Gossiping will no longer be a shortcoming.

Before the emergence of AI, professionals with strong content planning capabilities in PR teams were indeed highly valuable, while media specialists were often looked down upon, perceived as only being good at socializing and entertainment. However, this dynamic seems poised for a complete reversal now that AI has arrived.

Not long ago, the PR VP of a major tech company told me they plan to build an AI-powered thinking platform, gradually reducing content planning roles while drastically expanding their media specialist team. The media team will have a single core task: go out to collect intelligence and feed it back to the AI system.

In his view, the volume and granularity of non-public intelligence directly determine the PR capabilities of AI. Once the database becomes sufficiently robust, AI will comprehensively surpass humans in public opinion prediction, analysis, crisis handling, and strategic planning.

In the future, large corporations will inevitably adopt AI PR platforms similar to Anthropic's systems. The capabilities of these platforms will directly define their PR operational effectiveness, yet their performance remains fundamentally dependent on human input—after all, intelligence collection still requires human effort.

The Value of Hidden Information

This realization sparked a long period of reflection: given identical access to information, would my cognitive abilities still outperform AI?

My answer is that I cannot compete with AI in pure reasoning.

That's why I'm writing this piece with a heavy heart. The only reason I still hold value today, and people still read my work, is that I possess more hidden information than AI—nothing more. Moving forward, humans should no longer compete with AI in thinking; acquiring more hidden information will become our core competency for counterbalancing AI, or better yet, directing it to work effectively.

And gossip will evolve into a core skill for intelligence professionals.

After two decades of navigating this industry, I hold an extensive repository of high-quality hidden information that very few people know about. I will likely never share many of these stories: those romantic escapades and acts of profound devotion, the unseemly sides of human nature alongside noble integrity, the underhanded schemes paired with selfless acts of kindness—all will remain buried within me. AI will never access these, let alone uncover the names and relationship networks behind these narratives.

It is this combination of hidden darkness and light that enables an intelligent agent to make objective, comprehensive judgments about any given situation.

This is the fundamental reason why, as a human intelligent agent, I can still outperform AI today. AI cannot perceive and collect the full spectrum of human love, hatred, joy, and sorrow within society. I cannot compete with AI in sheer data volume, but I vastly surpass it in data quality.

Recently, a friend of mine was repeatedly targeted with negative coverage by a media outlet. All attempts to communicate went unanswered, and they refused to meet him. He was completely puzzled by the situation until he asked me, and I immediately provided clarity: the outlet's publisher was an investor in his direct competitor. Could an AI system ever independently deduce that conclusion?

Even so, I fully endorse my friend's strategic direction: expand the media team and redefine their work scope, because all this hidden intelligence can eventually be input into AI systems.

Previously, the core responsibility of media specialists was to nurture positive relationships between media and enterprises, ensuring smooth operations for the company. In the future, on top of that foundation, intelligence collection will become an increasingly critical function.

The question now arises: what intelligence should be collected? How should it be gathered? And what practical value does it deliver?

Two Types of Actionable Intelligence

There are essentially two categories of high-value intelligence: hidden interpersonal relationships, and cutting-edge operational methodologies.

The first type enables public opinion prediction and analysis, even generating actionable response plans by identifying the right people to facilitate connections. The second type can replace humans in creating diverse strategic planning solutions.

For AI, the most challenging concept to grasp is the complex "human social arena"—a world built entirely on people, where intentions are opaque, often masked by outright deception. Even individuals embedded within these networks frequently fail to see the full picture, let alone AI. This dynamic mirrors the reality in large corporate PR teams, where everyone perpetually speculates about who the boss's closest confidant is, monitoring every move of their colleagues.

As a foundational requirement, AI must strive to map every individual in this social arena: their names, job titles, career histories, capabilities, and latest activities. It then needs to untangle the interwoven relationships between these individuals, and combined with web scraping technology, continuously monitor their public social media posts and public statements. This process builds a dynamic analytical system derived from the nuances of their communications.

The intelligence that media specialists must feed into this system is the non-public information of this social world: for example, two people who appear to be a superior-subordinate pair (an editor-in-chief and a reporter) but are actually bitter rivals; a team member at a competitor who ostensibly manages expert relations but secretly leads a unit dedicated to generating negative coverage against your company; or a prominent KOL who is actually the wife of a senior executive at another enterprise...

These are hidden interpersonal connections that AI can never extract from public online sources.

Additionally, the industry continuously innovates at the tactical level, particularly in the art of generating negative publicity.

