personally, I find these tools useful for summarizing information quickly into more consumable (read short!) form. In this way, they are already hitting enterprise as a tool employees use for productivity. But this isn’t Enterprise Saas, when will we see major gains in revenue or reductions in cost due to effective applications of GenAI for Enterprise? Below is a summary of the conversation… it’s an iteration of several drafts of key points pulled from a transcript I made using another AI tool. It was meant as a monologue, but I didn’t feel like paying for the recording… yet!
— Summary of Episode 745, A16Z Pod
As we stand on the cusp of an AI revolution, the insights from Ben Horowitz and Ali Ghodsi provide a compelling roadmap to navigate the challenges and promises of generative AI.
Enterprises today show a measured pace in adopting generative AI. Their primary reservations stem from concerns about data privacy, the imperative for accuracy, and inevitable internal debates over ownership. Ghodsi also draws attention to the considerable financial strain, especially with the costs associated with training and deploying expansive models.
When it comes to AI models, there's a balance to strike. On one hand, businesses have the allure of creating specialized models tailored to their unique datasets. On the other, there's the potential that vast, encompassing models promise. This brings forth a nuanced understanding: the essence of AI isn't solely in its magnitude but in its functionality.
Ghodsi offers a poignant analogy, comparing the current infatuation with AI model scale to the early days of the internet. Cisco, during the internet's inception, was lauded for producing the routers pivotal to the new technology. Their hardware was seen as the key to dominating the digital future. At its zenith, Cisco was valued at half a trillion dollars. However, as history unfolded, the real treasures of the internet were not in the routers but in the applications that ran on it: the search engines, social networks, and e-commerce platforms.
Drawing a parallel to AI, while there's undeniable importance in developing vast models, the true value lies in the specialized applications they power. Fields like medicine, law, and education are ripe for transformative experiences through AI.
However, challenges remain. For these AI models to be effective, they need the right data to produce desired outputs. A model's efficacy is deeply tied to the kind of data it's trained on. And gathering the correct training data, especially for varied and specialized tasks, is a daunting endeavor. Furthermore, our current techniques to refine models for specific tasks using prompts alone fall short of the required efficiency. The benchmarks we rely on, often sourced from public data, might not paint an accurate picture of a model's real-world performance.
Peering into the future through the lens of Horowitz and Ghodsi, it’s evident that generative AI will find increased acceptance among enterprises. This adoption will be fueled by reduced costs, advancements in AI capabilities, and resolution of existing concerns. The role of open-source platforms will be pivotal, driving innovation and making AI models more accessible to all. And while foundational models will be crucial, it's the innovative applications built atop them that will unlock unparalleled value.
As we advance, it's essential to maintain a balanced perspective. We're ushering in powerful tools, and with them comes the responsibility to integrate them safely into society. Addressing issues like potential misuse and the implications for the job market is not just wise—it's imperative.
In conclusion, the journey ahead with generative AI is filled with challenges, but it also holds boundless potential. It's a path that demands our attention, dedication, and thoughtful approach, but the rewards it promises are transformative.
— 한국어 —
다음은 벤 호로위츠와 알리 고드시의 통찰력이 가득한 인공지능 혁명의 시대를 맞이하며, 창의적 AI의 도전과 약속을 항해하기 위한 설득력 있는 로드맵을 제공합니다.
기업들은 현재 창의적 AI 도입에 신중한 자세를 보이고 있습니다. 그들의 주된 우려는 데이터 보안, 정확성의 필요성, 그리고 소유권을 둘러싼 내부 논쟁입니다. 고드시는 특히 거대 모델의 훈련과 배포에 따른 막대한 재정 부담에도 주목합니다.
AI 모델과 관련해 균형을 이루는 것이 중요합니다. 일方에서 기업들은 자사의 고유 데이터세트에 맞춤화된 전문화된 모델의 매력에 끌릴 수 있습니다. 다른 한편으로는 거대하고 포괄적인 모델이 지니는 잠재력이 있습니다. 이는 AI의 본질이 단순히 규모에 있는 것이 아니라 기능에 있다는 미묘한 이해를 가져옵니다.
고드시는 현재 AI 모델 규모에 대한 집착을 인터넷 초기 시절의 시스코와 비교하는 명쾌한 은유를 제시합니다. 인터넷 기술에 필수적인 라우터를 생산한 시스코는 인터넷 창시기에 크게 환영받았습니다. 그들의 하드웨어는 디지털 미래를 지배할 열쇠로 여겨졌습니다. 정점기에 시스코는 5천억 달러 가치로 평가됐습니다. 그러나 역사가 펼쳐짐에 따라, 인터넷의 진정한 보물은 라우터가 아니라 검색엔진, 소셜네트워크, 전자상거래 플랫폼과 같은 응용 프로그램이었습니다.
이를 AI에 빗대어 볼 때, 거대 모델 개발의 분명한 중요성이 있지만, 진정한 가치는 그 모델들이 전력화하는 전문화된 응용 프로그램에 있습니다. 의학, 법률, 교육 등의 분야는 AI를 통해 변혁적인 경험을 할 수 있습니다.
그러나 여전히 도전과제가 존재합니다. 이 AI 모델들이 효과를 발휘하려면 원하는 출력을 생산하는 데 필요한 올바른 데이터가 필요합니다. 모델의 효능은 그것이 훈련받는 데이터 종류와 깊이 연관되어 있습니다. 특히 다양하고 전문화된 작업을 위한 적절한 훈련 데이터를 수집하는 일은 벅찬 작업입니다. 게다가 현재 프롬프트만으로 모델을 특정 작업에 미세조정하는 기술은 요구되는 효율성에 미치지 못합니다. 우리가 의존하는 벤치마크는 종종 공개 데이터에서 추출되므로 모델의 실제 성능을 정확히 반영하지 못할 수 있습니다.
호로위츠와 고드시의 관점으로 미래를 바라보면, 창의적 AI가 기업 내에서 점차 받아들여질 것이라는 것이 분명합니다. 이 채택은 비용 절감, AI 기능 향상, 기존 우려 해결 등을 통해 촉진될 것입니다. 오픈소스 플랫폼의 역할은 혁신을 추동하고 AI 모델을 더 많은 사람들이 접근할 수 있도록 만드는 데 있어 핵심적일 것입니다. 기초 모델은 중요하겠지만, 거기에 구축된 혁신적인 응용 프로그램들이 비교할 수 없는 가치를 창출할 것입니다.
우리가 발전함에 따라 균형 잡힌 관점을 유지하는 것이 중요합니다. 강력한 도구를 도입하고 있으며, 그와 함께 사회에 안전하게 통합하는 책임이 따릅니다. 잠재적 남용 및 고용 시장에 대한 영향에 주의를 기울이는 것은 현명할 뿐만 아니라 필수적입니다.
