Read More《豆包手机助手调整权限:AI手机是洪水，但不是猛兽》
Positive Reviews: The AI Mobile Assistant Is the Prelude to an Efficiency Revolution, Driving the Upgrade of the Human-Machine Collaboration Model
The controversy triggered by the Doubao Mobile Assistant is, in essence, a technological exploration and ecological collision regarding “whether AI can deeply proxy users’ operations.” Despite facing numerous doubts at present, the technological value and future potential it demonstrates are still important signals driving the human-machine interaction revolution.
First of all, the core value of the AI mobile assistant lies in “liberating repetitive labor,” which is the fundamental rationality for its existence. As mentioned in the news, users’ core demand for AI Agents is not “how smart they are,” but whether they can replace mechanical operations with fixed rules but time-consuming processes, such as ordering food, grabbing tickets, and organizing messages. This demand is essentially the same as that for tools like dishwashers and vacuum cleaners – the original intention of human beings to invent tools is to liberate themselves from low-value repetitive labor and focus on creative activities. For example, when users use an AI assistant to automatically complete processes such as ordering takeout and paying bills, the time saved can be used for work or study. The value of this efficiency improvement for individuals and society cannot be underestimated. The “non-sensitive scenario proxy” function retained by Doubao after adjusting its permissions (such as daily message organization and form submission) precisely meets users’ real needs, reflecting the essence of technology serving people.
Secondly, the technological exploration of the AI mobile assistant has promoted the “capability interface transformation” of the mobile Internet ecosystem. The core logic of the traditional Internet ecosystem is the “interface entry” – users must enter apps through steps such as searching, clicking, and jumping, and each step is related to the platform’s traffic distribution and commercialization (such as advertising exposure and user stay time). However, the emergence of AI Agents has broken this logic: users only need to issue instructions, and the AI can directly call app functions to complete operations, bypassing traditional interface interactions. This change may seem to impact the existing ecosystem, but in fact, it forces Internet companies to shift from “relying on interface entries” to “opening up capability interfaces.” The future collaboration models such as the “MCP protocol” and “digital signature authorization” mentioned in the news are specific directions for this transformation. For example, Apple’s SiriKit already allows apps to directly respond to voice instructions through interfaces. In the future, the collaboration between AI mobile phones and apps may be even deeper – the AI serves as the “instruction center,” the apps provide “functional components,” and users can have the experience of “completing everything with one sentence.” This collaboration model will not eliminate the value of Internet companies. Instead, it will enable them to focus more on their core service capabilities (such as payment security and content quality) rather than competing for interface traffic, which is beneficial to the professional division of labor in the ecosystem in the long run.
Finally, the adjustments made to the Doubao Mobile Assistant reflect the “trial-and-error spirit” of technological exploration and have accumulated valuable experience for the industry. From initially opening up full-scenario proxy to actively restricting functions such as score brushing, financial proxy operations, and game competitions, Doubao’s dynamic adjustments are not compromises, but a cautious exploration of the technological boundaries. This strategy of “verifying first and then optimizing” respects users’ demand for efficiency and also responds to the concerns of ecological partners about fairness and security. For example, restricting proxy operations for financial applications avoids the risk of fund losses caused by AI misoperations; suspending the use of AI in game competition scenarios maintains the fairness of the game. These adjustments provide a reference for the standardization of subsequent AI Agents – technological innovation requires “bold assumptions and careful verification,” and Doubao’s practice embodies this principle.
Negative Reviews: The Ecological Conflicts and Potential Risks of AI Mobile Assistants Expose the Contradiction between Technological Advancement and Rule Lag
Although the technological value of AI mobile assistants is worthy of recognition, the current controversy it has triggered also reveals the deep-seated contradiction between technological advancement and rule lag. From the perspective of the existing logic of the mobile Internet ecosystem, the “ahead-of-its-time” nature of the Doubao Mobile Assistant brings at least three challenges.
Firstly, the conflict with the existing traffic system may shake the commercial foundation of Internet companies. The core revenue sources of traditional Internet platforms (such as advertising, memberships, and transaction commissions) highly depend on users’ interactions with interfaces: advertising requires exposure, memberships require users to actively browse content, and transaction commissions require users to complete processes through the platform. The “interface-free operation” of AI Agents may reduce the exposure rate of these links to almost zero – users do not need to click on ads or browse product pages, and the AI directly completes the order, resulting in a significant reduction in the platform’s advertising revenue and traffic distribution income. The statement in the news that “third-party applications cannot determine whether the permission is authorized by the user or induced by phishing software” essentially reflects the platform’s anxiety about “losing control of traffic.” For example, if users directly skip ads and place takeout orders through an AI assistant, the advertising revenue of the takeout platform will be affected; if the AI automatically selects the lowest-priced product, the value of the e-commerce platform’s “recommendation algorithm” will be weakened. If this conflict cannot be resolved, it may lead to Internet companies collectively resisting AI Agent technology, hindering its popularization.
