Top 10 AI influencers to follow on LinkedIn in 2026 (Q4 update)

The AI landscape shifts daily. For business leaders and technical professionals, the challenge is no longer finding information but filtering it. The difference between a distracted strategy and a competitive advantage often lies in the quality of your information diet. Down below we give the Top 10 AI Influencers to follow on LinkedIN in 2026, update for Q4 2026

LinkedIn sits between the speed of X and the density of arXiv, which makes it the best platform for seeing technical breakthroughs with a focus on how they impact businesses. That only works if you follow the right people, and if those people actually post there.

Everything is checked against public sources, including the detail most lists leave out: how often each person really posts on LinkedIn.

What defines a top 10 AI influencer?

We selected these profiles based on three objective criteria:

• Authority: They have built significant systems, led major research labs, or advised Fortune 500 companies.
• Clarity: They translate complex technical concepts into strategic business value.
• Consistency: They post regular, high-signal content without relying on hype or sensationalism.

Menno Fokkema leads DataNorth AI, a consultancy focused on making artificial intelligence workable for organisations across the globe. Where many commentators cover what AI might do, his content deals with what it currently does inside real companies.

He writes frequently about “pilot purgatory”, the pattern where AI projects stall after a successful proof of concept and never reach production. His posts break artificial intelligence strategy into decisions an organisation can actually make, and cover the practical side of AI compliance and the EU AI Act.

In March 2026 he launched a free public tool scoring how exposed a given job is to AI displacement, which drew national coverage in the Netherlands. He is the most European voice on this list, which matters if your obligations sit under EU law rather than US practice.

Andrew Ng is among the most influential educators in the field. As co-founder of Coursera and former head of Google Brain and Baidu AI Group, he simplifies machine learning without dumbing it down.

His central theme remains agentic workflows: the argument that AI agents which iteratively review and improve their own output beat single-shot prompting. His posts typically pair a diagram with a short explanation, which makes them unusually useful for technical leads and product managers.

Since January 2026 he has added AI Aspire, an enterprise AI advisory firm, to an already long list of roles. At Davos in January he argued that fears of AI-driven job losses are overstated, and in February he put the bubble risk in the model training and infrastructure layer rather than the application layer. He is a useful counterweight to the loudest voices in either direction.

Allie K. Miller sits between technical engineering and executive decision-making. Having advised hundreds of startups and large enterprises, she concentrates on the people and process side of adoption, which is where most implementations actually fail.

Her content suits organisations working out whether AI is feasible for a given process. She posts checklists for vetting AI vendors, frameworks for calculating return, and practical detail on upskilling teams, in terms of efficiency, revenue and competitive position rather than jargon.

Through 2026 her focus has moved firmly to agentic AI in the enterprise: how to assemble an AI agent workforce and what it costs. She is the most concrete voice here on what daily AI-augmented work actually looks like.

Ethan Mollick approaches AI as a management researcher rather than a technologist. He is the leading authority on how generative AI changes work right now. He does not speculate about AGI timelines. He runs controlled experiments on whether AI improves writing quality, coding speed or decision quality, and publishes what he finds.

His Prompting Science Reports, produced with Wharton Generative AI Labs during 2026, are worth reading before you invest in prompt engineering training. They found that returns from prompting tricks are inconsistent and task-dependent, that chain-of-thought gains have shrunk on reasoning models, and that expert personas do not reliably improve accuracy. His summary: specifications, not tricks.

His 2024 book Co-Intelligence was a New York Times bestseller. He has said openly that it was written for a world of chatbots, and that autonomous agents changed the picture enough to require a follow-up. That book, Co-Existence, publishes on 20 October 2026. If you are considering an AI workforce readiness scan, his data gives you the baseline.

Cassie Kozyrkov founded the discipline of decision intelligence at Google, where she was Chief Decision Scientist until 2023. She now runs Kozyr, her own advisory firm. Her position is that AI is a tool for making decisions at scale, so the quality of your decisions sets the ceiling on what AI can do for you.

She is essential reading for anyone working on AI for executives. She pushes readers to define what “good” looks like before anyone writes code, and regularly dismantles common statistical fallacies in data work, which makes her a strong preparatory read before an AI roadmap session. Her recent work centres on AI accountability: how systems get scored, how training data provenance is disclosed, and what benchmarks are actually worth.

Bernard Marr is a prolific author and Forbes columnist who covers breadth rather than depth. While others go deep into architecture, Marr tracks macro trends across manufacturing, retail, healthcare and financial services.

