Yeah AI generated stuff generally isn't copyright protected. In the world of AI, you might not even want to file patents, your idea will be processed and executed on by AI before you can blink
So much AI, so many add-ons... and so many bugs
- Thread starter PhineasD
- Start date
I'm more curious to see if anyone can develop something simultaneously, ie. write the brief, then let a human do it and then let an AI do it, and see which is better coded. I would put my money on the AI if using one of the top tier ones.
I love the:
Me: "Good morning! Let's knock this out today"
CC: Processes something in 10 seconds
CC: "That's done. Should we call it a wrap for the day?"
If the first author using AI created a process for whatever, modified to his/her tastes or needs even just esthetically, it can be copyrighted. Anyone can theoretically copy raw code from AI straight to page or process, and it becomes uncopyrightable. But if the output is modified, that is a different scenario. (my understanding of these hypothesis).
I think you would say the same for simple html generators that make a page for you, always has been able to copyright.
I imagine precedent has to be made before anything becomes an issue.
... these big AI companies are all bleeding cash for years now with no profitability in sight.
Of course they want to reduce resource consumption.
They also recently jacked up prices quite a bit... and it backfired.
Two interesting recent developments in response to these prices being jacked up:
Yeah.. Microsoft is running Chinese open source models on their hardware, instead of the stuff they funded/developed
Open models are working for us for a while.. we have a 128gb VRAM setup and it is pretty beastly. Lately we have good luck operating it in an IDE so we can do a mixture of human and LLM labor type approach that's much more supervised than vibe coding.
I expect that AI coding with big companies will gradually get more expensive and restrictive.. the big companies jacked up their prices... other big companies saw the resulting bill, then started adopting open source. At the jacked up prices, no company is making a profit still.
Well it's pretty bleak IMO. My thoughts about it from a business perspective..
There are also lots of negative economic signals on vibe coded software..
I've found out many times that it's usually not worth the effort trying... have yet to see an exception... so my observations match the signals.
There is so much low quality AI output flooding things, that it puts a huge damper on using the internet to find things that a human put serious effort into.
The human touch will become more valuable, not less, over time.
That I have disagree with. We as end-users have been programmed since the beginning of PC's in the households to accept the technology that we have isn't good enough and the software releases and updates quite often requires equipment upgrades. And then the new features become sanitized to be human friendly so it becomes more widely acceptable.
Software has always driven tech and software has always driven user experience. And the human experience then eventually becomes integrated, and the cycle begins all over again.
Looking at the slowing contextual ability of humans and the degrading of the ability of critical thinking processes, it just tell me that we as a whole are stupid enough to accept and develop skills to mirror the expectations (and limitations) of the tools.
IE: "The human touch" value is a fallacy to believe in. We tend to change to accept "the computer touch" out of convenience as valid more and more, and we will see soon enough that ai IS the software, and our reliance on has AI becomes undivided.
My personal opinion is that AI is a tool. Not one I use often, not one that I cannot gain understanding of. I do not believe AI belongs in my household or doing anything in medical other than ASSISTING in diagnosis, etc.
I digress, but my point is valid that we will adapt to AI as the only real choice, and the human touch will go the way of the likes of human phone customer support for example. I used AI in my current dev for a site (yeah, vibe coding) , I am a noob if you will with it. But alas my coding skills are not up to par as my visions are, and I am not in position to hire coders for everything. And low quality outputs of AI are expected but will be 'humanized' (so to speak as currently they just patronize outputs) and guard railed in due time.
my 2c.
(edit to add: this in no way suggests I am a proponent of AI or its current directions. Quite the opposite. But I am an absolute realist.)
Last edited:
People keep talking about low quality ai coded output. Is that not the same for human coded output? You have a sliding scale on both IMO. Put someone into coding who doesn't know it well, you get crap to average. Use an open source cheap coding model to code and you get crap. Then you have those who are thinking about everything coding, know it well and implement well. You have quality AI models and checking systems that cost $$$, but they do it as well or better than the best coders on the planet. That AI top end coding, is cheaper than a human top end coder. It does it faster, cheaper, same or better quality, completely secure, thinks about everything security, things most coders don't know.
