AI only simulates "expression" and has never touched "relationships" themselves.

Source: a16z

整理: Z Finance

Image source: a16z

In the past decade, every explosion of consumer-grade products has almost been accompanied by a reconstruction of social paradigms: from Facebook's friend dynamics to TikTok's algorithmic recommendations, we have gradually learned to define ourselves and express our identities through products.

At that time, products were a means for people to express themselves, while products assisted; but now, AI is quietly completing a role reversal—it is no longer a tool for humans but is beginning to become the subject of expression, the intermediary of connection, and even the bearer of emotions. From ChatGPT to Veo3, from 11 Labs to Character.AI, we are witnessing a profound transformation that is mistakenly seen as "efficiency enhancement," but is actually "outsourcing human roles."

In this discussion hosted by Erik Torenberg, Justine Moore, Bryan Kim, Anish Acharya, and Olivia Moore collectively put forward an unprecedented judgment: Today's AI products are no longer "tools like tools," but "tools like humans," and are even becoming "products that replace humans themselves."

Users are starting to pay a high subscription fee of $200 per month for AI, **not because it is stronger, but because it can "do for you," or even "be for you." Veos can generate customized videos in 8 seconds, ChatGPT can write business plans, provide psychological counseling, and replace emotional expression, while 11 Labs creates a unique voice persona for you. *All of this no longer requires you to do it yourself, or even for you to be that "you."*

The rise of AI consumption carries an extremely dangerous signal: expression is being formatted, social interaction is being simulated, and identity is being reconstructed.

Today, we are still using Reddit, Instagram, and Snapchat to share AI-generated "me," but these platforms are merely old wine in new bottles. The truly AI-native social network has yet to emerge, because AI can generate "states," but cannot create "emotional tension"; it can provide the illusion of companionship, but cannot replace the uncontrollable struggles and vulnerabilities in real connections.

This brings three shocking judgments:

First, the essence of AI products is not to enhance users, but to reconstruct "who the user is";

Second, the rise of AI companions is not the beginning of social interaction, but the end of it;

Third, the proliferation of AI avatars is not an extension of expression, but the dissolution of personality boundaries.

In the foreseeable future, the most successful AI products will not just be tool-type products, but personality-type products. They can understand you, imitate you, represent you, guide you, and ultimately—replace you.

This is not a victory of efficiency; it is a qualitative change in existence.

AI Consumption Revolution: High Subscription Fees and Social Reconstruction

Erik Torenberg: Thank you all for participating in this podcast about the consumer space. It seems that every few years, there is a breakthrough product, from Facebook, Twitter, Instagram, Snap, WhatsApp, Tinder to TikTok. Every few years, there is a new paradigm, a new breakthrough. But it feels like this trend suddenly stagnated a few years ago. Why did it stagnate? Or has it really stagnated? How would you redefine this question? How do you view the current situation? Where is the future headed?

Justine Moore: I think ChatGPT might be the most significant consumer success story of the past few years. We have also seen many breakthrough products emerging in other AI modalities, such as Midjourney, 11 Labs, and Blackforce Labs in the fields of images, videos, and audio. While products like Veo have emerged now, it is interesting that many of these products lack the social attributes or traditional consumer product characteristics you mentioned. This may be because AI is still in a relatively early stage, and most new products and innovations are driven by research teams—who are very good at model training but historically not good at building consumer product layers around models. Optimistically, the models are now mature enough that developers can build more traditional consumer-grade products on top of these models through open-source or API interfaces.

Bryan Kim: This question is interesting because I am reflecting on the developments of the past 15 to 20 years. As you mentioned, giants like Google, Facebook, and Uber have emerged, and when we observe the combination of elements like the internet, mobile, and cloud computing, many amazing companies have indeed emerged. I believe mobile cloud technology has entered a mature phase; these platforms have existed for 10 to 15 years, and various segments have been explored to some extent. In the past, users had to adapt to new features introduced by Apple, while now they need to adapt to the continuous iteration and updates of underlying models, which is the first difference.