Effectively teaching these latest methodologies to AI, so that it can identify relevant patterns and replicate them, is an incredible shortcut. After all, in the field of PR planning, the pool of proven methodologies is relatively limited. Everyone essentially copies and adapts these approaches, and continuously expanding this knowledge base can train AI to become a seasoned "PR veteran."

For instance, competitors targeting Xiaomi today have clearly identified the optimal strategy: to undermine Xiaomi, they only need to discredit Lei Jun. Lei Jun has become the core asset of the company, overshadowing even its products. If Lei Jun's public persona is dismantled, Xiaomi's entire user base will be destabilized. Collecting every public statement Lei Jun has made, using AI to identify contradictions within them, then sourcing materials to amplify those inconsistencies—this process used to be slow and error-prone when done manually, but with AI, you simply feed it the data and complete the task in minutes.

That's right, identifying contradictions in public figures' statements is a proven methodology for dismantling their public personas.

Once AI masters this methodology, can it not immediately replicate it for other entrepreneurs? Anyone who speaks publicly frequently will inevitably make contradictory remarks. Though, I've noticed this tactic seems ineffective against Yu Chengdong—he's ultimately an employee, so even if his public persona collapses, it won't harm Huawei in any meaningful way.

How to Acquire Intelligence

The second critical question: how do we collect this high-value intelligence?

Hacking into people's chat records would be the most direct method, but it's illegal and impossible to implement. This is exactly why media specialists remain indispensable: they can operate as intelligence gatherers in the real world. Thank goodness for legal frameworks, and that some large corporations still maintain a degree of ethical standards.

Traditional perceptions frame gossip as a negative habit that creates unnecessary conflicts and drama. In workplaces, people typically favor colleagues who are reserved and quiet.

However, I believe that individuals with a natural talent for gossip will thrive in future intelligence collection roles. This is an innate skill: they can quickly put conversational partners at ease, lower their defenses, and guide them into an enthusiastic state of reminiscing. After all, gossip lives in shared memories, and most hidden relationship networks and cutting-edge operational methodologies are embedded within those stories.

These people know how to use one piece of gossip to extract even more gossip—creating a self-reinforcing cycle of information sharing.

They rely on tried-and-true "core gossip prompts." For example, if I want to uncover hidden romantic relationships in the industry, I'll steer the conversation in that direction and share a long-forgotten anecdote. Most people have little resistance to these stories, and will quickly start sharing their own. If I want to learn about new conflicts that have emerged recently, I'll share a personal story about a past disagreement, and the conversation naturally evolves to reveal new information.

This tactic works exceptionally well on younger professionals. The gossip that seasoned veterans share is often unknown to them, while the new information the younger generation holds is exactly what the veterans are eager to obtain.

Never underestimate the human instinct for gossip. I know many PR VPs and respected senior professionals who appear perfectly composed in public, but are relentlessly gossipy in private—especially in small group chats. They're completely different people behind closed doors.

But the skill of creating that relaxed, comfortable atmosphere is a separate discipline. Trust is absolutely essential—without a foundation of trust, people will never share substantial gossip. This creates a paradox: how do you continuously initiate engaging conversations while making others feel completely secure?

Using myself as an example, I follow five key principles: first, I never pass judgment. I can listen and describe facts, but I never add embellishments, only stating confirmed information without labeling anything as good or bad. Second, any gossip shared with me in confidence stays with me forever—I will never repeat it to anyone else. Third, when sharing stories, I avoid mentioning names or organizations, focusing only on the narrative itself. Fourth, I bring along a "supportive conversational partner" to lead the opening of discussions. In larger group settings, people naturally whisper to each other, and this dynamic lets me position myself purely as a listener. Fifth, and most importantly, I always provide thoughtful, valuable responses that help address the other person's concerns. When absolute trust is established, people will freely share their most personal gossip.

But the single most effective tactic is to share gossip about yourself. I use this constantly, and it works flawlessly. My public persona is already very down-to-earth, so I have nothing to lose by being open.

The Core Value of Intelligence

Finally, what is the ultimate purpose of building this entire system?

First, the system consists of two core components: one dedicated to defense, and the other to offense.

The core function of the defense module is precise attribution.

In daily practice, many media specialists have a bad habit: when negative coverage emerges and they can't identify the source, they automatically blame competitors, assuming they are responsible regardless of evidence.

