1
Claude Code, an impressive and possibly game-changing “coding agent” for programmers to write code faster is the single biggest advance in AI since the LLM. And the thing is, Claude Code is NOT a pure LLM. And it’s not pure deep learning.
Home / LinkedIn creators / Gary Marcus
Gary Marcus
Scientist, author (6 books, including Rebooting AI (Forbes 7 Must Read Books About AI), and Founder (Geometric Intelligence. Acquired by Uber, and Robust.AI). Professor Emeritus, NYU.
We read 209 of their posts. Here is what they actually do.
linkedin.com/in/gary-marcus-b6384b4 ↗37KFollowers
2.4Posts / week
159Avg likes
95Median likes
29Avg comments
1.8K likesBest post
Engagement over time
Average likes per post, by month.
2025-07Average likes per month2026-07
Their hooks, word for word
The opening line of their best-performing posts. This is what stops the scroll.
- Claude Code, an impressive and possibly game-changing “coding agent” for programmers to write code faster is the single biggest advance in AI since the LLM. ↗1.8K likes
- Why is the AI backlash growing? ↗1.4K likes
- Five things most people don't seem to understand about DeepSeek. ↗953 likes
- 👉 DeepSeek may well have broken OpenAI’s Terms of Service and distilled their intellectual property without permission ↗912 likes
- Huge victory for neurosymbolic AI and an embarrassment for pure LLMs. ↗755 likes
- Why things will eventually fall apart: ↗625 likes
Content pillars
What they post about, with a line from a real post as evidence.
- Neurosymbolic AI Vindication
Highlighted Claude Code leak proving classical symbolic code is required for agent reliability.
- AI Market & Tech Scepticism
Analyzed why DeepSeek is an economic disruption rather than an AGI breakthrough.
- Societal Downsides and Governance
Detailing the growing backlash against GenAI harms and regulatory failures.
Signature moves and voice
The repeatable moves that show up across their best posts.
- Open with a provocative contrarian verdict on breaking industry news.
- Frame market shifts using numbered lists of economic and technical consequences.
- Reference long-held personal theses being proven right by new technical evidence.
Tone of voiceUncompromising, contrarian, academic, and sharp with an urgent, authoritative tone calling out industry hype.
What works for them
Patterns that separate their top posts from the rest.
- Technical teardowns claiming validation of long-held contrarian theories drive massive peak engagement.
- Numbered lists dissecting market overhype and missing economic moats reliably spark discussions.
- Direct, blunt challenges targeting industry leaders outperform vague philosophical commentary.
When they post
Every post, grouped by day of week.
Mon
137
Tue
203
Wed
122
Thu
157
Fri
129
Sat
191
Sun
180
Posts publishedAverage likes
Best day is Tuesday at 203 average likes. Wednesday is their weakest at 122.
What format works
Text, image, video or carousel, ranked by what actually lands for them.
| Format | Posts | Avg likes | |
|---|---|---|---|
| Image | 16 | 220 | |
| Text only | 193 | 154 |
Steal this
Moves you can apply to your own posts this week.
- Break down breaking tech news using foundational principles rather than following common hype.
- Structure contrarian takes into structured bullet points analyzing economic reality versus expectations.
- Directly name industry giants and challenge their core assumptions with deterministic counterarguments.
Best-performing posts
Ranked by likes, across everything we analysed.
2
Why is the AI backlash growing? Outside of coding (where there is clear value), and a handful of other domains (e.g. brainstorming), Generative AI has been a net negative for society.
3
Five things most people don't seem to understand about DeepSeek. • It's not smarter than earlier models, just trained more cheaply • It doesn't solve hallucinations or problems with reliability.
4
👉 DeepSeek may well have broken OpenAI’s Terms of Service and distilled their intellectual property without permission 👉 OpenAI may well have done analogous things to YouTube, New York Times, and countless artists and writers. 👉 Karma is a bitch.
5
Huge victory for neurosymbolic AI and an embarrassment for pure LLMs. Claude Code only works because it uses an old fashioned symbolic parser. update, pending, per Victor Bajanov’s comments/thread below the main thing seems to be grep (still symbolic) not the parser.
Similar creators
Comparable follower counts, analysed the same way.
Stop paying $3,000 for what 2pr does in 1 minute.
free trial, no limits on requests — start growing on LinkedIn today.
Built from public LinkedIn posts, last refreshed 2026-11-07. Follower counts and engagement change over time. Gary Marcus is not affiliated with 2pr.io.