News Jensen Huang advises against learning to code - leave it up to AI

Status
Not open for further replies.

Findecanor

Distinguished
Apr 7, 2015
269
186
18,860
Manager with little sense are going to ingurgitate these words, and use them as reason for laying off more software developers.
The remaining staff is going to be overworked and overstressed (as if they weren't already) and quality will suffer.
When everyone in the modern world depends on software for their daily lives, everyone will suffer.
 

CmdrShepard

Prominent
Dec 18, 2023
314
234
560
It is our job to create computing technology such that nobody has to program. And that the programming language is human
There's 7,139 human languages, he should've been a bit more specific.
Everybody in the world is now a programmer. This is the miracle of artificial intelligence.
Nice try, but current models aren't intelligence -- they are just advanced auto-complete which is wrong more often than not.

"Everybody in the world" will never be a programmer.

Programming requires skills beyond that of knowing the language syntax. You need to be good at organizing, tracking multiple things at once, breaking down complex tasks into simple self-contained steps, and creating high-level abstractions to avoid repetition.

Inform 7 already enables you to write computer code in plain English. The thing is, if you can't write code in a regular programming language you probably won't be any better in English either.
 

DiegoSynth

Prominent
Jan 15, 2023
13
21
515
I'm totally fine with this, I very much accept it.
Now, that being said:
1) Jensen, man, it's time to retire. Soon you'll be needing the diapers, someone to feed you, clean you, and these things.
2) Jensen, while you are still there, fix my situation: cover my expenses while your AI does the coding. Because this also applies to every profession. I would be a carpenter. But nowadays everything is particle boards, prefabs, mdf.
So if you replace people, then maintain us while we enjoy life. Deal?
 

vanadiel007

Distinguished
Oct 21, 2015
234
228
18,960
Nvidia CEO Jensen Huang has shared advice for young professionals trying to steer a career path during the fast-changing AI era. In a nutshell, Huang appealed to those forging their careers to “Dedicate yourself to learning all the time and doing the best possible work you can” and said that he's loved every job he's worked — including cleaning bathrooms.

https://www.tomshardware.com/pc-com...-out-career-tips-for-the-fast-changing-ai-era

I guess he changes his mind and wants everyone to stop learning right now.
The best customer is an uninformed customer. He's smarter than I thought.
 

35below0

Commendable
Jan 3, 2024
1,279
564
1,590
If he loved cleaning, he's dead lying. That is not a forever job. That is a turd sandwich.

It's an important job, and one that cannot be given to a machine. It's valid experience and requires dealing with stress and fatigue. Also, it is the pits and no human being could find it tolerable, much less love it.
Do it for 1-2 years, tops.

He's another guy saying stupid things in order to be called out on it. Don't bother.
 
  • Like
Reactions: artk2219

Phyzzi

Commendable
Oct 26, 2021
8
10
1,515
"Why learn math? Just use a calculator. "

Simpleton. 🙄
I think there is a really valid argument that learning a bunch of syntax or even the idiosyncrasies of most programming languages will soon be as important as knowing long division or how to use a slide rule. Probably everyone should learn basics of at least logic, case statements, and loops, but I hope my kids never have to deal with a pointer that won't clear. And, just like a calculator doesn't tell you how to set up a problem to solve something in the real world, and is pretty unlikely to do a good job with something like gradients if you don't understand how something that complicated works well enough to modestly anticipate the inputs and outputs, programming will continue to require people who understand more or less what the computer is probably actually doing in order to create even adequate results.

Given a half century developing higher level programming languages, game engines, libraries for OOL's, "text editors" that flag syntax errors, drag and drop "visual" or "block code" languages and other programming tools, there isn't actually that much room for AI to bring much more than a little efficiency and possibly a new avenue to "throw stuff at a wall and see what sticks".

On the other hand, I feel like AI has a huge opportunity to evaluate management and eliminate ineffectual and toxic leaders, and help investors know when higher level leaders are not worth their costs. Say, when c-suite leaders start reviewing their competition or talking about how their product will supplant (instead of supplement) human creativity in the near future. Not that I'm thinking of any PARTICULAR C-suite doofs or anything.
 

bit_user

Polypheme
Ambassador
I think we can take lessons from the effects EDA (Electronic Design Automation) had on hardware design. Electrical engineers still learn how to draw schematics and design circuits. They still learn what the tools should be doing and what constraints they have to negotiate. This can be used to validate some of the decisions they make, as well as guiding higher-level decisions about chip design.