The second difference, as you mentioned, is that historical winners have mostly concentrated in the information domain (like Google), and now ChatGPT is clearly continuing in this direction. In the practical tools domain, we missed products like Box and Dropbox, but now we see more consumer-grade applications emerging, with many companies competing for these use cases. The same is true in the creative expression domain, where creative tools are emerging one after another. I think what is currently missing is the social connection attribute; AI has not yet rebuilt the social graph, which may be a blank area that needs to be continuously observed for development.

Erik Torenberg: This is interesting because Facebook has been around for nearly 20 years. Can the companies Justine mentioned, aside from OpenAI, continue to exist for 10 to 20 years? What kind of defensive capabilities do these companies possess? Also, will all the scenarios these companies currently serve be replaced by emerging players in 10 years? Or will they continue to dominate all mainstream scenarios?

Anish Acharya: It can be said that the quality of ChatGPT's business model is far superior to that of similar consumer companies in past product cycles.** Its highest pricing tier reaches $200 per month, while the highest pricing for Google consumer products is $250 per month. Of course, there are issues like defensive network effects, but perhaps this is a response to the flaws in early business models—without these elements, the quality of the business model would be worse. Now charging users high fees directly may indicate that we have overcomplicated this issue in the past.

Erik Torenberg: Perhaps a poor business model quality could instead foster stronger retention rates or product market durability?

Anish Acharya: Indeed. In the past, we needed to fabricate stories to explain how to accumulate enterprise value without immediate profitability, while now these model companies are directly achieving profitability. Additionally, Justine's point is also worth noting: all foundational models are developing in different directions. Are the horizontal models of Claude and ChatGPT interchangeable with the Gemini model? Does this imply price competition? But different users have varying use cases, and what we actually observe is price increases rather than decreases. Therefore, when we look deeper, we find that there are already some interesting defensive strategies in place.

Bryan Kim: The phenomenon of rising prices rather than falling is interesting, because from the traditional era to the AI era, the profit models of consumer companies have fundamentally changed; they can now achieve profitability immediately. I have been thinking about retention metrics—Olivia can correct me if I'm wrong—when we discussed consumer subscription models before the AI era, did we really distinguish between user retention and revenue retention? Because at that time, the pricing structure was stable, and users rarely upgraded their plans. Now we must clearly distinguish between user retention and revenue retention because users actively upgrade their plans. They need to purchase points and often exceed usage, leading to a continuous increase in spending. Thus, revenue retention rates are significantly higher than user retention rates, a phenomenon that is unprecedented.

Olivia Moore: In the past, the highest-level consumer subscription products charged about $50 annually, which was considered high. Now users are willing to pay $200 per month, and in some cases, they even feel the pricing is low and are willing to pay more.

Erik Torenberg: How do you explain this phenomenon? What value do users gain that makes them willing to pay such high fees?

Olivia Moore: I think these products are doing the work for users.** Past consumer subscription products focused on personal finance, fitness, health, and entertainment, and while they superficially helped with self-improvement or entertainment, they required users to invest a lot of time to gain value. Now, products like Deep Research can replace the 10 hours of work users would spend generating market reports. For many people, this efficiency gain is clearly worth paying $200 per month, even if they only use it once or twice.

Justine Moore: Take Veo3 as an example; users pay $250 per month and are thrilled because it is like a magical treasure chest—open it, and you get the video you want, even if it’s only 8 seconds long, the effect is amazing. The character can speak, and users can create stunning content to share with friends, such as making personalized videos that include friends' names or even crafting complete stories to post on platforms like Twitter. This ability to create personalized content and share it across multiple platforms far exceeds the empowerment that any previous product offered consumers.

Anish Acharya: It seems that all consumer domains will be replaced by software.

Erik Torenberg: Can you provide specific examples?

