Since the explosion of AI last year, many people feel that all fields can be reshaped with AI, and this is the case in the social field that everyone valued before. AI social products are emerging in an endless stream, and everyone wants to get a piece of the social pie, but what is the effect of AI plus social networking?
From the point of view that existence is reasonable, the market is indeed constantly launching a lot of AI social apps, and there are many voices discussing. I've looked at some of them, but I feel like what they're talking about doesn't touch the essence of the product.
It just so happens that the author has not only been responsible for traditional matchmaking social apps, but also has rich trading experience in AI products. I wondered if I could break down the AI social product from the perspective of Product Market Fit. The products discussed in this discussion mainly include Hoshino, Glow, Tsukiyujima, Character AI and other leading AI social products at home and abroad, and the products involving male edge rubbing are not within the scope of discussion for the time being.
1. Target users
Although the target users of traditional matchmaking social apps are single men and women, the fact is that "high-quality women" are the so-called "high-quality resources", and the source of cash flow is single men. The male-to-female ratio of most social apps is 2:1, or even 3:1.
In AI social software, most of the target users are Generation Z women, who have little love experience or are mainly single mothers and fetuses. The main types of interests are subcultures such as two-dimensional and PIA opera. The reason is closely related to user needs and the current product form.
In terms of target users, the users of these two types of products almost do not coincide.
Second, the core experience
AI social mainly includes the following three core user experiences:
- Custom Virtual Roles/Use Virtual Roles Defined by Others
- Engage in conversational interactions with virtual characters
- A card experience derived around agents
Among them, the most valuable experience occurs in the part of immersive interaction with virtual characters such as text and voice.
Traditional social apps mainly include the following core user experiences:
The most valuable experiences happen in the parts where the two parties are able to have a continuous conversation.
Although the experience of both takes place in conversations, the supply of "ideal" in traditional social apps is limited, and the topic is biased towards real life.
The supply of "ideal" in AI social networking is unlimited, and topics can also break through the three-dimensional dimension and become more imaginative. The latter is largely free from the problems of dialogue pressure and social stereotypes.
Third, the demand is satisfied
According to the core experience of AI social products, the satisfaction of users' needs is analyzed, which is mainly divided into the following categories:
1. Deeply customized face pinching
It corresponds to the projection of the user's desire to create and a certain emotional desire. For example, you can bring your favorite appearance, voice, occupation, personality, etc. into customized AI characters.
2. 浪漫角色扮演（Romantic Roleplay）
It corresponds to the emotional projection of the sexes that cannot be realized in reality. The user's demand is to experience idealized love and friendship stories, such as bosses, gangsters, sick women, werewolves, and so on.
3. Hope to be recognized by the small circle of friends
Users like to recommend their own agents in social networks, as well as the feeling of dialogue with them, showing a strong tendency to be recognized by others, but this need to be recognized is only limited to the current circle.
Most of the needs met by traditional social apps are hormonal needs in short-term relationships, which may include some love needs.
To sum up, from the perspective of demand, the demand for AI social satisfaction is indeed real in the small circle.
However, once the crowd expands (for example, to non-ACG enthusiasts), the current needs and solutions may not match the users. That is, if the current (in that direction) demand continues to be met, the upper limit of the user scale is obvious.
Fourth, the business model
In addition to the problem of user scale caused by niche demand, AI social products also have certain problems in the business model.
The current commercialization model of AI social products may be divided into the following types:
1. Value-added experience
That is, the subscription service, then the problem is to focus on the membership benefits to create items that are strongly related to the core experience, but the fee is not objectionable to users. At present, the AI social products of various heads do not give particularly strong conversion projects in this part
2. Build a trading system
By building a trading ecosystem, users can purchase props and derivative products such as Talkie's AI card trading marketplace. This is the classic monetization model of otome games. But there are differences
1) The difference between PGC and UGC
The reason why the cards in the otome game have a sense of high value is the result of the operation of the game publisher's artists, marketing, festivals, etc., and is a "limited edition product". However, the cards in AI social are AI paintings, which can essentially be drawn an unlimited number of times, and the quality is lower than that of the artist. Therefore, the inability to monopolize production will lead to a decrease in the value of the trading system, and it is not easy to achieve high fee transformation.
2) Lack of inter-user transaction system
In addition to the above-mentioned problem of weak transactions between users and platforms. Since users only interact with AI agents, there is also a lack of demand for transactions between real users. The construction of the transaction system can only be based on the relationship between the AI agent and the user, and it is difficult to design transactions with high ARUP value.
3. To C业务做声量 To B业务做营收
For example, companies like Minimax that develop their own large models are also doing AI social applications. If the C-end application is not for the purpose of revenue, it is also a feasible path to obtain revenue by providing large model capabilities to other applications. However, other large-scale model buyers still have to consider the issue of monetization, especially start-ups, whose monetization pressure cannot be passed downward.
From the perspective of commercialization, the current head AI social products will not take the measurement of production ratio as the focus of the current work, and the current focus is still on scale.
After summarizing the above aspects, I will have an idea about the commercialization of the dialogue ability of large models, do AI social products have some inherent competition with otome games, or other RPG games?
Advantage:
1. From the perspective of enterprises, AI social products have a better cost model (the profit model has been mentioned above). The content of otome games generally includes card faces, skins, plots, etc., and gamers have very high requirements for content, and the cost of content production is very heavy. From PGC to UGC, the cost is reduced to zero, and the main expenditure of AI social products is the computing power cost brought by interaction.
2. Degree of personalization Theoretically speaking, the degree of customization of AI characters exceeds that of all previous RPG games, and there is no fixed character designated by the game company.
3. Freedom of user interaction
Any otome game has a clear main line and world view, which can give users relatively little independent decision-making power, and AI social interaction is more jumpy, we can see that many users in addition to confession, love and other conventional plots, but also developed plots such as jealousy, pregnancy, and childbirth...... It gives users a great deal of autonomy and room to explore.
Inferior position:
1. Memory Due to the limited ability of the large model itself to extract content from the context, applications such as Hoshino will also have problems such as a. forgetting the above, b. repeating the same paragraph, c. relieving from the persona, and so on. This is an inevitable problem brought about by the current stage of AI development.
2. Stability Due to the instability of the content produced by the large model (face collapse, read random answers, etc.), this instability has a relatively large damage to the immersion that is most important in role-playing. Although products like Hoshino have designed functions such as memory, backtracking, retelling, and evaluation to provide certain feedback and correction to the content given by AI, in terms of the current development of large models, there is definitely a distance from the silky dialogue of interactive novels.
Sum up:
1.AI social networking cannot be regarded as a false proposition in niche groups and niche circles, but when the circle is expanded, the universality of demand will decline.
2. From the perspective of business model exploration, it is still difficult to directly complete the 0-1 commercial verification through AI social products, and the revenue target consistent with the user volume has been achieved.
This article was originally published by @只做好20% of Things on Everyone is a Product Manager. Reproduction without the permission of the author is prohibited
The title image is from Unsplash and is licensed under CC0
The views in this article only represent the author's own, everyone is a product manager, and the platform only provides information storage space services