I can imagine a world where we use even higher-level languages to avoid unintentional ambiguity in how we specify what we want AI to build us. Natural languages, in general, aren't great for specifying exactly what you want. Especially English. Just think back to some legal contracts you've read, and how many hoops lawyers have to jump through to try and say exactly what they mean (and sometimes they still get it wrong)!

I think there will still be plenty of specialized niches, where coding by hand remains justifiable. Compilers didn't completely eliminate assembly language. High-level languages didn't completely kill low-level ones. I think AI won't kill off programming, but I could see it take a big chunk out of that job market.
 

toffty

Distinguished
Feb 11, 2015
71
55
18,610
So many ignorant statements being made here...

As a software engineer with 20 years experience in many regiemes, I can't agree more with Huang.

I've been using copilot since it's been in beta and the difference from then to now is astonishing - ~1 year.

He's talking within 10 years. Certainly. I'll luckily be ready to retire in that time frame and very much expect to be replaced by AI around that time frame.

Will AI make mistakes? Sure. But much fewer than humans.

What changes? How software is developed. Instead of humans coding, they ask for code to be written and verify it's what they want by testing it. If it's not, they ask for corrections and the code is revised. The integration will remain with humans for a while longer after that point but that will be automated soon after.

Example: already we have AI that can build basic websites in their entirety.

DL AI has a doubling time of roughly 6 months. How many doublings will it take to get from beginner programmer to superior programmer? 2, 4, 10? Then that's 1, 2, or 5 years time.

And remember, doubling means exponential growth
 
Last edited:

USAFRet

Titan
Moderator
Instead of humans coding, they ask for code to be written and verify it's what they want.
And unless the human knows good from bad, they cannot verify.

I've literally seen this at work.

Junior member gets some code from random AI, presents it to the team.
It sort of works, but absolutely not a good solution.

The issue is...he did not know it sucked.
 
Definitely not without merit, but given how the coding world is today, I doubt many of us would live long enough to see any AI replace regular humans for coding, testing and even design (architectural).

There's many more pieces to software development than just being a code monkey and if AI was to replace humans in that very specific role, many things would need to change in terms of process and approaches to delivery (SDLC, in particular).

Also, as a side note. People can barely write using their own respective languages and communicate effectively. I sincerely doubt using AI will actually work unless humans learn to use their own friggen language as intended.

Regards.
 

ThomasKinsley

Prominent
Oct 4, 2023
226
223
460
This advice sounds good to non-coders, but upon deeper thought it sounds terrible. Human experts will still be needed to build future versions of AI, to verify AI code, and to determine future visions of programs. It's the difference between an artisan steak burger and a McDonald's cheeseburger. You get what you pay for.
 

vertuallinsanity

Prominent
May 11, 2022
34
14
535
So many ignorant statements being made here...

As a software engineer with 20 years experience in many regiemes, I can't agree more with Huang.

I've been using copilot since it's been in beta and the difference from then to now is astonishing - ~1 year.

He's talking within 10 years. Certainly. I'll luckily be ready to retire in that time frame and very much expect to be replaced by AI around that time frame.

Will AI make mistakes? Sure. But much fewer than humans.

What changes? How software is developed. Instead of humans coding, they ask for code to be written and verify it's what they want by testing it. If it's not, they ask for corrections and the code is revised. The integration will remain with humans for a while longer after that point but that will be automated soon after.

Example: already we have AI that can build basic websites in their entirety.

DL AI has a doubling time of roughly 6 months. How many doublings will it take to get from beginner programmer to superior programmer? 2, 4, 10? Then that's 1, 2, or 5 years time.

And remember, doubling means exponential growth

He made the statement in Dubai. Dubai, UAE. Have you heard the details regarding "The Line"?

To move there you need to give up access to a huge amount of personal data. Personal, private (used to be), data.. Which will, of course, only be used to improve the space.

Complete B.S. Of course they dont want coders. It's Facebook, Google, Apple, etc pushed to the limits of data collection and, in response, human control.

Replacing human coders with AI is a stupid concept. So, that's it? 640 kilobytes of human coding/coders is all we need? :)

iykyk..
 
  • Like
Reactions: artk2219

toffty

Distinguished
Feb 11, 2015
71
55
18,610
And unless the human knows good from bad, they cannot verify.

I've literally seen this at work.

Junior member gets some code from random AI, presents it to the team.
It sort of works, but absolutely not a good solution.

The issue is...he did not know it sucked.
And you're talking about today's AI. I am not. I'm talking about 20 doubling times from now 😉
 
  • Like
Reactions: artk2219
Status
Not open for further replies.