Thought leadership three years ago was a content marketing function. Executives published opinion pieces and book chapters, gave conference talks, and built audiences on LinkedIn through consistent posts. The work scaled linearly. More posts, more speeches, more views. The framework was relatively stable.
That framework broke in 2023 and is still breaking. AI tools made it trivial to generate competent-sounding content at unlimited scale, which collapsed the value of competent content. AI search products started routing audiences to specific cited sources, which changed which thought leaders show up in research. Audiences got better at recognizing AI-generated content and turned skeptical of anything that sounds like it. The new environment rewards different work than the old one did.
This piece is for the executive, founder, consultant, or operator trying to build authority in the new environment. Specific moves that work, specific traps to avoid, and the long view on what thought leadership becomes when machine-generated content is free.
What changed
The flood of AI-generated content is the most visible shift but not the most important one. The deeper shift is in how audiences discover and evaluate sources.
Audiences used to discover thought leaders through search, social, and word of mouth. They read a few pieces, formed an impression, and decided whether to follow or hire the person. The discovery loop took weeks or months and rewarded consistent publishing because each new piece reinforced the impression.
Audiences now compress that discovery into a single AI query. “Tell me about Sarah Chen, the operator who writes about pricing.” The AI returns a synthesized summary in 30 seconds. The summary is built from whatever the AI can find: the LinkedIn bio, the recent posts, the podcast appearances, the news mentions, the Wikipedia entry if one exists. The audience’s first impression of the thought leader is not formed from reading her work directly. It is formed from the AI’s synthesis of what other people have said about her.
This change is profound. Thought leadership used to be evaluated by what you wrote. It is now increasingly evaluated by what the AI says you wrote. The two are correlated but not identical, and the gap is what most thought leaders have not yet adjusted for.
The second change is the median quality drop. AI tools made it possible for executives, agencies, and ghostwriters to produce competent thought-leadership content at scale. LinkedIn, Medium, and corporate blogs filled with this content within 18 months. Audiences responded by tuning out. The piece that would have stood out in 2022 because it was thoughtful now blurs into the AI-generated noise because it sounds the same.
The third change is in voice. Audiences became sensitive to the AI tells (overly balanced phrasing, em dashes, “delve into,” “navigate the landscape,” “in today’s digital age”). Content that includes these tells gets discounted regardless of who actually wrote it. The defense against the discount is writing in a recognizable human voice with real specificity and willing edges, not the smooth corporate voice that AI tools default to.
What still works
The thought leaders who are breaking through in this environment share specific qualities.
They have actual experience that they reference in concrete detail. Not “in my career I have learned” but “when we tried to launch in Germany, the customer acquisition cost was triple our model assumption and we cut the program in six months.” The specific story is the unfakeable signal. AI cannot generate it, ghostwriters need to extract it, and audiences treat it as evidence of credibility.
They have positions that are specific enough to be wrong. Not “leadership matters” but “most CEOs underestimate how much they should be the brand themselves rather than building a separate marketing brand.” The specific claim creates engagement. Some readers agree, some disagree, some have stories that complicate the position. All of them remember the writer.
They write in a voice that is recognizably theirs. Some thought leaders are dry and analytical. Some are warm and conversational. Some are blunt to the point of rudeness. None of them sound like a corporate communications team. The voice is the recognition signal that survives across platforms and across time.
They publish consistently in a few places rather than spraying across every channel. The thought leader who publishes weekly in two places (LinkedIn and a newsletter, or LinkedIn and a podcast) compounds faster than the one who publishes occasionally everywhere. The audience develops a habit of looking for the work, which only happens when the cadence is predictable.
They engage rather than broadcast. Replying to comments, responding to other thought leaders’ pieces, participating in conversations rather than only initiating them. The engagement signal feeds the algorithm, builds relationships with peers, and keeps the work fresh because the writer is in conversation with their actual audience.
The AEO layer for thought leaders
The work to show up well in AI search results is real and most thought leaders have not done it. Specific moves that compound:
Build a clean About page on your website. The About page is what AI products read first when summarizing a person. It should include your name as the canonical phrase, a one-paragraph positioning statement that names what you actually do, your three to five most credible accomplishments with specific dates and outcomes, the companies you have worked at with brief descriptions of what they do, and your education. Write it in third person so the AI can extract it cleanly. Keep it under 600 words.
Maintain your LinkedIn profile carefully. LinkedIn data flows into many AI products’ synthesis. The headline should match how you describe yourself elsewhere. The about section should be substantive (250 to 400 words) and free of buzzwords that AI products discount. The experience section should be specific about each role and outcome.
Get a Wikipedia entry if you legitimately qualify for one. The Wikipedia notability bar is high, and most thought leaders do not meet it. If you do (significant book, major company, sustained press coverage in mainstream outlets), the Wikipedia entry becomes the canonical source AI products reference. Editing your own Wikipedia entry is forbidden, so the move is making sure the underlying coverage exists for an editor to pull from.