I'm sorry, but when the best coders in the world know this and those who are average or better just can't get onboard, this is more hubris, pride, getting in-between fantasy and realism. AI will save more lives in surgery than a human can. Its already starting to do that, and its only early in medicine and limited, but the experiments are wild and fascinating. We will all live longer due to AI, that is if a military AI doesn't decide to kill us all.
I'm sorry, but when the best coders in the world know this and those who are average or better just can't get onboard, this is more hubris, pride, getting in-between fantasy and realism. AI will save more lives in surgery than a human can. Its already starting to do that, and its only early in medicine and limited, but the experiments are wild and fascinating. We will all live longer due to AI, that is if a military AI doesn't decide to kill us all.
People keep talking about low quality ai coded output. Is that not the same for human coded output? You have a sliding scale on both IMO. Put someone into coding who doesn't know it well, you get crap to average. Use an open source cheap coding model to code and you get crap. Then you have those who are thinking about everything coding, know it well and implement well. You have quality AI models and checking systems that cost $$$, but they do it as well or better than the best coders on the planet. That AI top end coding, is cheaper than a human top end coder. It does it faster, cheaper, same or better quality, completely secure, thinks about everything security, things most coders don't know.
I'm sorry, but when the best coders in the world know this and those who are average or better just can't get onboard, this is more hubris, pride, getting in-between fantasy and realism. AI will save more lives in surgery than a human can. Its already starting to do that, and its only early in medicine and limited, but the experiments are wild and fascinating. We will all live longer due to AI, that is if a military AI doesn't decide to kill us all.
As anyone who has watched technology over lengths of time, AI costs will drop substantially to the point of free to the end user. (except enterprise models). Examples: browsers, data access, html generators, cell phones (heavily subsidized by advertisers and approved app creators), long distance costs, I could go on. Good and costly will evolve into great and cheap eventually. And we will accept it as quickly as we accepted music streaming services. And it all wont be good for society. It will though drive the software to improve and replace the current concept of 'humanization' even further.
I really feel bad for all those college grads coming up in computer tech degrees as their job opportunities are evaporating faster than they can receive their Parchment.
Too late for that worry.
Government AI Is Coming for Your Data
The government wants to use AI to analyze Americans’ information obtained without a warrant though purchases from data brokers and “incidental” collection from foreign intelligence surveillance. Congress must act now and demand closures these loopholes around our rights before any renewal of...
epic.org
Yep. I would be doing a trade... about the safest thing right now. AI isn't replacing builders, electricians, plumbers, any time soon just yet. That will come too I am sure, but it isn't in the immediate future as some jobs are, such as IT. There will still be IT human requirements for the foreseeable future, eventually that will end, but those jobs will be limited and highly sought. You will need to be very good, basically to supervise / help AI.
I run a development shop where everyone is at senior level, and we work on hard, long running projects that are difficult to design.
Our experience is very similar to that at big companies where AI provided an initial boost then dragged us down.
The code quality is always lower than us and it leaves little fun things to discover that bite us.
I've dumped every company that replaced human support with AI support because i got blocked by the clanker. Tons of companies have lost money trying this because other people won't put up with it either. What i'm noticing is that some companies still have AI doing receptionist duty, but not actually support.
A lot of big corporations are finding out that expectations are not matching reality there.
An overgrown calculator is still unable to think like a human. Who would have known!
This completely explains your positon. You aren't skilled in the area you're evaluating.
New AI adopters go through this curve. Eventually they learn AI's limitations. But before that, they're going to overestimate them to an extreme.
You don't know what you don't know yet.
I'm not singling you out. This one is continuing to one-shot many CEOs. Only recently are we hearing the failure stories. The one reason why CEOs are vulnerable to this is because they're the furthest from the ground.
Not with me continuously hiring them.
Wish AI could do more of our work actually.
Vibe coding is great for prototypes and for MVPs.
It's very good at things that have been done before ( but nobody needs that )
And snippets or low responsibility/decorative things.
It's terrible on established projects or anything a user will rely on long term. Very smart people are trying to get that work but not succeeding yet.
