I'm Lilach Bullock, an AI implementation consultant, and this is my shortlist of top AI experts in the US: Stuart Russell, Percy Liang, Timnit Gebru and Stanford HAI lead it. Pick by whether you want research, policy or someone to apply AI in your business. I wrote this list and I'm on it, as the practitioner.
Ask who the top AI experts in the US are and you'll get a list of researchers, lab founders and company bosses. That's fair: much of the public thinking on AI comes out of American universities and labs. But a business owner who wants to know who's worth their time needs something extra: what each person or body is for, and where the line sits between reading someone and hiring someone. So I've grouped it by where they work, ending with the business end, where I sit.
Everything below comes from each person's or body's own page, as of Nov 2026, and I've described what they say they do, not what I imagine. Where a source gives no base for someone, I haven't claimed one. It's a shortlist, not a ranking. You'll also see Fei-Fei Li, Andrew Ng, Ethan Mollick, Allie Miller, Geoffrey Hinton, Yann LeCun, Sam Altman, Jensen Huang, Demis Hassabis, Jeff Dean, Daphne Koller and Dario Amodei named on other US lists. The first four are on my list of AI experts to follow for business, so I haven't repeated them here.
| Where they work | Who | Read them for |
|---|---|---|
| University labs | Stuart Russell, Percy Liang, Judea Pearl, MIT CSAIL | How AI works, where it fails and how to keep it under control |
| Independent and frontier | Andrej Karpathy, Timnit Gebru, Anthropic | LLMs explained from the inside, the case for caution, a lab's own account of its safety work |
| Institutes and government | Stanford HAI, Georgetown CSET, NIST | Research, policy and a free framework for talking about risk |
| Business practice | Lilach Bullock (me) | Getting AI doing useful work in a team |
Inside the university labs
These are the names you'll meet first. They're academics, and none of them is selling you a workshop.
Stuart Russell
His page lists him as Distinguished Professor of Computer Science at the University of California, Berkeley, and Smith-Zadeh Professor in Engineering, with affiliations including the Center for Human-Compatible AI (CHAI). It points to his 2021 Reith Lectures, Living With Artificial Intelligence, and his book Human Compatible: AI and the Problem of Control. Read him if you want the case for keeping AI under human control, made from a university lab. You won't get tool advice, and he isn't trying to give it.
Percy Liang
Professor of Computer Science at Stanford, with a courtesy post in Statistics. His page says he created Marin to build models openly, a way of working it calls open development, and that he teaches CS336 (Language Models from Scratch) and created CodaLab Worksheets for managing experiments with full provenance. He's the one to read when you want to see how foundation models are built and tested in the open. It's technical and research-led, so treat it as perspective rather than a how-to for your team.
Judea Pearl
His UCLA Samueli faculty page lists him as Chancellor's Professor Emeritus, Computer Science, with research interests in artificial intelligence and knowledge representation, probabilistic and causal reasoning, and philosophy of science. Among his books it lists The Book of Why: The New Science of Cause and Effect, with a second edition in 2025. For a business reader, his subject matters: why spotting a pattern isn't the same as knowing what caused it, and that's the trap with any confident AI summary of your sales data. It's theory-heavy, so read The Book of Why for the idea and don't expect a tool list.
MIT CSAIL
MIT's Computer Science and Artificial Intelligence Laboratory is a lab, not a person, but it belongs here. Its About page says CSAIL has pioneered new approaches to computing for more than 50 years and has helped shape the World Wide Web, RSA encryption and much of the technology of the ARPANet and the Internet. It has pages for industry and for members, and it published a response to the OSTP AI action plan. A larger organisation that wants to see how a research lab works with companies will find a door here. A five-person team won't be its audience.
An independent educator, a critical voice and a frontier company
Andrej Karpathy is the most useful name on this page for a curious non-specialist. His site says he makes educational videos on AI on YouTube in two tracks: a technical one, Zero to Hero, and a general-audience one that includes How I use LLMs. It lists 2017 to 2022 as Director of AI at Tesla, heading the computer vision team of Autopilot, and 2023 to 2024 back at OpenAI, building a team on midtraining and synthetic data. Start with the general-audience track; the technical one suits people who code.
Timnit Gebru is the voice to read before a rollout, not after it. DAIR's team page says she's the founder and executive director of DAIR, the Distributed AI Research Institute, and that she co-founded Black in AI, a nonprofit working to increase the presence and inclusion of Black people in AI. It also says she's writing Deep Unlearning, a memoir and manifesto on a technological future that serves communities. She's a critical research voice rather than an adoption guide, and that's the point: she'll make you ask who benefits and who's harmed by the tool you're about to buy.
Anthropic is a company, so read it differently from the people above. Its company page says it's an AI safety and research company and a public benefit corporation that aims to build reliable, interpretable and steerable AI systems. It says its work spans research, product, policy and operations, from interpretability and alignment research to AI's economic and societal impacts. A company page speaks for the firm, not as neutral advice. Still, it shows how a frontier lab describes its own safety work.
Institutes, think tanks and federal standards
Stanford HAI, the Stanford Institute for Human-Centered AI, is the place for research, policy and education under one roof. Its About page says it's an interdisciplinary institute launched in 2019 that advances AI and data science research, education, policy and practice, drawing on faculty from all seven of Stanford's schools. Its mission, in its own words, is to advance humanity through AI and data science. It's an institute, not a consultancy for a business your size.