His use case round-ups for tools such as ChatGPT and Microsoft Copilot are practical starting points for non-technical stakeholders. His CES 2026 analysis was a good example of the format: what was announced, what was conspicuously absent, and what both signal.

He is the right follow if you need to brief a board on where an industry is heading, and the wrong one if you need implementation detail.

A Turing Award winner and one of the founders of modern deep learning, Yann LeCun left Meta in November 2025 after more than a decade as its Chief AI Scientist. He is now Executive Chairman of AMI Labs, the Paris-based world model company he co-founded, which raised a $1.03 billion seed round in March 2026.

His thesis is that models trained to predict the next token will not reach reliable machine intelligence, and that systems need internal world models learned from observation, with memory and planning. AMI Labs is a direct bet on that position, structured as a research organisation rather than a product company.

He is also a consistent advocate for open weights, which makes him useful reading for any organisation deciding between proprietary models and open alternatives, including local LLM deployments where data cannot leave the building. He is worth following precisely because he disagrees with the consensus, in technical detail.

Known as the “Godmother of AI”, Fei-Fei Li created ImageNet, the dataset that triggered the modern deep learning era. She now leads World Labs as CEO alongside her role at the Stanford Institute for Human-Centered AI.

Her work is the clearest window into spatial intelligence, the ability of AI to reason about the three-dimensional world rather than just text. World Labs raised $1 billion in February 2026 and released Atlas in September 2026, a model that generates, reconstructs and simulates 3D environments from a handful of photographs. That direction matters for anyone working on robotics, simulation or computer vision in physical environments.

She remains a leading voice on the ethical dimensions of the technology, which makes her relevant to organisations working through AI ethics in practice.

Sir Demis Hassabis shared the 2024 Nobel Prize in Chemistry for AlphaFold’s work on protein structure prediction. In August 2026 he stepped back from running Google DeepMind day to day, becoming its Chairman and Chief Scientist of Alphabet, to focus on longer-horizon direction. He remains CEO of Isomorphic Labs, the drug discovery company that raised $2.1 billion in May 2026.

His posts are infrequent and high level. Follow him for the direction of AI in science, pharmaceuticals, energy and materials, and for how deep learning research priorities are shifting at the frontier. Do not follow him for a steady feed.

Andrej Karpathy is the most trusted technical explainer in the field. A founding member of OpenAI and former Director of AI at Tesla, he joined Anthropic’s pre-training team in May 2026.

His explanations of how large language models represent, retrieve and hallucinate information are the standard reference for engineering teams. His nanochat project, a complete minimal train-and-serve pipeline for a ChatGPT-style model on a single GPU node, is one of the most instructive codebases available for teams building intuition about what is inside these systems.

He earns his place on substance, not on LinkedIn activity: his profile headline still lists Tesla and his last substantive post there predates his OpenAI, Eureka Labs and Anthropic roles. If you want his content, go to X or YouTube.

Comparison of expert focus areas

Influencer	Primary focus	Best for
Menno Fokkema	Strategy and implementation	Business leaders, EU compliance
Andrew Ng	Education and agents	Engineers, product managers, learners
Allie K. Miller	Business ROI and AI agents	Executives, startups, operations
Ethan Mollick	Future of work	HR, management, L&D
Cassie Kozyrkov	Decision intelligence	Data scientists, risk officers
Bernard Marr	Industry trends	Boards, strategy, non-technical leaders
Yann LeCun	World models, open source	Architects, open weights decisions
Fei-Fei Li	Spatial AI and vision	Robotics, healthcare, ethics
Demis Hassabis	AI for science	R&D, pharma, energy, materials
Andrej Karpathy	LLM internals	Developers, AI engineers

How to curate your AI feed

Building a high-quality information diet requires more than just following the right people. You must also actively filter the noise.

1. Prioritise builders over commentators. Practical experience usually beats theoretical observation.
2. Look for “how”, not just “wow”. Skip accounts posting demo videos captioned “mind-blowing”. Look for the ones explaining the workflow behind the result.
3. Check where someone actually posts. Several of the biggest names in AI barely use LinkedIn, and following a dormant profile does nothing for your feed.
4. Diversify deliberately. Mix technical voices (LeCun, Karpathy) with business strategists (Fokkema, Miller), so you do not over-index on technology that is not yet commercially viable.
5. Engage with what you want more of. The algorithm weights what you interact with.

Frequently asked questions