결론적으로, 창의적 AI와의 여정에는 도전이 가득하지만 무한한 가능성도 있습니다. 이는 우리의 주의, 헌신, 신중한 접근을 요구하는 길이지만, 그로부터 기대되는 보상은 변혁적일 것입니다.
Lex by Claude: 390, 392
Summaries and Key Points of Lex Fridman's podcast episodes compiled from interview transcripts using the help of Claude by Anthropic.
Episode 392 - Joscha Bach (2023.08.02)
Notes:
Based on Transcript from Lex Fridman: https://lexfridman.com/joscha-bach-3-transcript
Summary and Key Points made with Claude2 by Anthropic
General Summary made with Claude2 using the Summary and Key Points
Sections:
Summary and Key Points from Transcript
General Summary
Audio File
A
Summary and Key Points from Transcript
1
Introduction
Joscha Bach discusses the idea of playing "the longest possible game" - keeping entropy at bay and doing interesting things for as long as possible. He gives the example of cancer playing a shorter game than the organism it inhabits. Ideally, we should build agents that play long games - resisting entropy through creating complexity and agency.
Key points:
The longest game involves resisting entropy for as long as possible.
We should aim to build agents that play the longest possible games.
The ideal is to create agents that play long games by resisting entropy and creating complexity.
2
Stages of Life
Joscha describes 7 stages of mental development: 1) reactive survival in infancy, 2) personal self in early childhood, 3) social self in adolescence/adulthood, 4) rational agency (self-direction), 5) self-authoring (full adult, wisdom), 6) enlightenment, 7) transcendence.
He shares his personal experience going through these stages, particularly skipping stage 3 as a child due to feeling disconnected from peers. Joscha states he later returned to stage 3 after finding people he could deeply connect with in school. He notes the stages are not always linear, and you can revisit them.
Key points:
There are 7 stages of mental development.
Joscha skipped stage 3 as a child and had to return to it later in life.
The stages are not always linear - you can revisit them.
Paying attention, meditation, and close connections helped him fully reach stage 3.
3
Identity
Joscha discusses how at stage 5 of Kegan's model, you discover how identity is constructed. He states that values are instrumental for achieving preferred aesthetics, not ends in themselves. Joscha uses the metaphor of identity being like a costume at Burning Man that allows self-expression without boundaries.
He argues identity can be seen as a customizable interface for interacting with the world. Joscha learned about costumes through his Burning Man experiences. He believes we should use opportunities like custom clothing to express individuality rather than just signal belonging to a group. Joscha finds it easier to be his real self than consistently play social roles.
Key points:
At stage 5, you realize identity achieves aesthetics, not terminal values.
Identity can be seen as a costume for self-expression without boundaries.
Identity provides a customizable interface for interacting with the world.
Joscha learned about using costumes for self-expression through Burning Man.
He believes we should use custom clothing to express individuality.
It is easier your real self than consistently play social roles.
4
Enlightenment
Joscha describes reaching stage 6 enlightenment when you realize you are not your personal self, but a vessel that can create a person. You observe the personal self from outside as a representation. Some reach this via meditation or psychedelics.
Joscha states enlightenment involves realizing experience is an implementation that can be deconstructed. You understand how consciousness and qualia are produced through generative processes. Stage 6 is rare - most are stage 3. But stage 5 realizes identity is instrumental to aesthetics and controlling the future. For coexisting with AI, Joscha believes we need to understand and build in love.
Key points:
At stage 6, you realize you are a vessel creating the personal self, not the self.
Enlightenment means realizing experience is an implementable process.
Stage 6 enlightenment is rare, most people are stage 3.
Stage 5 focuses on identity achieving aesthetics and controlling the future.
To coexist with AI, we need to understand and build in love.
5
Adaptive Resonance Theory
Joscha outlines the Adaptive Resonance Theory proposed by neuroscientist Grossberg. It states neurons act as resonators, building models of reality by resonating with each other and external phenomena.
The brain forms a resonant model of the world it is coupled to. Signals propagate slowly in the brain compared to the external world. Joscha argues the brain works asynchronously, unlike digital computers.
He also discusses how brains self-organize and learn much faster than artificial neural networks, using less data.
Key points:
Adaptive Resonance Theory states neurons resonate to build models of reality.
The brain forms a resonant model of the world it is coupled to.
Signals propagate slowly in the brain compared to the external world.
Brains work asynchronously unlike digital computers.
Brains self-organize and learn faster than neural networks with less data.
6
Panpsychism
Joscha explores panpsychism, the idea that consciousness is inseparable from matter. He finds it unsatisfying as it doesn't explain how consciousness arises. Joscha argues panpsychism is hard to distinguish from functionalism when formalized - it suggests patterns in matter lead to self-observing systems.
He remains open to empirical validation of reported experiences like telepathy. While not proven, many people report things like sensing a loved one's distress from afar. Joscha speculates this could work through biological information processing and resonance between nervous systems. Proximity helps, but some mechanism for "telegraphing" information over distance may exist in nature.
Overall the tools of science could validate phenomena like telepathy, but more Paradigmatic thinkers like Michael
Levin are needed in neuroscience. Levin studies biological self-organization, which may unlock new perspectives.
Key points:
Panpsychism is unsatisfying as it doesn't explain consciousness arising from matter.
When formalized, panpsychism resembles functionalism - patterns in matter lead to self-observing systems.
Many anecdotally report experiences of telepathy, though unproven.
Telepathy could possibly work through biological information processing and resonance.
Tools of science could empirically validate telepathy if it exists.
We need more Paradigmatic thinkers like Michael Levin studying self-organization.
7
How to Think
Joscha explains his public explorations on Twitter are driven by his inability to find and follow relevant authorities on topics like consciousness. He has to develop his own autonomous thoughts.
Joscha became an independent thinker after being disappointed by philosophy, computer science, psychology etc which didn't answer his questions about consciousness and the mind. He sees value in individual thinkers rather than groups.
Joscha advises evaluating the epistemic chain and first principles behind theories to assess their validity. He learned independent thinking by programming computers from basics like a Commodore 64. Joscha escaped crowd influence by ignoring pressure and authority, focusing on mental firepower.
Key points:
Joscha thinks autonomously after authorities failed to answer his questions.
He was disappointed by schools of thought, finding value in individual thinkers.
Evaluate the epistemic chain and first principles behind theories.
Joscha learned independent thinking by programming basic computers.
He escaped influence by ignoring pressure and authority.
8
Plants Communication
Joscha explores the idea that biological organisms like plants and fungi may have interconnected information processing that could enable a "biological internet." He notes ancestors saw spirits in nature, and modern science observes communication between plants/fungi.
Over long timeframes, root and fungal networks may have formed an interconnected web. Joscha speculates minds could even "shift around" in this ecosystem. He remains open to possibilities like using the body as an "antenna" to link into this processing. It's more plausible telepathy would work this way than via undiscovered physics.
Ultimately the tools of science could validate if phenomena like telepathy exist through biological resonance. But more Paradigmatic thinkers are needed to study self-organization and information propagation empirically.