Set up a Wikidata entry. Wikidata has a much lower notability bar than Wikipedia and serves as a structured data layer for AI products. A clean Wikidata entry with your name, occupations, employers, and notable works makes you machine-readable in a way that propagates across many AI surfaces.
Maintain your Crunchbase, Forbes Profile, AngelList, and other directory entries. The data quality on these surfaces feeds AI synthesis. Inaccurate or outdated entries produce inaccurate AI summaries.
Get cited by sources the AI trusts. Earned media in mainstream outlets remains the highest-trust signal. Podcast appearances on shows the AI has seen referenced in trusted sources. Articles on Medium publications, Forbes, Inc., or trade publications in your category. The path is making sure your real work gets covered in places the AI has learned to weight.
What to write about
The topic selection that compounds in the new environment is more selective than it used to be.
Pick a small number of recurring themes rather than spraying across topics. The thought leader known for one or two specific obsessions builds recognition faster than the one who comments on whatever is trending. Pick three topics maximum and stay with them for at least 18 months. Audiences develop the mental shortcut “this person is the one to read about pricing strategy” only when they have seen you publish about pricing strategy ten times.
Take positions that competitors will not take. The unique value of thought leadership is the position itself. Content that summarizes consensus has zero value because consensus is freely available. The thought leadership that gets cited and shared is the position that has bite. The risk is being wrong publicly. The reward is being remembered.
Write longer pieces less often rather than short pieces constantly. The 1500 to 2500 word pieces that develop an argument compound far better than the 200 word LinkedIn posts. The longer pieces become the citation surface. They get pulled into AI summaries. They get linked from other thought leadership. They survive the algorithm changes.
Write pieces that someone could disagree with. The most-shared thought leadership in 2025 onward is the work that triggers disagreement, debate, and follow-up posts from peers. Safe pieces get a like and disappear. Pieces with edges create the conversation that sustains attention.
Avoid AI commentary specifically. The AI commentary lane is saturated to the point of meaningless. Unless you have specific operating experience inside an AI lab or an AI-first company, your AI commentary blurs into noise. Better to write about your actual domain and let AI references appear naturally when they fit.
The voice problem and the ghostwriter question
Most executives do not have time to write the volume of thought leadership their position calls for. The ghostwriter market expanded in response. The new environment changes how to work with ghostwriters.
The ghostwriting that works has the executive deeply involved in the source material. Long interviews with the executive, real positions documented from those interviews, drafts reviewed line by line. The ghostwriter’s job is structuring and writing, not generating ideas the executive then approves blindly. Pieces produced this way still sound like the executive because they originate from real material.
The ghostwriting that fails has the ghostwriter generating ideas, drafting pieces, and the executive approving with a quick read. The pieces sound generic because they are not anchored to specific experience. The audience picks up on this within months and discounts the executive’s brand as inauthentic.
Pure AI-generated content does not work for thought leadership in 2026. Audiences identify it within a paragraph and discount the source. Executives who use AI tools should use them for outlining, structural editing, fact-checking, and brainstorming, not for prose generation. The actual writing should pass through a human (the executive themselves or a real ghostwriter working from interview material).
Distribution that still works
LinkedIn remains the highest-impact distribution surface for B2B thought leadership. The audience is buying when they read, the algorithm rewards consistency, and the comments produce relationships that translate into business. The format that works is text posts with one or two specific points and a willingness to take a position. Carousels work for some audiences but text posts have the best signal-to-effort ratio.
Newsletters compound faster than they used to. Substack, beehiiv, and personal newsletters give thought leaders an owned channel that survives platform changes. The subscriber list is the most durable asset in the thought leadership stack. Building it should be a deliberate project alongside the LinkedIn work.
Podcasts work for the right thought leaders. Long-form audio transfers credibility unusually well. The thought leader appearing on three to five well-targeted podcasts per quarter often sees more business impact than the one publishing 20 LinkedIn posts. The quality of the host audience matters more than the size.
Speaking engagements still produce inbound, particularly at conferences with engaged audiences. The shift since 2022 is that speaking is less about reach and more about relationships. The 40 people who walked up after the talk are worth more than the 400 in the room.
The longer view
Thought leadership in the AI era will look different than it did in 2018, but the fundamentals remain. People want to learn from people who have done the thing. They want positions they can argue with. They want voices they recognize. They want depth that surprises them.
The flood of AI-generated content makes that work harder to find but more valuable when audiences encounter it. The thought leaders who commit to specificity, voice, and consistency over the next 24 months will compound the advantage as the noise gets worse. The ones who try to win on volume with AI-assisted output will lose ground because the audience can tell the difference and is increasingly unwilling to spend attention on content that does not earn it.
Pick the harder path. Write less and more carefully. Take positions. Show specific work. Show up consistently in a small number of places. The thought leadership that will matter in 2028 looks more like 2008 than 2023. The cycle has come back around.