The signal-to-noise problem on X is real. Here are the 15 voices actually worth your morning scroll.

AI moves at a pace that outstrips every communication channel built to capture it. By the time a Medium post is published, five new research papers have dropped. By the time a newsletter lands in your inbox, the discourse has moved on. X remains the closest thing we have to real-time AI discourse — but the problem is ancient and inescapable: separating signal from noise in a 500-million-post-per-day feed.

I've spent the last year actively curating my feed, testing which voices actually move my thinking forward and which ones just generate engagement theater. The result is this list of 15 people I check every morning — not out of obligation, but because I genuinely learn something from their perspective every few days.

The goal of this post is simple: help you build a smarter feed faster. Pick 3-5 of these voices, follow for a week, and notice how your understanding of the AI landscape shifts. That's the test.

The Signal Problem

Here's the reality: if you follow "AI" on X, 60% of what you see is:

  • Repackaged news from bigger accounts
  • Technical breakdowns of things already covered elsewhere
  • Hot takes designed to go viral, not clarify
  • Self-promotion disguised as insight

The remaining 40% is actual signal — new perspectives, early warnings, deeper understanding. The voices below represent that 40%.

The pattern is consistent: signal comes from people doing the work, not talking about people doing the work. Original research. Shipping products. Managing teams. Deploying at scale. That's where the real insights live.

Research & Foundations: The People Pushing the Frontier

1. Andrej Karpathy (@karpathy) — ~1.4M followers

Role: AI researcher, former Tesla Director of AI, early OpenAI team member

Why follow: Karpathy is one of the few people who will publicly admit that the pace of change has outstripped his own ability to keep up. That's not weakness — that's intellectual honesty. His recent post ("I've never felt this much behind as a programmer") resonated because it came from someone who has been at the forefront of deep learning for a decade.

What you'll learn: The shift from manual coding to orchestrating AI systems. Not just what's new, but what's fundamentally *different* about how software gets built now.

The insight: Follow Karpathy when you want to understand the **profession-level** implications of AI, not just the technology level.

2. Ilya Sutskever (@ilyasut) — ~500K followers

Role: Co-founder & Chief Scientist at OpenAI

Why follow: Ilya is a core voice in generative AI research from both the theory side and the shipping side. He touches on foundational architectures, scaling behaviors, and the kinds of insights that only come from running large training runs.

What you'll learn: Where the research frontier actually is. Not blog posts about AI — actual research directions being pursued by people with billion-dollar budgets and top talent.

3. Yann LeCun (@ylecun) — ~972K followers

Role: NYU professor & Meta's Chief AI Scientist

Why follow: Yann blends cutting-edge research with the willingness to challenge conventional narratives. He's one of the few senior figures who will publicly push back on hype (including hype around his own field).

What you'll learn: How to think critically about AI capabilities, limitations, and what actually matters for long-term progress vs. what's just hype.

4. Pedro Domingos (@pmddomingos) — ~108K followers

Role: ML researcher and University of Washington professor

Why follow: Pedro is both a theorist and a communicator. His work helps demystify machine learning fundamentals in a world obsessed with scale.

What you'll learn: The deeper ideas that often get lost amid the noise. Not everything is about the next $100M training run.

Product & Application: Shipping Real Tools

5. Arvind Srinivas (@arvindsrinivas) — ~320K followers

Role: CEO of Perplexity AI

Why follow: Arvind offers founder-level thinking on building AI products that users actually want. Not research; product lessons from someone shipping to millions.

What you'll learn: How AI moves from labs into applications people pay for. The gap between "technically possible" and "users value this" is vast.

6. Logan Kilpatrick (@logankilpatrick) — ~230K followers

Role: OpenAI developer relations lead and educator

Why follow: Logan bridges deep technical knowledge with community building. He shares practical guides, API insights, and real use cases.

What you'll learn: How to actually *use* the tools being released. Not just "what's new" but "here's how to build with it."

7. Linus Ekenstam (@linusekenstam) — ~220K followers

Role: Product designer and entrepreneur

Why follow: Linus brings a design and product lens to AI. He thinks deeply about how humans interact with models and what makes an AI product *feel* good to use.

What you'll learn: UX for AI isn't about interfaces — it's about understanding model outputs, uncertainty, and building products that respect user agency.

