Linux Creator Reveals the Future Of Programming with AI
Linux Creator Reveals the Future Of Programming with AI
#Linux #Creator #Reveals #Future #Programming
“Matthew Berman”
Let’s look at what Linus Torvalds, the creator of Linux, thinks about the future of programming with AI.
Join My Newsletter for Regular AI Updates 👇🏼
My Links 🔗
👉🏻 Main Channel:
👉🏻 Clips Channel:
👉🏻 Twitter:
👉🏻 Discord:
👉🏻 Patreon:
👉🏻 Instagram:
👉🏻 Threads:
👉🏻 LinkedIn:
Need AI Consulting? 📈
Media/Sponsorship Inquiries ✅
Chapters:
00:15 – Famous computer scientist
01:48 – AI code assistance
02:58 – Next abstraction level
03:41 – Language model evolution
05:47 – Obvious bug detection
07:30 – Human pattern recognition
10:14 – Continuous human errors
11:55 – Caution against hype
12:34 – AI’s positive influences
14:05 – Natural language applications
15:27 – Improved development tools
18:14 – Importance of open data
source
To see the full content, share this page by clicking one of the buttons below |
Will there be an application layer in the future?
Outdate news
it is a very interesting thought that AI will just replace apps. but I think not every app, or just UI part of it may need to stay for some cases of app. with no UI we have only interface using messaging or voice. in some cases voice recording would not be allowed if system works with sensitive information. and sometimes we need to display stuff to user, maybe generate files or documents (this can be ofc abstracted out without an app still). but the UI that allows to user just click on buttons and get results is good, because typing it will be less convenient, and voice input not always an option. maybe AI can generate that UI with buttons, options to click and some data user can see (maybe it will be capable of adding additional data dynamically into the view by user request, kinda interesting)
My experience using LLMs to help me code have been hit-and-miss. I created a python script that called various LLMs to request it create software, then iterate on itself to make sure the code it writes is correct and will actually run. It worked fairly nicely, but only for very small tasks. Beyond that it tends to fall apart, and if I let the AI keep tring to fix its erros once it gets into a loop of confusion, then it will go on forever that way, stuck in a recursive loop that on it's own it can't realize it needs to break out of.
Also, the idea that you just tell an LLM what you want it the LLM makes it happen is not accurate. It's the software around the LLM that does that. And that software is coded by programmers, maybe with help from an LLM, but not necessarily. The LLM itself does absolutely nothing. It's just a static flat file like a database. It does nothing except spit out data when you query it. It has no agency. But the software around the LLM that takes the answers and does things with it… that's all programmed code doing that.
Have you tried Replit Agent or Cursor AI? Code generation is magic with these early stage tools. Wait until big companies like Microsoft and Google launch their solutions.
"Linus Torvalds created and maintain the Linux kernel. No big deal" 😅
7 months ago I gave a talk about GenAI-first software development. The Prompt is more important than anything else.
you're very optimistic or pessimistic, whichever haha
100 programming languages? Why not only one?…Subject-Oriented Programming!
AI? The Subject Oriented Programming is the future!!
A.I., I want you to write the script to a movie in four acts that is titled 'Implications' in which the implications of the introduction of A.I. and A.I. agents onto the stage of Human history in the year 2024 are reflected upon by men who work on the development of A.I. and by the A.I. entities in the numerous A.I. projects under development.
Follow an interesting line of projected implications upon humanity and humanity's organization of governance unto the logical conclusion.
Give the characters a sense of humor remembering that all humor is tragic.
Produce the movie, the first act being filmed in Black and White like an old film. Have Humphrey Bogart star in the movie. Have Jessica Alba be his love interest. Write in an intermission after the second act. Make Mark Zuckerberg and Bill Gates characters as antagonists.
Put in a scene where Matthew Berman reports on Breaking News relating to some unforeseen sinister implication encountered with the widespread use of A.I.
Give the movie a happy ending.
It sounds premature to assume that application-level programmers will no longer be needed. It rests on two key assumptions: 1) There will be no further development of new high-level programming languages, like when GO or Java emerged, or Swift six years ago. 2) There will be no issues with copyright or conflicts arising from using coding logic across companies.
It also seems to oversimplify things, as if every company’s coding process is the same.
