The rapid evolution of AI search paradigms has taken the digital landscape by storm, marking a significant shift in how users interact with information onlineCompanies like Moonlight, Zhizhu, Perplexity, and OpenAI have unveiled enhanced AI search reasoning capabilities, fundamentally altering the user experience.

Unlike the past, where users would struggle with mere keyword searches akin to “finding a needle in a haystack,” today’s advancements allow AI to dissect questions into manageable steps, conduct tailored searches, and present answers in a format that feels more intuitiveThis development resonates with the principles of knowledge management, whereby raw data lacks clarity until it evolves into information and then into knowledge, and ultimately into wisdomThe integration of techniques like Chain of Thought (CoT) signifies that AI searches have entered a transformative 2.0 era, moving from simple information gathering to summarizing patterns and aiding decision-making.

The increasing dependence on AI search solutions has exceeded preliminary expectationsFor instance, Perplexity claims to handle a staggering 100 million queries weekly, accumulating to approximately 400 million monthlyThis competition for users, particularly against tech giants like Google, poses compelling implications for the future of online searchNotably, recent reports suggest a historic dip in Google’s search ad market share, dipping below 50% for the first time in a decade, underscoring the shift towards AI-driven alternatives.

Various players in the digital space are keenly eyeing the opportunities left by traditional giants like Google, envisioning two primary pathways for advancementThe first path involves creating general-purpose AI search tools targeting consumer (C-end) markets by selling usage rights and promoting ads to businesses (B-end). The second path focuses on embedding AI search functionalities into applications, effectively bypassing the traditional search engine middlemen by developing proprietary distribution channels

Advertisements

This method enhances user experiences while promoting a more profitable business model that limits intermediary costs.

Among these players, Moonlight has leveraged a strategy of creating urgency and excitement around its status as the first AI search engine by aggressively marketing its capabilitiesMinimax, having capitalized on its initial successes, is shifting its focus toward AI application recommendationsEven WeChat, historically slow to adopt new technologies, is now experimenting with an AI Q&A feature that links its extensive ecosystem with new functionalities.

A common sentiment across the industry reflects a growing consensus that as more players enter the fray, competition will intensifyAs experts indicate, “The more consensus, the higher the barrier to entry,” suggesting that without distinctiveness, many offerings may fall short against established platforms like Google.

The demand for deeper reasoning in AI search systems has redefined search as merely a querying actionUnderneath this evolution lies a complex logic chain, seamlessly combining problem identification and resolutionIn practical terms, when users pose intricate questions that may involve extensive data retrieval and multiple intentions, modern AI systems refrain from immediate answersInstead, they first dissect the question, methodically breaking down the task into steps, culminating in a thorough “search + analysis” process before delivering a solution.

This framework of deep reasoning epitomizes an evolution from a simplistic AI that mechanically retrieves information based on keywords to an adaptable AI capable of discerning patterns and constructing thoughtful conclusionsFor example, when tasked to summarize key points from all of Jeff Bezos’s annual shareholder letters, an AI would first locate and extract the relevant documents, followed by synthesizing summarized findings into a comprehensive tableFollowing this, advanced tools like Kimi would perform a reflection stage, ensuring a fully rounded answer.

The shift to “search + deep reasoning” transforms AI search into an actionable agent by yielding operational steps rather than merely presenting keywords

Advertisements

It represents a logical pivot from merely “finding” to “solving,” where the act of searching becomes merely one component of a broader problem-solving toolkitThis perspective not only broadens the search spectrum but also enhances the overall quality of AI searches by allowing them to tap into various document formats, knowledge repositories, and software applications.

However, it is crucial to acknowledge that AI search agents currently face inherent limitationsFor example, Light Photon Planet has identified that an overemphasis on the completeness of search protocols can yield correct processes but misleading resultsThis highlights a disconnect in the correlation between components within the CoT, potentially leading to cycles of repeat search and analysis that waste computational resources.

Moreover, in situations where the reasoning process elevates a straightforward problem, transforming it into overly complicated tasks, the efficacy of search can diminishThus, the quality of constructed reasoning models must take precedence over sheer quantity.