Secondly, the controversy over fairness may pose challenges to public order and good customs. Taking the “red envelope assistant” as an example, the news points out that fully automated operation is essentially a “cheat” – WeChat allows mobile phones to notify users of red envelopes in advance but requires users to click manually to maintain “fairness in hand speed.” The high-frequency and high-speed operations of AI assistants (such as ticket grabbing and game competitions) may break this balance: some users gain an “operational advantage” far beyond that of ordinary users through AI, leading to unfair resource allocation (such as popular concert tickets being grabbed by AI, leaving no tickets for ordinary users) or imbalance in game competitions (AI players crushing manual players). This unfairness brought about by “technological empowerment” may trigger negative emotions among users towards AI assistants (such as accusations of “technological bullying”) and even lead to regulatory intervention. For example, if AI ticket-grabbing tools are abused, it may trigger restrictions on “technological ticket grabbing” by transportation departments, which in turn restricts the normal needs of ordinary users.
Thirdly, the potential risks to user privacy and security intensify the trust crisis. Although Doubao emphasizes that “users need to give clear authorization, and sensitive operations require secondary confirmation,” users’ concerns about AI “overstepping its bounds” still exist. The question in the news, “Does the mobile phone still need an owner? Is it the AI assistant or is it monitoring the user?” reflects the public’s deep-seated anxiety about AI proxy operations: AI needs to obtain data such as screen content and operation records to complete tasks. Is the collection, storage, and use of these data safe? If the AI system is hacked, will it lead to the leakage of user privacy or financial losses? The current Internet risk control system has not adapted to the “AI proxy” scenario and can only avoid risks by “banning services across the board” (such as financial apps suspending support). This “technological black box” further intensifies users’ distrust. For example, users may abandon using relevant functions due to concerns about AI misoperations leading to payment errors, which limits the release of the technological value.
Suggestions for Entrepreneurs: Find the “Breakthrough Point” between Technological Innovation and Ecological Collaboration
The controversy over the Doubao Mobile Assistant provides important inspiration for entrepreneurs: the implementation of cutting-edge technologies such as AI Agents needs to balance technological advancement and ecological compatibility. Here are some specific suggestions:
- Clarify technological boundaries and respect existing ecological rules: Entrepreneurs need to realize that AI proxy operations are not “unrestricted freedom,” but need to explore value within the existing ecological rules. For example, in sensitive areas such as finance and gaming, they should actively limit the fully automated operation of AI and retain the user’s manual confirmation process; in advertising and traffic distribution scenarios, they should avoid designing functions that “skip all ads” and instead explore a “precision advertising recommendation” model in cooperation with platforms (such as AI recommending highly relevant ads based on user preferences to improve conversion rates rather than simply blocking ads).
- Promote collaborative innovation between “technology and rules” and participate in standard setting: The contradiction between technological advancement and rule lag needs to be resolved through collaboration. Entrepreneurs can actively communicate with Internet platforms and regulatory agencies and participate in formulating interaction standards for AI Agents (such as digital signature authorization and operation whitelist systems). For example, they can cooperate with financial apps to develop an “AI operation authentication interface” to confirm the legitimacy of AI proxies through encryption protocols; they can jointly establish a “fair operation agreement” with game manufacturers to clarify the scenarios in which AI can participate (such as automatic tasks in non-competitive games) and prohibited scenarios (such as PVP competitions).
- Strengthen user trust building and design a transparent authorization mechanism: To address users’ concerns about privacy and security, entrepreneurs need to enhance transparency through technological means and product design. For example, they can provide a “operation log review” function so that users can check the AI’s historical operation records at any time; they can adopt the “principle of least privilege” and only obtain the necessary data to complete tasks; they can confirm sensitive operations through pop-ups, voice, and other multi-modal methods to ensure that users “know and control” the AI’s actions.
- Focus on “incremental value” and avoid direct competition with existing models: The core advantage of AI Agents is “efficiency improvement,” not “replacing users’ decision-making.” Entrepreneurs should focus on mechanical labor that “users don’t want to do or it’s not worth doing” (such as form filling and message classification), rather than creative activities that “users want to do but are not good at” (such as content creation and decision-making). For example, they can develop a function for “automatically organizing meeting minutes and reminding key information” instead of replacing users in meetings; they can design a tool for “automatically comparing business travel itineraries and recommending the best plan” instead of directly placing orders for users (retaining the user confirmation process).
- Pay attention to industry trends and layout the “capability interface” collaboration model: In the future, the relationship between AI Agents and apps will shift from “opposition” to “collaboration.” Entrepreneurs need to layout the “capability interface” ecosystem in advance. For example, they can develop AI components that support multi-platform calls based on the MCP protocol, allowing users to call different app functions through unified instructions; they can cooperate with app developers to design “AI-friendly” functional modules (such as simplified API interfaces and standardized operation instructions) to lower the technical threshold for AI proxy operations.
Conclusion: The controversy over the Doubao Mobile Assistant is a microcosm of the human-machine interaction revolution in the AI era. The “flood” of technological innovation will surely come, but it is not a “beast” – as long as entrepreneurs find a balance between efficiency improvement and ecological collaboration, and between user needs and rule boundaries, AI Agents will surely become a key force in promoting the upgrading of social efficiency.
