Soldato
- Joined
- 24 Sep 2007
- Posts
- 6,069
Yes I know it's good for programming. I'm just saying it's not intelligent, it's just advanced computing, i.e. everything it's doing is ultimately a calculation.
AI Section in forums
- Thread starter ginjaninja
- Start date
Associate
- Joined
- 3 Oct 2014
- Posts
- 1,784
You mean like the human brain. Basically just a bio computer?
How do you define intelligence?
If it can create a full application without me coding and by just describing in English what I want? what kind of human would I need to hire to complete that task and in the same time? I think I would have to find one of the best programmers on Earth to have it do what I did in that short of time.
Maybe in a few years I think we will find an eerily close relationship to how AI functions to how we function.
Last edited:
Creatives already use AI to be creative. It's enhancing their creativity. They can't lose a job they don't have.
It's programmers who don't seem to realize they are writing themselves out of their own careers. Tech lay offs are everywhere.
Most programmers write garbage applications on their own. It's why they usually need a team of non programmers with common sense to guide them.
Associate
- Joined
- 3 Oct 2014
- Posts
- 1,784
Bizarre thought you just made up.
Last edited:
But no denial...
The cost per seat of AI is eyewatering. We have to track usage and productivity gains from its use across the organization. We are actively looking for any opportunities to leverage it to justify the cost. If people aren't using it, they may lose their license to cut costs.
We've completed a few AI projects with some success. Been useful for some miscellaneous automation and documentation work that was tedious. I expect it will dominate the next quarter..
The cost per seat of AI is eyewatering. We have to track usage and productivity gains from its use across the organization. We are actively looking for any opportunities to leverage it to justify the cost. If people aren't using it, they may lose their license to cut costs.
We've completed a few AI projects with some success. Been useful for some miscellaneous automation and documentation work that was tedious. I expect it will dominate the next quarter..
Most of them do. They love reinventing the square wheel.
AI that hallucinates. Not a bug but a feature.
Last edited:
That isn't even remotely true, sure there are some for various reasons and sadly with the increase of it being a career there are some increasingly poor approaches by people with little real care for what they are doing.
The vast majority of software outside of large packages or bespoke industrial applications are often the work of small teams or even individuals often heavily loaded by programmers.
EDIT: In fact it is a horrible stereotype - sure there are a small number of "savants" who can be extraordinary coders but lacking overall development but your average good coder tends to conversely be well above average when it comes to common sense.
The vast majority of software outside of large packages or bespoke industrial applications are often the work of small teams or even individuals often heavily loaded by programmers.
EDIT: In fact it is a horrible stereotype - sure there are a small number of "savants" who can be extraordinary coders but lacking overall development but your average good coder tends to conversely be well above average when it comes to common sense.
Last edited:
I would say the opposite is true. The ratio of really good developers to mediocre ones is really low. Problem the majority think that are gods gift. They aren't.
But they think they are good at UX, UI management, everything, business analysis. The don't stay in their lane and they produce awful applications as a result.
This is why you have software like MS Teams where everything is illogical. You can see this in AI where they are so excited to see what they can do they don't stop to think of the wider implications. It gives wrong information so often and none of them thinks that's a problem.
But they think they are good at UX, UI management, everything, business analysis. The don't stay in their lane and they produce awful applications as a result.
This is why you have software like MS Teams where everything is illogical. You can see this in AI where they are so excited to see what they can do they don't stop to think of the wider implications. It gives wrong information so often and none of them thinks that's a problem.
Last edited:
There does seem to be something a bit off with the way its being financed though, this is worth a watch
Said every speculator during every bubble ever
Being a bubble doesnt mean the technology isnt useful or wont suevive, it just means the amount of current speculation based on future rewards isnt grounded in reality and that sums up AI to a T atm.
There are areas of coding, especially anything esoteric or innovative where AI falls flat on it's face - for example I recently wanted to use an ATTiny85 (AVR microcontroller) with USB monitoring but with modern Windows serial communications is largely a no go with the device (there are some hacked up V-USB drivers but that is a mess as well) but it does work as a USB HID device i.e. keyboard or game controller so my solution was to create a program that takes advantage of raw input's ability to create a background hook to a raw input device to handle the ATTiny as a joystick and stuff data into button and slider inputs (if I revisit the project I might look at creating a customised HID driver more suited to the task).
Whole program optimisation especially things like clever ways of packing data, etc. are still something AI struggles with - even with the newer models which can devote long time periods to reasoning and self-correction, etc. to improve the final product.
I used AI to create an small bit of automation recently, created the error handlers, file handling, progress bars, documented the code, all the tedious stuff done perfectly. How ever the actual task I was automating it couldn't do. Because it wasn't a generic task. I could correct it and thus the job was done in about 10% of the time it would have if I'd had to do it from scratch in a technology I'd not used in years. But if it's not in the AI library or lexicon it falls over.
Ironically some financial gurus had tried to it before me and failed. But that's because they weren't coders and didn't understand the framework or library.
The task arose as an export from a major financial system. However while the system could have done it, they hadn't a licence for those features, AI etc because it was too expensive. It was cheaper to use a person to join the dots.
I think there's a foreshadowing there.
Last edited:
There does seem to be something a bit off with the way its being financed though, this is worth a watch
Said every speculator during every bubble ever
Being a bubble doesnt mean the technology isnt useful or wont suevive, it just means the amount of current speculation based on future rewards isnt grounded in reality and that sums up AI to a T atm.
The funding model is not actually that unusual across many industries that require significant capital investment. Nvidia relied upon the success of Openai/Anthropic and by association companies mike core weave, and vice-versa. They all have vested interest in making themselves succeed together. Then there is competition between some of them and a landgrab for users and use cases, so they cannot afford not to invest sufficiently and miss out. There are plenty of 3rd parties and investors that want in on the action financially, which allow those actually creating the technology to get favorable moans and investments.
So sure, you can can make diagrams of interlinked deals that look suspect on the outside, but the deals themselves are sound. You will see the same with things like transportation projects, energy, construction, auto-industry.