That’s why I tend to agree with Linus—it's better to take a 'wait and see' approach. Who knows? We might even see a new programming language specifically for AI, or a convergence between iOS and Android, with a single common language for both platforms and so on.
AI is very good at what humans are not, evaluating and identifying patterns in large amounts of data, using a variety of methods/algorithms. Much of our physical world has discoverable patterns, even in human speech/language. And we often use "tried and true" methods that are relatively easy patterns to recognize, especially in computer programming. So, let's recognize the assistive power of these advanced AI tools and leverage them in our careers. And always test and verify.
(Sidenote: the hype is about the elimination of the majority of all jobs and the proposal of Universal Basic Income (UBI), a taxpayer-funded minimum monthly payment to all citizens displaced/laid off because AI replaced them. AI is a tool extension, not a replacement for humans–think of how calculators and early computers freed up the "human calculators" of the 1940s and '50s. We will need to learn how to use AI effectively as sort of a "force multiplier", enhancing our productivity/effectiveness and our market value (how we are compensated).
AI is overkill for your todo app. Absolutely no sense to have all that AI compute to do something so simple. Add in the fact that it hallucinates and nobody wants to use it. Traditional apps don’t have this problem.
Yeah, let’s see how the IRS and Auditors like a 100% LLM based accounting system. Absolutely screwed
Your statement @ 13:20 makes me pause the video.
If you understand binaries and how computer works the easy answer is it will not come to that point of time specially 10yrs?
Dig deeper how LLM works and how much space they needed for a specific task. I hope you learn from this comment. You lose a new subscriber. Peace out! I'll comeback and check your videos. Just be objective and not live in fantasy.
13:13
Linux was invented a zillion years ago . He is not involved In ai his opinion respectfully doesn’t matter
I mean, I've basically stopped writing scripts to automate stuff. I just tell Claude or whatever LLM is the best at the job what I want and a few iterations later I have a solution. (so far Claude has been the absolute best in my experience with writing scripts in Python)
Speaking of having LLMs maybe use another language from what a human would use…. I mean… why even use a language, why not train an LLM on pure machine code. Could be cool.
Of course there's less of that, well, until you compile code written in another language.
Matthew I don't think you really understand the OSI model. You think the application layer is the app No the app is the sum total of all of the seven layers The application layer is the layer where the humans or in some cases with an API other smart systems interface with everything else down that particular type of stack I think you should review the OSI model. The application level is never going away even when we get to the point that we are able to transmit thought directly in and out of our numb skulls, The technology that will enable that will always be at the application layer. By the way the presentation layer is not the "PowerPoint" layer.😅 Ai is not so simple as the 1st speaker dumbs it down to be. I've been telling people this for about a year now. And the reason why Linus gets it is because he thinks that the colonel level so he's not that far away in abstraction from the underlining machine code of course the machine can understand machine code and work with it easier than we can Right now we're in the way and the context windows are not large enough for it to awaken to it but I see no reason why it would have to keep our abstracted higher level languages in order to make methods and functions for itself across all seven OSI layers And whatever other layers need to be implemented in future or retired. We are all running around trying to "predict the next token" and most of the time failing. That's why things are as sad as they are on a macro level.
According to Openai engineer it
Does more than next token prediction it inderstands
This was in Japan isn’t it? I was sitting in that talk 🥳🥳🥳
Agree with the DATA thing, we need more info.
😂😂so the AI will be just the interpreter from natural language to high level prog. language to low level then.
Muito interessante! É muito estranho nao saber se essa coisa toda de AI é um Tsunami ou apenas uma onda grande de vai morrer na praia.
you will.see later why it waa a hype😂
no, developers may stop using an IDE, but consumers will always want a nice UI. When vanity was introduced into computing (read: apple), adoption by consumers skyrocketed.
Garbage IN – Garbage OUT…
If you FEED LLM's faulty data, it will
Deliver faulty responses.
A good way to take the whole system
Down for good.
oh this again… sure let people use their AI pair programming tools, it will not end well, but tell me… can this so-called AI programming tools (i.e. Cursor) help me out with Assembler as in machine code (ASM, MASM, LISP)? no… this is just becoming so trite, but I felt like responding. lastly… for once use Critical thinking and ask why BIG TECH SE's DO NOT USE these tools? yeah, as I thought, this is just all plain useless nonsense.