Venture & Startup Ecosystem: Early Signals

8. Bojan Tunguz (@bojantunguz) — ~255K followers

Role: Venture capitalist, entrepreneur, ex-physicist

Why follow: Bojan is exceptionally sharp at spotting emerging startups and structural shifts before they hit mainstream. His early takes on trends often come months ahead of consensus.

What you'll learn: Where VCs are actually placing bets and why. That's usually 6 months ahead of hype.

9. Varun Mayya (@varunmayya) — ~219K followers

Role: CEO of Avalon Labs, founder of JobSpire

Why follow: Varun offers founder-level commentary from someone building in the Indian startup ecosystem (a leading edge for AI applications in emerging markets). Real progress, not just positioning.

What you'll learn: Founder lessons backed by traction. How to think about building when capital is scarce but talent is abundant.

10. Rowan Cheung (@rowancheung) — ~567K followers

Role: Founder of The Rundown newsletter

Why follow: Rowan distills the week's highlights in AI — perfect if you want curated signal without drowning in noise. Think of this as a weekly compass setting.

What you'll learn: High-signal developments across all the above categories, filtered through a smart editor's lens.

Enterprise & Systems Thinking: How It Scales

11. Ronald van Loon (@ronaldvanloon) — ~342K followers

Role: AI, big data & enterprise trends commentator

Why follow: Ronald offers a high-level view of how AI intersects with enterprise infrastructure, cloud, IoT, and business strategy. If research is the frontier, enterprise is the lagging indicator of what *actually works*.

What you'll learn: What big companies are really doing with AI (not what they say they're doing in press releases).

12. Vin Vashishta (@vinvashishta) — ~29K followers

Role: ML strategist and engineer

Why follow: Vin talks about scaling ML systems, reliability, and how teams actually ship models at scale. His insights come from trenches, not theory.

What you'll learn: How AI works beyond research papers. The engineering practices that separate shipping teams from flailing ones.

13. Antonio Grasso (@antoniograsso) — ~350K followers

Role: Stanford professor and digital economy expert

Why follow: Antonio ties AI to economic and societal trends. He thinks about the systems-level implications, not just the tech.

What you'll learn: How AI drives economic shifts and market dynamics. The "why this matters beyond AI" perspective.

Critical Perspective & Ethics: The Necessary Counterweight

14. Gary Marcus (@garymarcus) — ~198K followers

Role: Entrepreneur and cognitive scientist

Why follow: Gary is renowned for pushing back on hype and raising hard questions about AI safety, cognition, and limitations. In a field obsessed with scaling, he's the skeptic we need.

What you'll learn: Where AI's actual limitations and risks live (not the dramatic sci-fi ones, the real ones that matter for deployment).

15. Fei-Fei Li (@drfeifei) — ~405K followers

Role: Stanford professor and former Google Cloud AI lead

Why follow: Fei-Fei blends technical depth with a strong emphasis on ethics and responsible AI development. She's building frameworks for thinking about AI *for people*, not just building AI.

What you'll learn: How to think about AI as a human-centered discipline. Technical excellence without ethical grounding is just optimization.

How to Use This List

Don't follow all 15. That defeats the purpose. Instead:

Step 1: Pick your category

  • If you care about research frontiers → Follow 2-3 from the Research section
  • If you're building products → Follow 2-3 from Product & Application
  • If you're evaluating trends → Follow 2-3 from Venture & Enterprise
  • If you need perspective → Follow 1-2 from Critical Perspective

Step 2: Follow for one week

Give the voices a week to shape your feed. Notice:

  • Which ones make you think differently?
  • Which ones do you disagree with productively?
  • Which ones surface ideas you hadn't considered?

Step 3: Iterate

Keep the ones that shift your thinking. Drop the ones that just echo. Your feed should evolve as your priorities shift.

The Meta Point

The AI landscape in 2026 is moving faster than any individual can track. The solution isn't to read more — it's to read better. Surround yourself with people doing the actual work: shipping products, running experiments, building teams, challenging assumptions.

Avoid the meta-commentators, the hype machines, the "AI will solve everything" crowd. Instead, follow the people who understand that AI is a tool, progress is nonlinear, and the interesting work happens in the details.

Your task is simple: Pick 3–5 of these voices, follow for one week, and notice how your feed changes.

The ones that stick with you after that week? Those are your signal.

Interested in translating AI insights into board-level outcomes? Read about why most AI strategies fail to produce ROI and how to fix the framing.