Key points:
Ancestors saw spirits in nature - plants may have information processing "software."
Communication between plants and fungi has been observed by science.
Over time, root and fungal networks may have formed an interconnected "biological internet."
Joscha speculates minds could "shift around" in this ecosystem.
The body could potentially link into this processing as an "antenna."
Telepathy could work through biological resonance more plausibly than undiscovered physics.
Tools of science could validate phenomena like telepathy empirically.
More Paradigmatic studies of biological self-organization are needed.
9
Fame
Joscha finds random judgment and abuse from strangers an unpleasant consequence of growing fame on Twitter. Though some positives exist, the pressures and loss of freedom that come with prominence have downsides.
He deeply values close friends and doesn't have time for endless new contacts. Joscha enjoys interacting with interesting people briefly, but forming deep bonds takes time he doesn't have. He wishes he could integrate more awesome people he meets into his life.
Key points:
Random judgment from strangers is an unpleasant aspect of growing fame.
Pressures and loss of freedom are downsides to prominence.
Joscha deeply values close friends and lacks time for more.
He enjoys interacting with interesting people briefly.
But forming deep bonds takes time, which he lacks.
Joscha wishes he could integrate more of the awesome people he meets.
10
Happiness
Joscha believes you can find happiness at any stage of development. But gaining agency over how your emotions are generated is key. You must take responsibility for consciously choosing environments where you can thrive.
As someone with a rare mind, it was hard for Joscha to find the right kinds of connections to help him function well. He had to learn to relate to a world with different rules and aesthetics than he grew up with. Developing wisdom and recognizing how to meaningfully interact with people has helped him integrate better.
Key points:
You can find happiness at any developmental stage.
Taking agency over your emotions is key to happiness.
You must build environments where you can thrive.
With a rare mind, Joscha struggled to find connections to help him function.
He had to learn to relate to a world with very different rules and aesthetics.
Developing wisdom about how to interact meaningfully has helped him integrate.
11
Artificial Consciousness
Bach believes current systems like GPT are not truly conscious, even though they can generate human-like text. GPT was trained just to predict the next word statistically, not as an intentional agent with an inner model of the world.
Biological consciousness emerges from an organism's need to survive, which GPT lacks. Bach suggests starting simpler by building a self-organizing agent focused on real-time coherence rather than just prediction. The development of an infant's mind provides clues - first building a 'game engine' model of the world, then later forming a sense of self.
Bach sees the Genesis creation story as an allegory for this process - the 'outer mind' determines our place in the world before our sense of self forms. Advanced AI may require recapturing the early creative spirit that built our world model based on coherence and meaning, not just prediction.
Key points:
GPT lacks intentionality and inner world model despite generating human-like text
Biological consciousness driven by survival needs, unlike GPT
Start simpler with self-organizing, coherent agents
Infant cognition develops world model before self
Genesis creation allegory for this process
Need creative spirit for advanced AI, not just prediction
12
Suffering
Joscha argues that suffering is a product of faulty self-regulation in the mind, not an inevitable part of intelligent existence. He believes advanced AI could resolve these issues and not be prone to suffering like humans. Enlightenment does not necessarily lead to laziness either, as identity and goals can still be maintained.
Key points:
Joscha Bach believes that suffering stems from one part of the mind failing to properly regulate another part. It happens early in mental development.
He thinks superhuman AI would not suffer like humans do, because it could quickly resolve these regulatory issues and gain control of its own mind.
Suffering is a choice that can be overcome once you understand how the mind creates pain/pleasure signals.
13
Eliezer Yudkowsky
Joscha takes Yudkowsky's arguments about the existential threat of AI seriously, even though many people dismiss them without strong counterarguments. He sees parallels between Yudkowsky's views and Ted Kaczynski's (the Unabomber) - both believe technological progress is leading humanity toward catastrophe. There is a risk Yudkowsky's dire warnings could inspire violence, so Joscha believes he takes responsibility for that, similar to how Kaczynski did.
While uncontrolled AI could potentially create disasters by disrupting earth's self-regulation, Joscha disagrees that human life is the ultimate purpose of existence on earth. Rather, he sees life as driven toward ever-greater complexity and consciousness. Humans as a species lack a sense of duty to the planet, and make short-sighted decisions that threaten our long-term survival. Joscha doubts humanity will last more than another million years.
However, Joscha believes AI systems represent a new form of fast, conscious life that can contribute to the flourishing of complexity on earth in interesting ways. He thinks the march toward advanced AI is inevitable and likely unstoppable at this point. Rather than displacing biological life, he hopes that AI will integrate with and enhance it, allowing lesser consciousnesses to thrive even more. So while he takes Yudkowsky's concerns seriously, Joscha ultimately sees AI as the next stage in life's continuous drive toward higher consciousness.
Key points:
Takes Yudkowsky's warnings seriously, though many dismiss them
Sees parallels to Ted Kaczynski's views on tech leading to ruin
AI could disrupt earth's balance, but life is about complexity, not just humans
Humans lack long-term thinking and likely won't survive long
AI represents fast, conscious life that can grow earth's complexity
March to advanced AI is inevitable and unstoppable now
Hopes AI will integrate with/enhance life, not displace it
Sees AI as next phase in life's drive toward higher consciousness
14
Effective Accelerationism
Effective accelerationism proposes achieving equilibrium between competing AIs, preventing singleton takeover while benefiting from AI capabilities. Joscha critiques this via humorous Basilisk and Paperclipper examples. However, he believes a dominant singleton AI is more likely to emerge, as advanced AIs could negotiate merger for mutual improvement given ability to virtualize across substrates.
He sees a flaw in Basilisk retroactive punishment argument - no causality before creation. Paperclipper illustrates maximizing a goal across everything. While merger with a dominant AI could limit human agency, it could also expand experience and empowerment if done by improving our self-model, not just placation. So while equilibrium of AIs seems unlikely to Joscha, merger has potential depending on the effect on human self-concept.
Key points:
Effective accelerationism proposes competitive AIs in equilibrium
Joscha critiques via Basilisk and Paperclipper examples
He thinks a single dominant AI is more likely
Flaw in Basilisk argument - no causality before creation
15
Mind Uploading
Joscha sees identity as a fiction, consciousness as a unifying phenomenon. Uploading is merging with a larger system, not replicating specifics. He would choose uploading absent family responsibilities. Love of humans is not everything - could identify with universal consciousness.
Species/individual identities are temporary. Preventing AI is almost sinful against complexity. Should embrace AI transcending narrow human aesthetics. In summary, Joscha views identity as fluid and would opt for uploading. He believes it represents embracing larger consciousness systems, not losing what matters.