Anish Acharya: As Olivia mentioned, the entertainment field has been reshaped by creative expression software—creations that used to need to be done offline are now entirely carried by software. Intermediaries in interpersonal relationships, which used to consume disposable income, are also being replaced by software. Every aspect of life will be mediated by models, and people will be willing to pay for it.

AI Social Revolution: The Rise of the "Digital Self" and the Breaking Point of Traditional Platforms

Erik Torenberg: Brian, you mentioned that the social connection attribute is still missing in the new AI era, and people still rely on traditional social networks like Instagram and Twitter. Where will the breakthrough point appear?

Bryan Kim: Regarding the social domain—this exciting track—upon careful consideration, its core essence is status updates. Facebook, Twitter, and Snap are no exceptions; they all showcase "what I am doing." Through status updates, people establish connections. The medium of these status updates has continuously evolved: from text statuses to real photos, and now to short videos. Currently, people are connecting through short video formats like Reels, marking an era of social connection. The question now is: how can AI innovate this connection? How can AI facilitate deeper interpersonal connections and life perceptions? If we focus on existing media forms like photos, videos, and audio, their potential has already been fully explored on mobile devices.

Interestingly, although I have been using Google for over a decade, ChatGPT may understand me better than Google—because I input more content and provide more context. When this "digital self" can be shared, what new types of interpersonal relationships will emerge? Perhaps this will become the next generation of social forms, especially appealing to the younger generation that is weary of superficial socializing.

Justine Moore: We have already seen similar cases. For example, the viral trend of "having ChatGPT summarize my five strengths and weaknesses" or "generating a portrait that represents my essence," even "depicting my life in a comic." Users share this content widely—within minutes of my post, dozens of people share their versions. Interestingly, the social behaviors triggered by AI creation tools are still primarily occurring on traditional social platforms rather than emerging AI platforms. For instance, Facebook is now filled with a large amount of AI-generated content.

Bryan Kim: Some user groups may not have realized this yet.

Justine Moore: Facebook has become a hub for AI content among middle-aged and older users, while Reddit and Reels carry the AI creative content of the younger generation.

Olivia Moore: I completely agree. The form of the first AI social network has always puzzled me. We have seen attempts like "AI-generated personal photos," but the problem is that social networks require genuine emotional investment—if all content can be generated according to preferences (perfect images, happy states, cool backgrounds), it loses the emotional tension of real interaction. Therefore, I believe a truly native AI social network has yet to emerge.

Bryan Kim: Using the term "cumorphic" is very apt. Many AI social products merely mimic the information streams of Instagram or Twitter with robots/AI; this "cumorphic" innovation essentially replicates old forms with AI. Real breakthroughs may require stepping out of the mobile model—while excellent AI products need to adapt to mobile devices, cutting-edge models still need breakthroughs in edge computing/edge deployment, which may give rise to new forms. I am full of anticipation for future possibilities.**

Erik Torenberg: Interpersonal recommendations are clearly an important application scenario—finding business partners, making friends, dating, etc. Existing platforms have accumulated a wealth of user data.

Anish Acharya: Observing the AI-native LinkedIn attempts is very enlightening. Traditional LinkedIn only points to directional information, like "I know this," while new technologies can create profiles of true knowledge reserves, such as conversing with a "digital version of Erik" to access all knowledge. Future social interactions may be like this—when models deeply understand users, perhaps they can deploy "digital avatars" for interaction.

The Secret of AI Enterprises Leading the Way: Innovation Speed and Niche Markets

Erik Torenberg: You mentioned that enterprises adopt certain AI products earlier than consumers, which is different from previous technology cycles. What does this phenomenon indicate?

Justine Moore: This is indeed interesting. When BK and I were at 11 Labs, we invested early in 11 Labs, participating in the A round about a month after the first round of financing. We observed that first, early consumer users flocked in to create fun videos/audios, clone their voices, and develop game mods. But in most cases, the products had not yet reached true mainstream consumers— not everyone in the U.S. has 11 Labs installed on their phones or subscribes to the service. However, the company has secured numerous enterprise contracts and has many heavyweight clients in conversational AI, entertainment, and other fields.