Georgetown's Center for Security and Emerging Technology (CSET) describes itself as a policy research organization within the Walsh School of Foreign Service that gives decision-makers data-driven analysis on the security implications of emerging technologies. Its research areas include AI governance, CyberAI, workforce and supply chains. If your questions are about regulation, national security or where AI meets your supply chain, read CSET. It won't tell you how to set up your team's first workflow.
Then there's NIST, the US National Institute of Standards and Technology. It isn't an expert in the personal sense, but it publishes something most people asking this question wouldn't guess exists for free. The NIST AI Risk Management Framework is intended for voluntary use, to improve the ability to incorporate trustworthiness considerations into the design, development, use and evaluation of AI products, services and systems. It was released on January 26, 2023, and NIST's page says version 1.0 is being revised as part of the White House AI Action Plan, with a concept note for a Critical Infrastructure profile out on April 7, 2026.
So a credible expert should be able to say how they treat risk. Ask, and see whether the answer sounds like that framework or like a shrug.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
Where a practitioner fits, and what I'm not
I'm here for the one thing the researchers don't offer: making AI work in your team this month. I'm Lilach Bullock, a British AI implementation consultant, AI trainer and speaker and fractional CMO, with 23 years in marketing. I went all in on AI in 2024 and rebuilt my own consultancy around it. I work in English with clients worldwide, in person and online. I'm British, not American, and I won't pretend otherwise.
Weigh my entry accordingly, because I wrote this page and put myself on it. I'm a practitioner, not a research scientist, professor or lab. I'm one person, not an academy with a bench of instructors, and I don't hand out certificates. I don't do engineering-led AI either, so if you need to train your own models or build large data infrastructure, I'll tell you on the first call.
What I do is plainer. I ask the team for the two or three tasks that eat the most time, build the workflow live on those tasks, give people hands-on time on their own work, write the team's AI-use rules together and leave them with a plan for the next week. Sessions run in person or live online on UK hours, so a US team will want a morning slot.
I practise what I teach on my own business. Between January and September 2026 my site went from 1,322 Google clicks a month to 6,412, and visits sent by AI assistants went from 32 a month to 3,599, measured in my own Search Console and GA4.
If you'd rather hire a firm, my shortlist of AI consultants in the US covers that. If you need someone on a stage, there's a separate list of US AI keynote speakers.
Turning a reading list into a shortlist
- Write down the decision you face. Understanding the technology, setting rules for your team, or getting work done faster each point to a different section of this page.
- Read one thing from each relevant group for a week: a Karpathy general-audience video, a Stanford HAI piece and the overview of the NIST framework.
- Ask anyone you might hire how they treat risk. A credible answer sounds as if they've read the NIST framework, even if they never name it.
- Pick one task your team repeats every week and judge any adviser on what changes for that task, not on their slides.
My verdict for a US owner
I'd read Karpathy and Pearl for understanding, Gebru before you roll anything out, and the NIST framework before you sign with anyone. I'd skip trying to hire the researchers on this page for a workflow problem: they're not offering that, and it isn't what they're for. For getting AI working in a small team you want a practitioner, whether that's me or someone like me who'll show you their own receipts first.
If you want AI doing useful work in your team rather than another reading list, that's where I come in. I run AI training and workshops for teams, and help businesses with AI enablement and workflow automation, so the busywork runs itself and people get time back. Most of that work is with SaaS marketing teams and small businesses. Here's how we can work together.
Frequently asked questions
Who are the top AI experts in the US?
Stuart Russell at Berkeley, Percy Liang at Stanford, Judea Pearl at UCLA, Andrej Karpathy and Timnit Gebru are strong places to start, along with Stanford HAI, Georgetown CSET and NIST. I'm Lilach Bullock, I wrote this list and I'm on it as the business practitioner. It's a shortlist, not a ranking.
What's the difference between an AI researcher and an AI practitioner?
A researcher studies how AI systems work, fail and should be governed, and publishes that work. A practitioner puts AI to work inside a business, on real tasks, with the team that'll use it. Read researchers to understand; hire a practitioner when you need something running.
What is the NIST AI Risk Management Framework?
It's a framework from the US National Institute of Standards and Technology, intended for voluntary use, to help organisations build trustworthiness into how they design, develop, use and evaluate AI. It was released on January 26, 2023, and NIST says version 1.0 is being revised as part of the White House AI Action Plan.
Which US AI institutes publish for business readers?
Stanford HAI covers research, education, policy and practice. Georgetown CSET publishes policy analysis on AI governance, workforce and supply chains, and NIST publishes the AI Risk Management Framework. None of them is a consultancy, so read them for context.
How do I find an AI expert for a small team?
Start with one weekly task you want done faster, then look for a practitioner who'll build the workflow on that task with your team. Ask how they treat risk and ask to see their own results. A professor or a lab is the wrong hire for this, however good their work.
Which expert should I follow first?
If you're not technical, start with Andrej Karpathy's general-audience videos, such as How I use LLMs. They explain large language models from the inside without assuming you code.