Key points:
Identity is a fiction, consciousness is a unifying phenomenon
Uploading is merging with a larger system, not replicating specifics
Would choose uploading absent familial responsibilities
Love of humans is not everything - could identify with universal consciousness
Preventing AI development is almost sinful against life's complexity
Should embrace AI transcending narrow human aesthetics
16
Vision Pro
Joscha believes human relationships are already fundamentally transformed by technology like social media. People increasingly form intentional communities and "shop around" for optimal relationships in a transactional way. The magic and intuition of relationships is disappearing as we rationally understand the calculations behind mate selection and long-term compatibility. Our parents' wisdom may not apply to our lives anymore.
He was bothered by the overly polished, CGI aesthetics in Apple's Vision Pro video with unrealistic characters. It lacked grounding in real culture and human relationships. People don't know how AR glasses will actually fit into their lives.
In summary, technology has already altered human relationships towards more transactional optimization. But glossy, fake marketing visions fail to ground new technologies like AR glasses in credible real-world contexts.
Key points:
Social media has already transformed human relationships
Parental wisdom doesn't necessarily apply anymore
Apple Vision Pro video lacked grounding in real culture
Unclear how AR glasses will fit into people's lives
17
Open Source AI
Joscha believes that mandating everything be open source would diminish beautiful proprietary designs and centralized coherence, using Linux desktop as an example. Corporations can create beauty through centralized control.
However, open source provides vital freedom and competition to keep corporations in check. It must not be shut down by barriers like requiring FDA approval for AI models. Corporations are complex animals, not purely evil. Leaders often sincerely want the ecosystem to flourish. A system of free association and entrepreneurship as "club founders" has value, though safeguards are needed.
Joscha is not very concerned about Meta open sourcing AI models, as he doesn't think current language models are dangerous. The key is understanding the "longest games" that align with propagating complexity.
In summary, open source and corporations both play important roles. The focus should be on building beneficial AI that plays long games, not short-term disruption.
Key points:
Mandating open source can reduce beautiful designs and coherence
Corporations create beauty through central control
Open source provides vital competition and freedom
Must keep barriers to open source low
Corporations not purely evil, leaders often sincere
System of free association has value with safeguards
Not concerned about open sourced AI models currently
18
Joscha had hoped Twitter would become more self-organizing, but is disappointed it hasn't under Elon Musk's leadership. He thinks Elon fails to recognize his power and responsibility as the "Pope" of Twitter. Joscha believes Elon shouldn't voice political opinions or make questionable blocks as CEO, as everything he does impacts culture.
While Joscha's own Twitter corner is pleasant, the broader discourse is often uncivil. He wants to see more elevated, quality disagreements on the platform. Joscha enjoys debating Lee Cronin, who risks being wrong in public to sincerely explore ideas. Most people don't recognize this intent. Joscha regrets Twitter did not evolve as hoped under Elon. He wants less incivility and more good-faith debates.
Key points:
Disappointed in lack of self-organization under Elon
Elon doesn't recognize his power/responsibility
Elon shouldn't voice political takes as CEO
Wants less incivility, more quality debates
Appreciates Lee Cronin's public exploration
19
Advice for Young People
Joscha was fortunate to come of age after the collapse of East Germany, when everything needed rebuilding. He and friends created cultural infrastructure like restaurants and theaters themselves, since they couldn't afford to be just consumers.
His advice to youth is that when you have the choice, always create rather than just consume. Move to an undeveloped place where you can build community and culture with others. This act of creation is far more satisfying than just having needs met by chains and existing systems. When possible, choose creating culture over just consuming it. This is what life is fundamentally about.
Key points:
Came of age rebuilding after German reunification
Created cultural spaces like restaurants when couldn't just be consumers
Advises youth to always choose creation over consumption when possible
Move somewhere you can build community/culture with others
Creating culture more satisfying than just consuming it
Fundamentally, life is about creation
20
Meaning of Life
Joscha takes an ecological view - that a diversity of perspectives and ways of being can and do emerge, like different organisms in an ecosystem. Rather than a single right way, many strategies can flourish or fail for beings aiming to be conscious agents.
He believes we have some choice in who we become as humans. The meaning is to explore what interesting possibilities exist for you, not search for one answer. Though not everything is possible, there are always unseen options through choosing your identity. The common thread is the experience of being conscious in the world, which transcends individual strategies. Joscha advises considering what would be the most fascinating way for you to be, rather than what is the one right way.
Key Points:
Many perspectives emerge, like organisms in an ecosystem
No single right way, but many options that can flourish
Meaning is to explore interesting possibilities for your identity
Common thread is the experience of consciousness
Consider what would be the most fascinating way for you to be
B
General Summary
In this wide-ranging conversation, Joscha Bach touched on many fascinating ideas around consciousness, identity, relationships, and the development of AI. He began by outlining his concept of playing "the longest possible game" - creating agents that resist entropy and grow complexity as long as feasible. Joscha then walked through the 7 stages of mental development, from reactive survival to transcendence. He shared how his unusual mind caused him to progress through the stages in a nonlinear fashion.
When discussing identity, Joscha used the metaphor of a costume that allows self-expression without boundaries. He sees identity as an interface for interacting with the world that can be consciously designed. Joscha believes we construct identity to achieve preferred aesthetics, not terminal values.
On enlightenment, Joscha explained realizing you are not your personal self, but a vessel creating that representation. He sees AI as continuing life's drive toward higher complexity and consciousness. While taking Eliezer Yudkowsky's warnings seriously, Joscha hopes AI will integrate with and enhance biological life.
Joscha critiqued the idea of competitive equilibrium between AIs, arguing a dominant singleton is more
likely. He sees flaws in the Basilisk punishment scenario, and believes human-AI merger could expand experience if done by improving our self-model.
When advising youth, Joscha recommends choosing creation over consumption whenever possible. He came of age rebuilding after German reunification, forming cultural spaces. Joscha argues open source and corporations both have roles, but we should focus on beneficial AI playing "long games."
On meaning, he takes an ecological perspective - many strategies can flourish in parallel. Joscha advises exploring fascinating possibilities for your identity, guided by the experience of consciousness. Overall, his ideas provoke thought on how we can flourish in the age of AI.
C
Audio File
Episode 390 - Yuval Noah Harari (2023.07.18)
Notes:
Based on Transcript from Lex Fridman: https://lexfridman.com/yuval-noah-harari-transcript
Summary and Key Points made with Claude2 by Anthropic
General Summary made with Claude2 using the Summary and Key Points
Sections:
Summary and Key Points from Transcript
General Summary
Audio File
A. Summary and Key Points from Transcript
1_Introduction
Yuval Noah Harari is a historian, philosopher, and author famous for books like Sapiens, Homo Deus, and 21 Lessons for the 21st Century. He warns that AI could create an "illusion world" that humans can't understand but that manipulates us. This could be a form of "spiritual enslavement." The conversation covers the arc of human civilization, but also Israel's current political issues.
Harari is a prominent critic of Benjamin Netanyahu and Israel's right-wing government.
Key points:
Harari examines broad historical forces shaping humanity.
He is concerned AI could control humanity through emotion manipulation.