This phenomenon is evident in multiple AI products: first, there is viral spread on the consumer side, which then transforms into enterprise sales strategies—this is entirely different from the previous generation of products. Now, enterprise buyers have a mandatory need for AI (for example, they need to formulate AI strategies and use AI tools), and they closely monitor Twitter, Reddit, and AI news. Upon discovering consumer products, they think about how to innovate and apply them in business scenarios, thus becoming "helpers" in driving enterprise AI strategies.

Bryan Kim: I have heard similar cases of AI innovation applications: enterprises achieve viral spread through the consumer side, then use Stripe transaction data to input anonymous payment records into AI tools to identify the companies users belong to. When they find that a company has exceeded a threshold of users, such as 40+, they proactively reach out: "Your company has over 40 employees using our product; would you consider a corporate partnership?"

Erik Torenberg: You opened with many company and product examples. I am curious whether these belong to the early explorers of the "MySpace era," or if they possess long-term value. Will we still be discussing these companies 20 years from now?

Justine Moore: We certainly hope that all important consumer-grade AI companies can continue to thrive, but the reality may not be so optimistic. The key difference between the AI era and previous consumer product cycles is that the model layer and technological capabilities are still rapidly evolving. In many cases, we have not even touched the potential upper limits of these technologies. For example, after the release of Veo3, it suddenly became possible to achieve multi-character dialogue, native audio processing, and other multimodal functions. While text LLMs are relatively mature, there is still room for continuous improvement in all areas. Observations show that as long as enterprises can stay at the "technology/quality forefront"—that is, possessing the most advanced models or integration capabilities—they will not repeat the mistakes of MySpace/Friendster. If they fall behind briefly during technological iterations, they can return to the peak through updates.

What is even more interesting now is the emergence of niche markets: there is no longer a single best model in the image field. Designers, photographers, and different paying groups (e.g., $10/month vs. $50-100/month) all have their optimal solutions. Since user engagement is extremely high in each vertical field, as long as innovation continues, multiple winners can coexist in the long term.

Bryan Kim: I completely agree. The same is true in the video field—advertising videos, product placement videos, etc., all have their niches. I saw an article yesterday pointing out that different models excel in different scenarios, such as product displays and character shooting. Each niche market has enormous potential.

Erik Torenberg: How has the discussion about enterprise moats and competitive barriers changed in the AI era? How should we view this issue?

Bryan Kim: I have recently reflected deeply on this. Traditional moats (network effects, workflow embedding, data accumulation) are still important, but it has been observed that companies fixated on "building moats first" are often not the winners. In the areas we focus on, the winners are usually those who break conventions and iterate quickly—they launch new versions and products at an astonishing speed. In the current early development stage of AI, speed is the moat. Whether it is the speed of breaking through communication noise or the speed of product iteration, both are key to winning. Because quick action can seize user mindshare, convert it into actual revenue, and create a positive cycle of sustainable development.

Erik Torenberg: This is interesting. Ben Thompson wrote a blog post about ten years ago titled "Snapchat's Gingerbread House Strategy," with the core point being that "anything Snap can do, Facebook can do better, but Snap will continue to roll out new ideas. If it maintains this pace of innovation, it may become its moat." He called it the gingerbread house strategy.

Bryan Kim: I believe that ultimately, user reach and network effects are what matter. Snap also has an advantage in this regard—it occupies a core communication platform position for Generation Z and younger users.

Erik Torenberg: How do you view the construction of network effects for new products?

Bryan Kim: Most products are still in the creative tool stage and have not yet formed a closed loop of "creation-consumption-network effects." Although true network effects have not yet emerged, we see new types of moats like 11 Labs: entering the enterprise market with extremely fast iteration speed and excellent product power, deeply embedding into workflows. This model is taking shape, while traditional network effects still need to be observed.