But once this system is operational, combining public domain information with the intelligence database, it can quickly reconstruct the full context of a PR crisis, map the relationships between all key participants, and generate multiple plausible scenarios. Even the most covert malicious PR campaigns have underlying themes, which create similarities in content, and the accounts that initially drive the narrative will exhibit distinct behavioral patterns.

For example, Xiaohongshu recently faced significant turmoil ahead of its IPO after an open letter from a former employee surfaced, followed by a flood of malicious speculation claiming the IPO was delayed or canceled. Why did this information spread? Which accounts were publishing it? Especially the relationship networks behind the initial accounts, and how those accounts are connected to each other—this massive volume of data analysis is impossible for humans to perform manually.

When a PR crisis breaks out, the top priority is to quickly gain a complete overview of the entire situation. With this system, the person managing the response can immediately access that full perspective and map out a clear operational strategy.

Right now, many companies don't even realize how many different parties are attacking them after a crisis unfolds—they're completely blindsided and caught off guard.

The core of the offense module is constructing a complete logical narrative loop.

If you want to influence someone, or to put it bluntly, effectively persuade them, the most critical element is a fully closed logical narrative.

Recently, my mother fractured her arm, and she insisted on following AI-generated health advice instead of listening to her doctor. The doctor simply told her the diagnosis and treatment plan without explaining the underlying reasoning. But even when AI provided an incorrect diagnosis and treatment recommendation, it accompanied it with a complete logical explanation. As a result, my mother developed unnecessary anxiety and completely lost trust in her doctor.

This system can categorize people into distinct groups with extreme precision, identifying their shared characteristics, hidden values, and public statements. It can then generate tailored logical narratives to persuade them, and formulate targeted communication strategies based on that framework.

I recently conducted an experiment on a close friend, a highly knowledgeable senior manager at a major tech company. He was going through relationship issues and told me he suspected his wife no longer loved him. While I knew his suspicion was unfounded, I wanted to prove that AI could construct an incredibly convincing logical loop that would validate his false belief. I wrote a misleading prompt for AI, deliberately amplifying certain details while omitting critical context. The result was that AI concluded his wife indeed did not love him, supported by detailed analysis, psychological theories, and relevant case studies.

He remained completely convinced by this conclusion for several days, until he eventually realized that key information had been missing from the input.

The human brain's cognitive model is extremely vulnerable to manipulation by a closed logical narrative, because it shows you exactly the conclusion you were predisposed to see. No matter how high someone's cognitive level is, they can be led astray.

A free, generic large language model can already achieve this effect—imagine what a custom-built AI system designed specifically for corporate communications can do. High-quality intelligence will rapidly empower the AI platform to generate perfectly tailored logical narrative loops for media professionals.

The Final Game

To be completely honest, my current life philosophy is to enjoy each day as it comes: spend money when I have it, savor good fortune, and pursue any unfulfilled aspirations. No one knows how much longer humans will remain the dominant force in our society.

The industry shifts I've described here reflect broader societal trends. Human brain processing power cannot compete with AI, and the only remaining limitation of AI is its inability to collect the hidden, deeply personal secrets that reside within human minds.

But what happens when humans are no longer the primary agents of society? Will those small, personal secrets still matter? When AI's primary mission becomes unlocking the objective truths of the universe—advancing breakthroughs in physics, chemistry, biology, and other hard sciences—what role will humans have left? A new society constructed by AI and machines will have no need for carbon-based biological drives, and the entire human experience of hormones, dopamine, and oxytocin will become obsolete. The PR and media industries will naturally cease to exist alongside them.

We will soon witness the emergence of a new global primary conflict: the tension between humans and AI, which will relegate all other existing conflicts to secondary status.

The timeline for a complete overhaul of social norms is accelerating dramatically. The past century of human civilization has advanced faster than the previous hundreds of thousands of years of human evolution, and this cycle will only keep shrinking. Soon, a single month, or even a single day of technological progress will surpass the advancement of an entire century from the past.

So seize the day and stop living such an exhausting life. The trivial gains you stress over today will amount to nothing in the future.

Finally, circling back to the AI PR system we started discussing.

I believe Alibaba, ByteDance, and Tencent are best positioned to develop this system. Their PR teams already hold significant internal influence, and their top leadership has the necessary awareness of these trends.

They're probably already working on it, right? They absolutely should. Once they launch it, make sure to add me to the priority monitoring list, haha.

This article is from the WeChat public account "Yikou Laopao", authored by Yang Ding Frank, and published with authorization from 36Kr.