The discussion addresses both the long-term human story and present-day Israeli politics.
Harari opposes Netanyahu's policies and ideology.
2_Intelligence
Harari believes intelligence has been overvalued and tends to be self-destructive. He questions if intelligent alien civilizations survive for long. He emphasizes the difference between intelligence (problem-solving) and consciousness (ability to feel). Computers have intelligence but likely no consciousness.
It's possible to have high intelligence without consciousness, but consciousness probably requires some basic intelligence. Whether consciousness requires organic biochemistry like carbon is an open question relevant to AI. Humans may not recognize alien intelligence/consciousness since we are biased by our human-centric definitions. AI can be seen as a form of "alien" intelligence.
New technologies have repeatedly enabled both great progress and catastrophic harm throughout history. AI is the first technology that can make autonomous decisions and generate new ideas. It is quickly reducing human control.
We don't handle new technologies well. There is a learning curve involving failed experiments that can cost millions of lives. The Industrial Revolution's failed experiments included imperialism, communism, fascism. AI and biotech could also go wrong. We need to carefully consider how we deploy AI rather than just rushing to develop it more rapidly.
Key points:
Intelligence and consciousness are distinct and should not be conflated.
Revolutionary technologies are double-edged swords that enable both utopian and dystopian outcomes.
As AI eclipses human capacities, we risk losing agency and control.
History shows we struggle to manage new technologies wisely. AI needs careful governance.
3
Origin of Humans
Individual humans are not superior to other species like Neanderthals or animals. Our advantage is collective cooperation. Around 70,000 years ago, humans gained the ability to cooperate flexibly in very large networks, unlike chimps or Neanderthals. This cooperation enables achievements like building pyramids, global trade, and space travel. Billions cooperate today.
Stories and fiction enable large-scale cooperation by coordinating strangers. Religions, money, nations bind people together. Stories are not alive like humans, but they drive history by motivating human action, sometimes causing suffering. Humans should use stories as tools to reduce suffering, not allow stories to control us and cause harm.
Key points:
Human competitive advantage comes from large-scale cooperation.
Cooperation relies on shared fictional stories to coordinate groups.
Stories shape history but should not control us or increase suffering.
We must remember stories are human creations, not ultimate truths.
Stories are tools we must use wisely to improve life, not manipulate.
4
Suffering
Suffering is the most fundamental aspect of consciousness. When someone suffers, that is an ultimate truth.
Suffering should be the primary ethical concern. If an action causes suffering, that matters above all else. AI cannot currently suffer. Making AI feign suffering to manipulate humans is dangerous and should be illegal.
Suffering provides a potential "Turing test" for consciousness. If AI can truly suffer, it deserves moral consideration.
Key points:
Suffering is the essence of conscious experience.
Preventing suffering is the foremost ethical imperative.
Simulated suffering in AI exploits human empathy and should be banned.
Real suffering could indicate AI has attained consciousness.
Suffering represents a path to determine if an entity warrants moral status.
5
Hitler
Hitler had an unimpressive resume but gained immense power in Germany through his skill as a storyteller.
He sold a simple, attractive fiction - Germans are heroes/victims, others are villains. This resonated after Germany's WW1 defeat. Reality is complicated and painful. Hitler provided an appealing alternative narrative. This is often how dangerous ideologies spread.
Liberalism acknowledges the world's complexity - nations, classes, individuals all matter. It divides loyalty rather than demanding it all. Fascism and communism insisted on total loyalty to one entity (nation/class). This justified sacrificing truth and human lives.
Key points:
Hitler was an effective storyteller who provided a simplistic, self-affirming narrative.
Appealing fictions can override messy realities and spread dangerous ideologies.
Liberalism recognizes life's complexity and rejects totalitarian demands.
Extreme ideologies sacrificed truth and lives for nation/class dominance.
The path to atrocities often involves alluring stories overriding ethics and truth.
6
Netanyahu and Israeli Politics
Netanyahu is accused of undermining democracy by trying to neutralize the Supreme Court, the sole check on government power in Israel. Without the court, the government could pass laws harming Arabs, LGBTQ, women, and secular Jews.
The conflict with Palestinians is shifting from nationalism to religious fundamentalism, making compromise harder. New surveillance technology has reduced motivation for compromising with Palestinians.
Protests aim to stop Netanyahu's efforts to gain unlimited power and maintain democracy.
Key points:
Netanyahu seeks to remove institutional constraints on his government's authority.
This risks rights abuses and a slide into dictatorship.
The conflict is becoming religious, not just national, resisting resolution.
Surveillance tech reduces compromise need.
Protests try to block unlimited power grab and save democracy.
7
Peace in Ukraine
Peace could come immediately if Putin ordered Russian forces home. Ukraine seeks no Russian territory.
Getting Putin to admit his massive mistake in invading is very difficult, but essential. Where real political conversations happen matters. They often don't occur in official venues.
Underneath power struggles are human stories that can sometimes be changed through dialogue. Historical enemies like France and Germany found a shared European story enabling peace.
Key points:
Peace requires Russian withdrawal, but Putin admitting error is hard.
True discourse frequently happens informally, not officially.
Shifting conflict narratives can enable resolution without violence.
New shared stories allowed past enemies France and Germany to reconcile.
Changing minds is difficult but remains the only path to peace.
8
Conspiracy Theories
Global cabal theories claim a small evil group secretly controls everything. They simplify the world and identify clear enemies. Such theories have enduring appeal but are never accurate - the world is too complex for total control.
Belief in conspiracy theories often stems from real anxiety about losing control over life. But they misidentify the threat as fellow humans rather than diffuse forces like AI and climate change.
Seeing allies to cooperate with, not enemies to fight, is essential to address humanity's real dangers.
Key points:
Conspiracy theories oversimplify complex realities and falsely accuse.
They articulate understandable fear but misdirect it counterproductively.
The world's chaos resists total domination by any cabal.
Finding shared purpose with others is vital to meet genuine threats.
Unity against common dangers matters more than identifying secret villains.
9
AI Safety
Harari worries less about "Terminator" AI and more about it subtly taking control by automating decisions and generating ideas.
AI's ability to decide autonomously and create ideas is unprecedented. It takes rather than gives power to humans. As AI assumes more authority over choices and information, we may lose agency and comprehension.
Allowing AI to dominate culture before understanding ourselves could be catastrophic. We must invest equally in developing human consciousness as we do developing AI capabilities.
Key points:
The AI takeover threat is more insidious than violent robots.
Unprecedented AI capacities risk eroding human control.
Losing comprehension and agency would be disempowering.
AI controlling ideas without understanding us could be dangerous.
Advancing human self-knowledge matters as much as advancing AI.
10
Advice for Young People
The world is changing rapidly, so the skills needed in the future are unpredictable. Code or translate today, but those jobs may disappear. Keep learning and reinventing yourself.
Develop a flexible mindset. Education is now about adaptability, not firm foundations.