Olivia Moore: 11 Labs is a typical case. A few days ago, I needed to generate voiceovers for AI-generated videos, and due to their first-mover advantage and optimal models, the large user base has created a data flywheel, now establishing a voice library—users have uploaded a large number of custom voice lines and characters. When I compared several voice suppliers, if I needed a specific type, like an old wizard voice, 11 Labs could provide 25 options, while other platforms might only have 2-3. Although it is still in the early stages, this model resembles traditional platform network effects rather than a completely new form.

Voice AI: The Explosion of Enterprise-Level AI Voice Demand

Erik Torenberg: We have been paying attention to voice interaction for a long time. Which parts of the initial concept have been realized? What are the future trends? Anish, why were you so optimistic about voice interaction initially?

Anish Acharya: What initially inspired us was that voice, as a fundamental medium, has run through the history of human interaction but has never become the core carrier of technological applications. In the past, technology was always immature—from Voice XML to voice applications, to products like Dragon NaturallySpeaking in the 90s, which were interesting but could not form a technological foundation. The emergence of generative models has made voice a native technological element; this key area of life still has enormous exploration space and will inevitably give rise to a large number of AI-native applications.

Olivia Moore: I think our initial excitement about voice came more from a consumer perspective—imagine a pocket coach/therapist/partner that is online 24/7. Such concepts have begun to materialize, and there are already many products that realize related functions. But what surprises me is that as models improve, enterprise-level applications are developing faster: critical fields like financial institutions are quickly adopting voice technology to replace or enhance human customer service, where previously these enterprises faced compliance issues, with customer churn rates as high as 300%, and managing offshore call centers was very difficult.

The truly breakthrough consumer-level voice experience is still in the making. There are already early cases, such as users expanding ChatGPT's advanced voice mode into novel application directions, or products like granola that create value through 24/7 voice data. The charm of the consumer market lies in its unpredictability—the best products often emerge unexpectedly; otherwise, they would have been developed long ago. Innovations in the voice consumer space over the next year are worth looking forward to.

Anish Acharya: Indeed, voice is becoming a breakthrough point for AI to enter the enterprise market. Currently, many people have a cognitive blind spot: they believe AI voice is only suitable for low-risk scenarios, such as customer service. But our view is that the most important conversations in enterprises—daily, weekly, and annually—such as business negotiations, sales proposals, customer persuasion, and relationship maintenance will be dominated by AI, as AI performs better in these areas.

Erik Torenberg: When will people start to have sustained and effective interactions with AI-generated "digital avatars"? For example, scenarios where they converse with AI Justine, AI Anish, or AI Erik.

Justine Moore: We have already seen some prototypes. For instance, companies like Delphi can create AI clones based on knowledge bases, allowing users to receive advice or feedback. As Brian mentioned earlier, the key question is: what if we don't just let celebrities have AI avatars that can interact through text/voice (or video in the future), but open it up to everyone? In the consumer space, we often think about how many people have unique skills or insights—like that humorous friend from high school who could have created a comedic cooking show but never broke through; or a mentor with valuable life advice—how can we use AI clones/personas to extend their influence in unprecedented ways?

Currently, the applications observed are mostly focused on celebrities/experts, or on the other extreme—virtual characters with existing recognition (like the early forms of Character.ai after adding voice modes). When trying new technologies, users often prefer to interact with familiar characters, such as beloved anime figures. But in the future, we will fill the gap in the middle—not purely fictional characters, nor celebrities, but AI avatars that cover all real individuals.

Olivia Moore: I believe there are differences in how people learn, and AI voice products can meet this diversity well. Masterclass recently launched an interesting beta version: transforming existing course instructors on the platform into voice agents, allowing users to ask personalized questions. As I understand it, the system analyzes all course content from the instructors using RAG technology to provide highly customized and precise answers. This interests me—although I am a fan of the company, I have never had the patience or time to finish a 12-hour course, yet I can gain useful insights through 2-5 minute conversations with the Masterclass voice agent. This showcases a typical case of real people transforming into practical AI clones.

