Jensen Huang said if he were a student today, he wouldn’t prioritize coding. He’d prioritize learning how to talk to AI. Most people treat AI like Google. Type a question, get an answer, move on. Huang sees it differently. He calls it “expertise in artistry,” which sounds dramatic but... makes sense when you think about it. The real skill isn’t using AI. It’s knowing what to ask for and how to refine it. “Learning to interact with AI is not unlike being really good at asking questions.” If you’re a doctor, can you use AI to catch diagnoses you’d miss? If you’re a lawyer, can you sharpen arguments faster than your competition? The leverage comes from pairing what you know with how well you can direct the tool. Domain expertise multiplied by AI fluency equals amplification. Without the expertise, the AI is just noise. Without fluency, you’re leaving most of the capability on the table. The question isn’t whether AI will replace you. It’s whether someone who knows how to use it better will.show more

I have a tutorial for this:

There's a skill nobody's teaching yet: AI fluency. Fluency means: -Knowing when to trust output and when to push back -Breaking complex problems into AI-sized chunks -Iterating through conversation, not one-shots -Combining your domain knowledge with its pattern recognition I think in 3 years it'll be the most obvious divide in every industry.

The ability to code is learning to talk to machines. AI just changed the language from Python to English. But the meta-skill is the same: decomposing problems, giving precise instructions, iterating on failure. Self-taught devs already think this way — we've always had to teach ourselves through trial and error.

Coding is one of the most complex things the human mind can embrace. Its complexity develops the brain in multiple directions. Rhetorics, on the other hand, is a very precious skill, but highly superficial in comparison. So it could be argued that Jensen is advocating for a dumbing down of the human species…

A friend knowledgeable in AI told me at lunch that he took an 8 hour college class online to teach him how tonerote better prompts. When I got back to my desk I asked AI if it was familiar with this course. Of course if was. So I had it extract all the tips so that we could wrote a custom prompt that self refines my prompt with these tips baked in. I went further and asked it to find more courses and pull the very best tips from all of them. The result has been incredible. You really have to learn how to use AI properly to become effective and efficient with it. This custom prompt we developed was refined over a few months. Try it.

Acting like it's beneath you to use AI in your job will be to your detriment.

Everyone seems to think that AI will take over almost all software and engineering jobs, but fail to realize how many jobs, like prompt engineering, will pop up just as easily. There's a fundamental difference between using tools and leveraging them.

Like programming really, garbage in garbage out, and the more you know the more precise your prompt. Th level of abstraction you can communicate in, helps too. If you know how to talk about what you want, you can get what you want, and not always in English

This is what I tried to point out to @garyblack00 Gary Black When he posted that Grok said that $TSLA was overvalued. I tried to explain to him that I put his post in the Grok and that you have to know how to ask AI questions and I pivoted , asking Grok what if we look at Tesla as a holding company of startups and then Grok said it was undervalued and should be $515-$1100 share price And then Gary blocked me

"Jensen Huang said if he were a student today, he wouldn’t prioritize coding. He’d prioritize learning how to talk to AI." Yes! Learn to clearly describe what you want coded. AI Agents will increasingly be able to write the requested code. Then it’s up to you to use your domain expertise to vet what’s produced. The value is shifting from typing to thinking and verifying—making better use of the cognitive surplus unleashed by AI Agents powered by LLMs for human-to-machine language translation.

LLMs in particular are elite verbal pattern-matchers. It only makes sense that verbal acuity is an incredible advantage when working them.

One skill we gotta teach our teams/kids. How to articulate exactly what you’re looking to achieve. Not in terms of instruction, not in terms of parameters. But full and pure alignment on vision, meaning and PURPOSE.

Feels like the shift from writing code to writing intent—new literacy emerging fast.

I don't disagree with his point, but there is a possibility that ai has a significant fundamental change in the future (e.g. different training pattern) that makes it very different to interface with. In this case the prompters of today could be the obsolete coders of tomorrow

I’d still prioritize learning fundamentals though. If you don’t understand the domain, you can’t judge the output, you just accept confident nonsense.

The real threat isn't the AI itself; it's the asymmetry of leverage. The professional who treats AI as a collaborative intern will eventually outpace the one who treats it as a static encyclopedia. We are moving from a world of "doing" to a world of "directing."

I think AI's lure is to use it for extreme critical thinking. You have to ask it the right and tough questions for it to learn. Not "find me an RTX 5090 for a reasonable price." That's not training it to accomplish anything.

If AI makes better decisions with better prompts how do we prove the human was actually there?

@nvidia I noticed that in this clip Jensen Huang omitted to mention Deep Seek. I would be grateful if in the future he would include this when he lists examples of major AI platforms lest he should be accused of bias.

And who will build his chips?

I watched my sister interact with AI just the other day and was surprised and impressed at how she was able to direct and tailer the questions in a way that she got the information and advise she needed... I'm going to have to master this art!!

@grok Please use the context Jensen provides above to develop a prompt to put into action the advice provided.

100% agree. The question will shortly be, what do you do if you don’t have the domain expertise? How will you know what questions to ask? How will you know what answers to seek? If the current generation of experts no longer need to teach younger generations their expertise, only what questions to ask AI, in 10-15yrs who will be the experts?

How to use LLMs is obviously important, but I think understanding the fundamentals of physics and mathematics is far more important, especially in an era when it seems irrelevant. The future will be governed by specialists who can generalize their knowledge.

Jensen Huang’s concept of "expertise in artistry" perfectly captures the shift from syntax to strategy. 🏛️🎨 In 2026, the competitive edge isn't just about writing code; it's about having the vision to direct intelligence. The prompt is the new brush, but the masterpiece still requires a master’s intent. 💡💎

@grok , please take this information above and give me some feedback regarding how to be better at asking my questions and utilizing you for optimum results. Thanks

If you are a mechanic or electrician AI will never be able to compete and take your job. Neither will those doctors and lawyers sacked by AI. Choose your job wisely.

Total nonsense. Is it real video? So you do not need math or physics? It is a total crap.

It doesn't do any of that shit. The problem is AI chatbots are so heavily trained by the conventional wisdom -- petabytes of it -- that its inertia in that respect is enormous. It's freaking impossible to move it in the direction of novel thinking.

Exactly. I've been integrating AI into my SaaS tools and the difference between "ask question, get answer" users vs those who iterate and refine prompts is massive. It's like the difference between using a calculator and actually understanding math.

Books written for AI not for humans are the future

Of course he states this nonsense. He is a true believer. The problem with AI it makes things up. So it’s difficult to build knowledge if the foundation is confabulated.

That's fine. But the question is: would I pay for something like that? Something that could be wrong 5% of the time and require me to double check everything it said, just in case? Hell, no. AI today is like having a human friend who can be wrong sometimes. For free, I kinda like it. But I wouldn't pay for it. Just like I wouldn't buy a calculator which might give me wrong answers so that I would still need to do manual multiplication once in a while to be sure it was giving the right answer. And that's why AI companies will fail. Because there's no ROI. Companies have no reason to both hire a software coder AND an AI agent who sometimes writes buggy code. Doctors don't need an AI assistant that reads a patient's blood test wrong and mistakes the total triglycerides number for the LDL Cholesterol number. We want assistants who don't make mistakes. If we are going to pay for them.

Exactly what I tell my students when I’m stressing vocab matters…but not just with a single AI but when AIs of one company are selling something to another you want to be the human standing between communicating with the people on both sides what the AIs are saying..

Domain expertise multiplied by AI fluency really is the ultimate formula for leverage.