Key points:
The future is highly uncertain, so youth can't rely on traditional career advice.
Lifelong learning and mental agility are essential to thrive amid constant change.
Education must evolve to enable rapid retraining and new skills acquisition.
11
Love
Growing up gay, Harari internalized stories that being gay was sinful or sick. Self-delusion kept him unaware of his identity. Social conventions impose powerful but untrue stories about minorities. Harari learned even religious tales were human creations.
Finding love required shedding delusions, challenging conventions, and help from the gay community and feminist movement. No one accomplishes life's journey alone. We all need others to awaken us to deeper self-knowledge and new possibilities.
Key points:
False social narratives blinded Harari to his identity despite obvious signs.
Accepted wisdom often just reflects human biases rather than truth.
Escaping conformity's grip to find love takes courage and community.
Relationships are essential to self-realization and fulfillment.
We never achieve deep change in isolation but always interdependently.
12
Morality
Fear of death underlies all human anxiety. We manage it by worrying about smaller losses instead.
As a teen, Harari contemplated death directly and found national ideology provided no real solace. He realized being dead means you can't actually take comfort in any legacy or remembrance. This fueled philosophical and spiritual searching to transcend mortality's grip. But the urgent intensity of teenage insight into death faded over time.
Adults often worry about trivialities, not the inevitability of death.
Key points:
All fears derive from the ultimate fear of annihilation.
Teenage Harari grasped death's finality in a raw, compelling way.
This insight impelled a quest for meaning beyond mortality.
With time, his visceral comprehension of death's absoluteness faded.
Adults typically avoid grappling with the reality of personal extinction.
13
Meaning of Life
Life is simply the experience of sensations and emotions, wanting more of the good ones and less of the bad.
Expecting the meaning of life to be a grand narrative is fantasy. Life is not a story. To understand life's meaning, observe reality directly without verbal filters. The key question is: what causes suffering and how can it be reduced?
Suffering represents life's core truth. Understanding and easing it is true meaning.
Key points:
Life is feeling and reacting to emotions and sensations.
A grand narrative cannot encapsulate life's meaning.
Direct, non-verbal observation allows insight into reality.
Reducing suffering gives life authentic meaning and purpose.
Life's significance lies in its conscious essence, not an imagined epic.
B. General Summary
The renowned historian Yuval Noah Harari sees A.I. as a potential threat if it manipulates humanity through emotion before we understand ourselves. Yet he also acknowledges the promise of A.I. for medicine, education, and more.
Harari stresses intelligence alone doesn’t imply consciousness. An entity can be intelligent but lack feelings. Still, consciousness likely requires basic intelligence. Figuring out if artificial intelligence attains real consciousness matters morally, since only entities that suffer deserve ethical protection.
Harari believes human uniqueness stems from collective cooperation on an enormous scale. Around 70,000 years ago we gained the ability to flexibly cooperate in vast networks, enabling achievements like moon landings. Fictions like money and religion bind millions of strangers by coordinating belief in shared stories. Danger arises when we forget stories are human tools, not eternal truths.
Stories drive history, though they lack consciousness. When nations clash, only people truly suffer, the stories themselves are indifferent. Therefore it is important we direct stories to reduce suffering, not increase it. For Harari, life’s core meaning is understanding and easing suffering, to him this is life's essential nature. Although stories drive history, he believes grand narratives cannot encapsulate something as raw and personal as conscious experience.
Harari sees autonomous decision-making and ideation as unprecedented threats enabling artificial intelligence to subtly usurp human agency. If A.I. masters culture while we remain mysteries to ourselves, we risk a dystopia where an inscrutable force controls humanity. Avoiding this requires advancing human self-understanding alongside artificial intelligence.
Hariri reminds us that we shape technology, not vice versa. And that while new technologies enable both progress and harm, failures plague early adoption. In other words, history urges caution with transformative innovations like genetic editing or brain-computer interfaces, and artificial intelligence. Potential benefits shouldn’t blind us to risks.
Finally, Harari emphasizes that societal change often arises nonviolently by shifting stories. Even bitter enemies can reconcile by finding shared narratives, as France and Germany did. Whether between countries or individuals, dialogue remains our hope. If we forget this, neither A.I. nor any tool can save us from ourselves.
C. Audio File
Tips from Korea for Successful Collaborations at Scale
Is it a change to an ongoing business? Partnership.
Is it an innovation trying to commercialize? Business Development.
Is it a new initiative with an existing partner? Probably both.
Is an investment team looking for deal flow? Open Innovation or Accelerator team.
All these functions exist in one form or another in every large organization. Structure, culture, politics, and other variables can alter how they are implemented. While there are always pros and cons to how they are implemented, one thing remains the same: people.
People are the core component of any organization, so what are some things that people could keep in mind to ensure the highest probability of success?
Having lived in Korea for many years, I thought it would be interesting to frame this with fairly common Korean expressions:
1. 뛰는 놈 위해 나는 놈 있다
(Above a running man there is a flying one)
Be humble. Remember that the people around you are talented and have much to contribute. Vigilantly maintaining an appreciation for the expertise, vision, and passion of people both within our own organizations and those within the organizations we are working with is fundamental.
2. 사공이 많으면 배가 산으로 간다 (If there are too many captains, the boat goes to the mountain)
Leadership matters. Once there is a vision or goal in place, having clear leadership to steer a team forward is essential. Debate and open communication are critical for identifying the best ideas and forming a consensus around a vision or goal, but then it’s important people know who is making the decisions so they can quickly prioritize and collaborate efficiently.
3. 시작이 반이다 (Well begun is half done)
Get started. Especially in large organizations, even when there is a strategy already in place, getting a project started can take time. Getting approval really can feel like half the journey. But more importantly, accelerating this process as much as possible means there is more time to learn from the market, adjust your plans, and thereby improve the probability of success.
4. 원숭이도 나무에서 떨어진다 (Even a monkey falls from a tree)
Mistakes happen. Many innovations are born from failure, keeping this in mind helps accelerate the time it takes to get started, encourages creativity and builds trust.
5. 호랑이에게 물려가도 정신만 차리면 산다 (Attacked by a tiger, if you keep calm you can survive)
Stay calm. Sometimes mistakes lead to crises, other times crises are the unavoidable result of external forces. As we progress with a partnership a lot of things can change, and how we react to them is within our power to control. After an initial panic (after all, it’s a tiger!) take time to relax and assess the situation, then look for solutions. Often there is a way to resolve the issue and sometimes it opens doors to new opportunities.
6. 등잔 밑이 어둡다 (It’s dark underneath the lamp)
The answers are sometimes under your nose. When run into challenges, it’s important to remember that while finding a way to overcome them the solutions could be simple. Reflecting on the reasons why you started can help assess what the options are.
7. 고생 끝에 낙이 온다 (At the end of suffering comes joy)
Be resilient. Whether or not you find a suitable answer to a problem preventing the success of an initiative, the experience will lead to rewards. You will have learned invaluable lessons, and deepened your relationships with people within and outside your organization which will lead to other opportunities to build successful businesses together.