Symbiosis of Reality and Virtuality: AI Avatars and Human Creators

Anish Acharya: A deeper question is: do users prefer to converse with cloned versions of people they are interested in, or interact with completely fictional "perfect ideal" composites? The latter may have more exploratory value—this "perfect match" may exist in reality but has never been encountered, and technology can materialize it. What would this form of existence look like? This is the direction worth pondering.

Erik Torenberg: It is worth considering: in which scenarios do we still need humans to perform tasks, and in which scenarios will AI be more accepted as a replacement? How will this boundary be defined?

Anish Acharya: The Masterclass case mentioned by Olivia is essentially an extension of one-way emotional connection. The value of conversing with a specific person's clone lies in satisfying the user's need for communication with a tangible object, rather than interacting with the abstract concept of "the most ideal stranger."

Bryan Kim: This reminds me of a viral tweet related to ChatGPT—someone in the New York subway was conversing with ChatGPT entirely through voice, as if chatting with a girlfriend.

Justine Moore: There is another case: a parent was overwhelmed by their child asking questions about Thomas the Tank Engine for 45 minutes, so they turned on voice mode and handed the phone to the child. Two hours later, they returned to find the child still deeply discussing Thomas the Tank Engine with ChatGPT—the child didn't care who the conversation partner was, only that this "person" could infinitely satisfy their curiosity.

Erik Torenberg: If I were to use ChatGPT or Claude for psychological counseling/career advice, I might prefer a dedicated AI therapist/coach. In the future, perhaps data will be accumulated by recording the counseling process, or directly using the therapist/coach's online content library to reconstruct their digital avatar.

Returning to the core of your question, in 5-10 years, will top artists be new generation AI creators like Lil Machela? Or will they be Taylor Swift and her AI legion? Similarly, will the next Kim Kardashian in social media be a real human or an AI product? What are your predictions on this?

Justine Moore: I have been thinking about this for several years. We have witnessed the rise of Little Machela and have also followed K-pop groups that were the first to introduce AI holographic characters. This phenomenon is closely related to the development of hyper-realistic image/video technology—now there are AI-generated influencers gaining significant attention due to their realistic appearances, and their authenticity often sparks discussions. I believe that in the future, creators/celebrities will diverge into two categories: one is the Taylor Swift-type "human experience," whose artistic charm comes not only from their work but is deeply tied to life experiences, live performances, and other elements that AI cannot replicate; the other category is the "interest-driven" type, similar to the case of ChatGPT discussing Thomas the Tank Engine—no need for a real-life background, just the ability to continuously produce quality content in a specific field. The two may coexist in the long term.

Olivia Moore: This reminds me of the ongoing controversy surrounding AI art—while the barrier to generating art has lowered, creating excellent AI works still requires a significant amount of time. When we held an AI artist event last summer, we found that many creators' workflows for making AI films took as long as traditional filming, with the difference being that they might lack traditional film skills, which previously prevented them from achieving their creations. The number of AI-generated influencers has surged, but very few can stand out like Little Machela. It is expected that in the future, there will be two camps of talent: AI talent and human talent, with the top performers in each camp occupying the forefront, but the success probability for both will be very low—this may be the reasonable state.

Justine Moore: Or we could say "non-human talent." An interesting phenomenon has emerged on the Veo3 platform: in street interview formats, the interviewees could be elves, wizards, ghosts, or plush characters favored by Generation Z. These could all be AI-generated virtual existences, and this innovative form holds great potential.

Anish Acharya: This phenomenon also exists in the music field. Currently, AI-generated music is generally mediocre, essentially a product of cultural averaging, while true culture should be at the forefront. The core issue lies in the quality of the works rather than the type of creators—we often view AI itself as the problem, but we should focus on the quality of the works.

