Our strategy bet is that UI is obsolete, and that AI and natural language interfaces are rug pulling the entire "frontend development space".
In this video I am managing our Kubernetes cluster, sending customised marketing emails, managing payments and customers, and the whole shebang from a custom GPT, connected to our internal management API.
It allows us to send payment links, create Kubernetes deployments, tear down deployments, and everything related to "administrating and managing our company" (we're a SaaS / PaaS company)
Is there a single thing you actually need a UI for today? Ignoring Photo Shop and such? Our bet is that at least for admin apps, AI will conquer UI completely ...
We crazy ...?
Thoughts ...?
Advantages are among other things that we got rid of 70% of technical debt, and the above Custom GPT was created in some 5 to 10 hours. Traditional software development (with UI) would require hundreds of hours, possibly thousands to deliver the same ...
Top comments (9)
I think AI is gradually replacing bad UI — and dramatically expediting the demise of really bad UI, but I think it will actually help bread new forms of better point and click UI.
Building UI is hard and one big impact of AI, in my opinion, will be the capacity to help people better handle the complexity of great UI.
What "great" is here is hard to get a handle on because we haven't seen it yet, but that is my prediction.
If you don't care about UX, then maybe.
Back office apps that just manage DBs might as well be replaced completely by AI solutions. Backend and frontend.
In fact no-code, AI generated backend is more likely to replace backend development in those cases. Users don't care what happens behind the scenes. But even for those kind of apps, they do care about what they see.
Also, customer facing products need a better UI/UX, that AI can't provide (yet).
Just as it creates human-like images with 6 fingers, it's UI/UX is for human-like creatures. (no offence robot)
You missed the point. The point wasn't AI generating frontend code replacing frontend development, the point was an entirely new user interface, based upon natural language, replacing UI as a whole.
Oh my bad.
Well, I've never seen a bot using natural language work as expected.
I almost always hang up the phone infuriated when I try to talk to the bank or get healthcare assistance from a bot.
And I always end up talking to a person when a chatbot tries to service my request.
Also I don't think that end users prefer oral commands or typing (eughhh) over clicks on a screen.
The day a bot will explain to me how my credit card's APR is calculated using natural language and be able to convince me to not to cancel my subscription seems far away!
Typing commands are a temporary problem, we're a year (tops), probably months away from being able to fluently speak to them, 1,000x better than Google's speech recognition and synthesis. As to your APR problem ...
The future is much closer than you think ... 😉
My point is that these kind of "stock/wiki" type of answers, do not help the average person.
I could have typed "How APR is calculated" in Google and get the same result.
It wouldn't help either.
Scanning through the internet and composing an answer with what you get is not Artificial Intelligence imho.
But anyway this is another discussion...
I wish this day comes sooner rather than later, but I am not going to be holding my breath.
Cheers!
There is some credit to your answer, but I suspect you under estimate the thing quite a lot, especially considering its motion, trajectory and momentum ...
In my ~15 years of software development, I have realized that it's just too complicated to instruct a processor to communicate with a human properly.
So I reckon that, if it's so difficult to make software that works well, without spending a lot of resources, it's not going to be that easy to have software create software for you for free (meaning human resources free not $$ per se)
Anyway, when that day comes, I think all humans are going to be happier, since a lot of our problems are going to be resolved.
This is why you use deep learning and not "instructions". According to a study on ChatGPT4, it not only provides 18% more accurate diagnostic services than the average GP, but it also scores 30% better on empathy ...
You've got one point though, which is that there will always be needed a smart human being capable of telling the machine what it wants for the machine to do - Which implies smart people (people like us), will (probably) always be sought after, to solve hard problems, using AI as cognitive assistants though I suspect ...