As AI search embarks on this 2.0 stage, a noticeable shift is evident: the process now amalgamates selections of refined content filtered through various dimensionsIn contemporary search scenarios, traditional keyword-based searches require active engagement from users to refine information through clicks on multiple pagesAI-driven searches not only automate the retrieval but ensure that the outputs align closely with user expectations.

Nonetheless, within this paradigm shift, user perception diverges significantlyTools like ChatGPT can respond primarily to basic “who, what, when, where” questionsThe primary advantage of this consolidated approach is its capacity to amalgamate essential knowledge without requiring navigation through countless webpages, leveraging prompts that can be mixed and matched to yield coherent answers.

Crucially, the transformative journey from data to wisdom unfolds within parameters of knowledge depth—where qualitative findings arise post extensive search and comparative analyses

Advertisements

Wisdom, being the next layer, entails actionable strategies derived from understanding patterns obtained from knowledgeThe AI search systems of this evolving 2.0 stage straddle the gap between knowledge and wisdom, allowing users to further inquire about the implications of identified trends or even upload contextual documents for tailored recommendations.

Despite the promise of AI-driven search capabilities, current implementations are not devoid of room for improvementThe difference between open-book and independent reflective searches remains minimal, as keyword-based queries can often yield comparably detailed responsesTransitioning from mere information aggregation to practical wisdom requires significant investment in training and reasoning protocols.

AI search systems can be enhanced through diverse approachesFirst is the expansion of search volumes—the Kimi exploratory version boasts tenfold search capabilities, handling over 500 pages in a single queryHowever, this remarkable breadth leads to a critical challenge highlighted by Kimi’s innovation lead: “If Kimi cannot find the information, users are unlikely to locate it elsewhere.” This statement not only exudes confidence in Kimi's search volume but also indicates an inherent risk in conflating volume with quality.

Hence, the second enhancement approach emerges: prioritizing quality in search outputsThis can be operationalized by pre-filtering substandard or irrelevant information, ensuring that only quality data fuels the search powerConstructing a search funnel separating basic information from tailored insights reflects this intention, employing filtering methods to enhance precision and relevance.

The final enhancement involves empowering users by allowing them to directly contribute data and documents to search frameworksPerplexity, for example, has opened avenues for users to build their AI research and collaboration centers, fortifying personal knowledge bases that, when combined with online searches, yield more meaningful results.

Looking ahead, the trend foregrounding traditional search's regression and AI search's ascension is unmistakable

AI has swiftly become a competitive terrain, and within this realm, recruitment trends reveal a strong emphasis on AI search reasoning among key players like MoonlightThis urgency is reflective of both the dynamic digital environment and the necessity for enterprises to adapt swiftly to harness emerging opportunities.

Many uncertainties loom regarding the duration of the new players’ window of opportunityThe ongoing accumulation of digital user bases is becoming an asset in the race to monetize search capabilitiesReports suggest that Perplexity is set to introduce ad placements within its app, including media sponsorship for “question answering” and featured video placementsIn this context, Kimi has revamped its user interface to incorporate advertising slots to drive revenue streams.

Who the users are will ultimately define how effectively new platforms can sell this accessed informationData indicates a significant portion of Perplexity's users are professionals with high-income backgroundsThis positions Perplexity as a formidable competitor in capturing premium ad spaces across specialized sectorsConversely, platforms targeting a more generalized audience, such as Kimi, encounter challenges due to their broader applications and reduced capitation from subsequent advertising opportunities.

Additionally, Perplexity’s collaborative model, akin to Notion, showcases the potential for AI-powered search capabilities to become versatile and tailored for small businesses seeking efficient solutionsThrough an open API, connections can be fostered with numerous products, amplifying user engagement while mitigating capability constraints.

As the revenue-driven phase gains momentum within the AI sector, search capabilities have been heralded as the most direct pathway to profitabilityIf AI search systems can become the pioneering force in large-scale model commercialization, the implications will resonate widelyThe pressing question remains: who will be the first to seize this emerging opportunity and “take the first bite of the cake”?

Advertisements

Advertisements