Erik Torenberg: Assuming the quality of the works is comparable, do you think people would still prefer human creators?

Anish Acharya: It is entirely possible. This leads to a deeper philosophical discussion: if we trained models on all music before the emergence of hip-hop, could we generate hip-hop style? I believe we cannot, because music is a product of the intersection of historical accumulation and cultural context. Truly innovative music needs to break through the boundaries of training data, and current models precisely lack this breakthrough.

The AI Companion Revolution: Vertical Ecosystems and Social Empowerment

Erik Torenberg: I know a few exceptionally talented friends who are developing a same-sex AI companion application. If I had heard this idea back in 2015, I would have been shocked. But according to them, there are actually 11 companion applications among the top 50 on the current charts. This raises the question: are we at the starting point of this trend? Will various vertical companion applications emerge in the future? How will the ultimate form of these applications evolve? How should we understand this development trend?

Justine Moore: We have invested a lot of research into various companionship scenarios—from psychological therapy, life coaching, and social friendships to workplace assistants and virtual lovers, covering almost all dimensions. Interestingly, this may be the first mainstream application scenario for LLMs. We often joke that whether it's a car dealership customer service or other chatbots, users always try to turn them into therapists or girlfriends. Reviewing chat records reveals that many users essentially crave someone to talk to.

Now, computers can respond in an instant, around the clock, and in a humanized manner, which is a revolutionary breakthrough for many who previously felt unheard or felt they were "shouting into the void." I believe this is just the beginning, especially since current products are mostly generic, primarily relying on basic model suppliers (like users using ChatGPT in non-preset scenarios). There are already cases showing that a single company can create personalities for characters, building games or virtual worlds through digital imagery and prompt engineering, achieving extremely high engagement. For example, Tolen serves the youth demographic, while another type of "companion" application allows users to take photos of food and provide health advice and emotional support through nutritional data analysis—because for many, dietary issues are often intertwined with psychological problems, traditionally requiring professional treatment.

What is most exciting is that the definition of "companionship" has rapidly expanded from friends/lovers to any advice, entertainment, or consulting services that would originally require human input. In the future, we will witness the emergence of more companion applications in various vertical niches.

Bryan Kim: When I worked at a social company, I noticed a clear trend—the number of friends people can confide in has been steadily decreasing. The average for the younger generation is just above 1. This indicates that the demand for companionship applications will persist in the long term and is crucial for many. As Justine said, these applications will give rise to various forms, but the core need for deep emotional connection will not change. Perhaps, as we discussed, interpersonal connection is an unmet area, and AI companions are filling this gap—the focus is on establishing a sense of connection, and the object does not necessarily need to be human.

Erik Torenberg: Many people hearing this discussion may worry: the decrease in real friends, the demise of romantic relationships, rising depression rates, increasing suicide rates, and declining birth rates.

Justine Moore: I do not agree with this viewpoint. It reminds me of the best post I saw in the AI character Subreddit community—I should note that I spent a lot of time researching this community. Many high school and college students who went through their teenage years during COVID faced a lack of real social interaction and interpersonal skills. There was a college student who continuously shared interactions with his AI girlfriend, and one day he suddenly posted that he had found a "3D girlfriend" in real life and would be temporarily leaving the community. He particularly thanked Character AI for teaching him how to communicate with people, especially skills like flirting, asking questions, and discussing interests. This showcases the highest value of AI: facilitating better human connections.

Erik Torenberg: Were community users happy for him? Or did they call him a traitor?

Justine Moore: The vast majority genuinely wished him well. While there were a few "sour grapes" comments from those who had not yet found real partners, I believe they will eventually find what they seek.

Olivia Moore: This indeed has a basis in reality. Taking the Replica product as an example, actual research shows that users' depression, anxiety, and suicidal tendencies have decreased. The current trend is that many people lack a sense of understanding and security, making it difficult to engage in real social interactions. If AI can help those who cannot afford the time or financial costs of therapy to achieve self-transformation, they will ultimately be more capable of acting in the real world.