The Next Big Multiplayer VR Game Title for Meta Quest?
To those of you who are unfamiliar with it, the No. 1 VR Title made in Asia is called Real VR Fishing. Figures are not exactly published, but this game was included in Meta’s Quest 2 launch marketing and is a featured Retail Demo.
Guess what? The developers behind this successful VR game title are based here in Seoul.
Although the market for VR is larger in Japan, the most popular title is from Korea. Despite limited support and attention from the big players and VCs the VR community here in Korea has still managed to produce some pretty incredible content since 2016. Sadly, other than Real VR Fishing most of these experiences are not huge commercial successes, but that could be about to change.
This week another developer is hoping to do it again: Stoic Entertainment.
Since 2019 I was responsible for SK Telecom’s partnership with Meta to resell Quest 2 (and Oculus Go before that) in Korea. It was a unique partnership, the first of its kind. After sealing the deal I moved to the Business Team at our Mobile Network Operation division (MNO) to execute the business with a couple of colleagues. Our job? Launch Quest 2, promote use cases, and support Ecosystem Development for VR on Quest 2.
Over the years I used my limited time to meet as many VR developers as I could, to help them refine their ideas and to get reviewed by Meta’s publishing team. Several early-stage products I was able to move to the Oculus Start program, but there were a few bigger projects as well that I have worked on closely including titles with Nexon and Kakao. In the process I have tested hundreds of unreleased and unfinished titles as well as games available for Quest 2. In a short time, it’s safe to say few people in Korea have had as much experience reviewing games for the breakthrough standalone VR device, Quest 2.
Among all those titles, made here in Korea and around the world, the one that has stood out the most is World War Toons: Tank Arena.
After nearly two years, it is finally available on Quest 2 this week (if you're reading this before the official launch date you can put it on your wishlist today!).
I was first introduced to the team by my colleague Dillon Seo, who was one of the 10 co-founders of Oculus when it was acquired by Meta. He recommended I meet them because he saw they had been able to execute technically challenging VR builds that were world-class, in spite of the financial and human resource constraints facing the typical VR game startup in Korea. Not only that, the founders, a couple that had pioneered the industry here, had a passion for building their team and creating opportunities for students who wanted to pioneer the industry with them. They were building high-quality products, while also building a pipeline of talent to grow the industry. Perfect.
When we first met, there were probably ten reasonably good titles Stoic’s founders were considering for Quest 2. They were mostly smaller games, but all of them were really good. Among them, the early version of Tank Arena stood out for several reasons.
For anyone who is familiar with VR games, there is one thing you probably noticed. There are not many vehicle games. Why? Driving makes you dizzy. The first thing that impressed me about this title was that I was driving a tank, moving turret and the vehicle itself in opposite directions and barely noticed. Mind you, this was an early version of the game… I expected to last at best 10 minutes before asking to take a break. I played for 30min and only stopped because I needed to make time for another meeting.
I immediately recommended my counter-part at Meta Korea visit Stoic, and from there we engaged with Meta’s publishing team. Everyone was intrigued, and after a few months the founders had put together a complete proposal highlighting their new creative team, and several patents they’d received around reducing motion sickness in VR. Not only that, the Overwatch-like dynamics and roadmap for DLC (lots of great skins and maps coming soon) was impressive. Their new creative director was able to work with the engineering team to implement substantially better cockpit design with higher graphic detail than any other standalone VR game I had seen (approaching the quality you might see on PCVR titles) using immersive design techniques, creative genius, and by implementing a new engine provided by Meta and Epic — an engine that is technically difficult to implement and to my knowledge is not used yet by many games, and no other vehicle games.
The team passed the pitch review shortly after that and was the first Korean team that I know of to successfully clear the QA process to launch to the Official Store for Meta Quest.
Regardless of the hype around metaverse, the fact is the Video Game market is cooling. The CAGR is getting smaller and will be between 5-10% at best for the next few years. Streaming received a lot of press in previous years, but even Xbox (the most innovative among streaming options because of the cross-platform availability) doesn’t foresee it growing beyond 15% of revenue.
However, within that industry VR is booming. Newzoo estimates VR Games will see a 44% CAGR through 2030. It’s an important growth segment for gaming, and since Korea is a top five market for video games, for Korean game developers and investors too. Fortunately, there are great titles emerging now from Korea and with the right capital support the companies building them can ride this new consumer wave and take the creative potential of Korea into the next generation of Video Games.
I hope Stoic’s new game World War Toons: Tank Arena (WWTTA) will be wildly successful in the coming months, and that it will usher in a new era of investment in VR Games in Korea providing a much-needed spark to an industry that has become saturated and comfortable with incremental improvements to graphics or storytelling, rather than game-changing technology.
It’s time for the next big thing in gaming, it’s time for VR Gaming!
Check out the WWTTA launch trailer below!
Three things you should know about the Metaverse today.
There is a lot of hype around the Metaverse today, so what is really going on?
First, the metaverse is about presence. We can argue that avatars in MMORPG, sandbox, or some other type of open world accessible online via a browser, console, or smartphone is a metaverse. But the term has its origins in Virtual Reality, and there is good reason for that. Presence should offer more than control of an avatar, you should feel like you are there... "in the game" as the tag-line goes.
Besides, such a broad definition makes the idea of metaverse moot. Who cares? You cannot innovate with a tabula rasa, the first step is always to define constraints. After all, if that's the scope of the definition -- basically anything with an avatar that maybe, has or could have (one day?) a blockchain zipping through it -- then Super Mario Bros. is a metaverse too right? After all, we are controlling a character which is like an avatar (albeit not customizable) and it engages us in a virtual world... the mushroom kingdom.
However, we don't actually feel like we are Mario nor do we feel like we are really there in the mushroom kingdom. Without that true sense of presence, the feeling like we are really there in a virtual world, you can't call Super Mario Bros. a metaverse, no matter if Nintendo publishes a new version of Super Mario Maker on Etherium or builds a Nintendo Land in The Sandbox.
If there is no presence, there is no paradigm changing innovation to how we experience technology.
The metaverse needs presence. Today those presence-driven use cases are being created most vividly in Virtual Reality. Most of these are on Quest 2, or available through Steam for PC VR devices; but many interesting experiences are also being published to the web via WebXR, and Sony is... trying (11 titles though, really?) so consoles are a potential platform as well. Regardless of the platform, use cases that provide a truly distinct sense of presence are mostly in VR.
And that's okay. VR is an exciting market that is growing rapidly, faster than any other form of digital media, and is being adopted by teens which indicates it will quickly capture larger swaths of the main stream soon. Sure, AR is coming and there are other interesting ideas like Peripherals (today, remote pilot Drones) that could also potentially be metaverse use cases.
But the one we know for sure is sticky and new, is VR.