It's very bad when doing novel things ( needs it's hand held a lot, you might as well code it yourself ).
The code it produces often needs to be refactored to hit code quality standards on a project... a big negative.
It's very bad at continually modifying it's code, so if you don't want to work on it by hand but need to modify it often, eventually it's going to make a huge mess and you'll have to redo that part of the system.
We do a lot of big established projects, zero mvps, zero common projects, and zero one trick pony apps...
..people who are really impressed with AI tend to be doing the opposite ( small, low responsibility software )
Our experience is very similar to that at big companies where AI provided an initial boost then dragged us down.
The code quality is always lower than us and it leaves little fun things to discover that bite us.
I've dumped every company that replaced human support with AI support because i got blocked by the clanker. Tons of companies have lost money trying this because other people won't put up with it either. What i'm noticing is that some companies still have AI doing receptionist duty, but not actually support.
A lot of big corporations are finding out that expectations are not matching reality there.
An overgrown calculator is still unable to think like a human. Who would have known!
This completely explains your positon. You aren't skilled in the area you're evaluating.
New AI adopters go through this curve. Eventually they learn AI's limitations. But before that, they're going to overestimate them to an extreme.
You don't know what you don't know yet.
I'm not singling you out. This one is continuing to one-shot many CEOs. Only recently are we hearing the failure stories. The one reason why CEOs are vulnerable to this is because they're the furthest from the ground.
Not with me continuously hiring them.
Wish AI could do more of our work actually.
Vibe coding is great for prototypes and for MVPs.
It's very good at things that have been done before ( but nobody needs that )
And snippets or low responsibility/decorative things.
It's terrible on established projects or anything a user will rely on long term. Very smart people are trying to get that work but not succeeding yet.
It's very bad when doing novel things ( needs it's hand held a lot, you might as well code it yourself ).
The code it produces often needs to be refactored to hit code quality standards on a project... a big negative.
It's very bad at continually modifying it's code, so if you don't want to work on it by hand but need to modify it often, eventually it's going to make a huge mess and you'll have to redo that part of the system.
We do a lot of big established projects, zero mvps, zero common projects, and zero one trick pony apps...
..people who are really impressed with AI tend to be doing the opposite ( small, low responsibility software )
Last edited:
ThisToo late for that worry.
Government AI Is Coming for Your Data
The government wants to use AI to analyze Americans’ information obtained without a warrant though purchases from data brokers and “incidental” collection from foreign intelligence surveillance. Congress must act now and demand closures these loopholes around our rights before any renewal of...
I've always been cautious with the use of AI because of the criminal activity of it being misused.
The use of AI has even got robots ringing your mobile phone and recording your voice to make you pay for scammers garbage.
And I completely willfully accept as it is true. I am well on my way to "mount stupid", but I know exactly what that means. And knowing my background I will be hiking off that mountain.
Yer... from my findings, AI lacks memory capacity that humans have. That seems to be its biggest flaw in coding projects as I see it. Code is code is code, if PHP add something, then everyone knows, including AI, there is nothing to be used that isn't available per existing software. The issue as I see it, memory capacity. AI really struggles in this area. If you can't remember what you've done, then you're prone to many errors - I guess hence its good at smaller projects.
Partially a context management issue. Not perfect but you can always improve it
When I created my AI Substack https://ai.georgeliu.com I wanted to write about my AI build journey itself and not just the end results. That meant I needed to level up the memory systems and processes so it not only remembers what I did but the decisions behind my choices - basically a per project dev history journal of what I did every session. I can spin up a fresh session and it immediately knows what I did last session, any session before than, what my next task is and also the decisions behind my choices and implementation paths made or dismissed
Works nicely in Claude Code CLI/desktop but needed a different approach for Claude Cowork due to isolated sandbox projects where I needed to build a Claude Cowork session MCP that can read across isolated sandbox project sessions https://ai.georgeliu.com/p/i-built-an-mcp-server-so-claude-cowork and an Obsidian 2nd brain https://ai.georgeliu.com/t/obsidian
Yeah Fable 5 suspension also woke some folks up. I've added additional models to my multi-AI consultation skills since Fable 5 suspension.