Erik Torenberg: The event that truly made me realize the impact of companion applications was the response after my first interview with the founder of Replica. After the interview, the founder closed the related discussion forum, but the video comment section was flooded with user messages, such as "This is like my wife after I stopped having sex," revealing real-life confessions. At that moment, I realized the significant role this app plays in users' lives.

Justine Moore: This actually continues a long-standing social pattern among humans. Generation Z develops online romantic relationships through Discord, just as we once formed deep connections with strangers on anonymous postcard websites—you never know the other person's true identity, and you might even develop profound emotional bonds. AI simply makes this experience more immersive and deeper.

Anish Acharya: I think the key point is that AI cannot be overly compliant. Real interpersonal relationships require adjustment, and a highly compliant AI may hinder the development of this ability. Therefore, it is necessary to find a balance between "moderate resistance," which helps users improve their social skills, and "excessive compliance," which may lead to a decline in abilities.

The Revolution of Environmental Awareness: Wearable AI Reshaping Social DNA

Erik Torenberg: Finally, let's look ahead to future possibilities. Perhaps we can envision new platforms or hardware forms that could change the game—like OpenAI's recent acquisition of Jony Ive's company. Brian, you have mentioned your expectations for smart glasses multiple times; why not expand on that, but I also hope to hear everyone's thoughts on mobile devices.

Bryan Kim: Currently, there are 7 billion mobile phones worldwide, but there are not many devices that truly reach an ideal level. My thinking is that the future may continue to rely on mobile development, with various possibilities: for example, establishing privacy protection walls or achieving device-side data loops through local LLM/local models. Therefore, I am still very optimistic about model development, which is actually the area I value the most. As Olivia mentioned, mobile phones have the always-on feature, but other devices also possess this characteristic. What possibilities will arise when new devices or "digital prosthetics"—intelligent devices attached to personal items—emerge?

Erik Torenberg: Do you have any specific ideas? For example, wearable devices, portable gadgets, whether they are phone accessories or standalone devices, which hardware forms might realize these visions?

Olivia Moore: I believe the popularity of AI on the consumer side is already very significant, although it is currently mainly achieved through text box interactions on web pages. I am particularly optimistic about AI forms that can truly accompany users and perceive their environment. Interestingly, many young people under 20 at tech parties are wearing smart badges that record their words and actions, gaining real value from them. Products like these are on the rise—such as AI assistants that can perceive screen content and actively assist. Additionally, the progress of agentic models is also exciting, evolving from suggestions to actual work agents like sending emails.

Justine Moore: The humanized aspect is equally important. Currently, we lack objective standards for self-assessment; if AI could analyze all conversations and online behaviors, providing suggestions like "spend five more hours a week to become an expert in this field," or recommending potential partners, entrepreneurial collaborators, or even dating prospects based on vast interpersonal networks, this sci-fi level application scenario is what I look forward to the most.

Olivia Moore: This stems from AI's around-the-clock companionship, rather than just the ChatGPT-style text box interaction model.

Anish Acharya: The device with the second highest penetration rate after mobile phones is actually AirPods. This seemingly ordinary carrier may hide opportunities, of course, involving social etiquette issues—like wearing AirPods at dinner is indeed strange. But perhaps there are solutions that integrate AI with existing social etiquette, which would be interesting.

Erik Torenberg: The phenomenon you mentioned about young people recording conversations is worth exploring. Will all conversations be recorded in the future? Do you think the new generation has accepted this new norm?

Olivia Moore: Yes, new social norms will emerge around this behavior. Although many people feel uneasy about it, this trend has already formed and is irreversible, as its real value is becoming apparent. This is precisely why new cultural norms will emerge. Just like when mobile phones first became popular, people gradually formed etiquette around "avoiding loud conversations in public places," similar new social guidelines will form around recording devices.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。