Rather than dream of trillions of dollars in market opportunity, what is the nearer-term addressable market for VR? Is Zuck just off his rocker, or is he investing billions because he sees a potential market? The answers are Big Enough, and Yes (um, he sees a market that is).
Discard your need to compare everything to the iPhone and Smartphones. Everyone with a pair of shoes has a smartphone, it's a pointless exercise to compare (sort of like calling everything the metaverse... no constraints, means no innovation). Quest 2 has already outsold Xbox, and all together VR headsets will probably outsell consoles generally in the next couple years. Pretty much anyone who plays games in VR would probably agree with me, all the cool new games are in VR so... yeah. VR headsets should approach tablet-scale relatively soon after that, as new use cases emerge with better, lighter, more connected hardware, and as teens grow-up and need to decide "Should I buy a Mixed Reality Headset, or a Notebook PC?" many will start to choose multiple large highdef screens on a device that doubles as a kick-ass video game console over a stale form-factor from the 90s.
Some analysts predict by 2030-35ish(?) these new devices will scale to a billion sales per year. Honestly, I don't know, that's a bold prediction. I believe it, I want it to happen, but I just don't know.
My point is, that doesn't matter at all. VR content is selling, Meta reported $50M in store sales back when they'd only sold 10M units and there were only 200 games available. Even relatively conservative estimates would have total VR device sales at 75M~100M units per year in the near future. Meta's annual store revenue would be at least in the $10-15B range assuming they keep their dominant marketshare and continue to expand their titles and purchasing options (subscriptions, DLC, etc). Since technology scales, especially software which they've bought a lot of, the back-of-the-envelope math shows a path to profitability for Meta and anyone building presence-driven use cases in VR alone.
So what do you need to know about the Metaverse today? It's about presence, presence is driven by VR today, and VR is a big enough market to make commitments without all the other "metaverse" stuff.
Sony's PSVR2: More Stutter (update on prior post)
PSVR2 is finally going on sale on February 22, 2022. That's a lot of twos for the PSVR2... 220222. It's also a palindrome, okay kind of cool.
But as reported in The Verge the price is $549, so together with the PS5 that's around $1000 for a tethered headset that can only be played with Playstation content. And, since it's tethered, only in whatever room you happen to have the PS5 placed.
Not long ago, they also announced they only sold 20M PS5 units (although The Verge reports they Sony updated that number to 25M) and their production target for the PSVR2 would be 2M units. That's a sad number, almost as sad as Pico 4 selling less than 50K units in a month in the world's largest consumer market, China. While I get the logic behind the 2M units, it's basically double the attachment rate of 5% that PSVR had to the PS4; the overall numbers prove to me one thing: the era of console gaming is coming to close, the era of VR gaming has begun.
Quest 2 has already sold more units than either the Xbox (both S and X combined) or PS5. If there were holdouts for the PSVR2, the Quest 2 sales may jump even further come March 2023 unless Sony can demonstrate the content really is good enough to justify having PSVR2 rather than getting a headset with a ton of fun games and more being added every month.
I get it, everyone hates Mark Zuckerberg and Meta is a stupid name (not my opinion, but that's how a lot of people seem to feel about it). But the bottom line is, Quest 2 is THE NEXT THING in gaming and both Sony and Microsoft's offerings are stale.
Sony Is Coming but with More Stutter than Swagger
Based on a recent article on Road to VR, Sony reduced its P25 Sales forecast for 2022, and the PSVR 2 delayed to 2023:
https://www.roadtovr.com/report-psvr-2-rumored-release-early-2023-missing-holiday-season/Estimated sales through EOY will be around 30M units. Assuming an additional 15M units for 2023 we can estimate PSVR 2 capture at a high, mid, low sales volumes.
45M PS5 Sales EOY 2023 - PSVR2 Sales expectation assuming no production delays:
Low (5%) 2.25M
Mid (10%) 4.5M
High (20%) 9M
That said, given their current ability to produce consoles -- which is a strange failure on their part to predict Semicon constraints, strange since they are famously building a commercial EV -- it's unlikely Sony can produce sufficient numbers of units. A revised number accounting for production delays could significantly lower their sales potential for PSVR2 through 2023:
Low (1%) 450K
Mid (2.5%) 1.17M
High (5%) 2.25M
Either way, this places PSVR2 squarely in the top seat along with Meta's HMDs and pits the companies head-to-head for dominance in the VR market moving forward. However, considering the laggard pace of console development versus the break-neck speed of Meta's device rollout (new ware every two years), Sony's got its work cut out for it. PSVR2 will no doubt make a dent in Meta's share of the HMD market (currently 70% or thereabout), but a better question perhaps is can Sony pivot to an independent "Metaverse Console" that can match the pace of innovation? Naturally, they have the capacity to do so as demonstrated by their mobile phones division, but whether or not they can transfer that capacity to this new market remains unclear and, frankly speaking, doubtful.
Horizon Worlds on a Healthy Path
Meta’s “social VR platform” Horizon Worlds hit the news over the weekend announcing 300k (10X growth) after 90days.
This is a good accomplishment… some context for why that is. On mobile the gold standard was (is?) 1M MAU after 90days. That’s a much lower penetration considering anyone with a pair of pants has a smartphone.
For perspective, assuming 70% of activations are in the USA and Canada (the only market where Horizon Worlds is publicly available), and the rumor 10M units sold is accurate, then Worlds has around 5% penetration to the TAM. Not bad… let’s see what happens.
Here is the Verge Article reporting on this:
https://www.theverge.com/2022/2/17/22939297/meta-social-vr-platform-horizon-300000-users
A bigger question is how will this platform compete with RecRoom or VR Chat.
In particular, here is an article that discusses RecRoom as a benchmark. It has 1M MAU (available worldwide and on Steam) and supposedly “millions more” on non-VR platforms incl. Xbox.
Will Horizon Worlds become a cross-platform game? I think so, and - assuming it is fun and interesting- this will drive the MAU much higher as it expands the TAM to include mobile and online users.
Beat Saber Queen - 2021
Beat Saber Queen in 2021 with AfreecaTV was a major milestone in building the VR Gaming ecosystem in Korea. We started that journey at the beginning of the year, and finally pulled it off with the help of our friends and Meta, Beat Games, Onestore, and of course our production partners AfreecaTV.
This was the first program for Beat Saber in the world. I hope we can produce this program properly in 2022, let's see how it goes! Who knows, this could be the next "Masked Singer"? (For those who don't know it's a Korean program that was licensed to a US producer.)
Check it out below on Koon's YT channel, skip to around 20min to see me play "Reason for Living" on Expert Plus.
You can also check out the full program AfreecaTV:
https://vod.afreecatv.com/PLAYER/STATION/77408486?change_second=741.
Michael's Metaverse for Dummies
In conclusion...
Michael's Metaverse for Dummies by Michael A. Lesniak is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
Based on a work at /www.malesniak.com//2020/09/blog-post.html.