To think of dedicating my next 5- 10 years for this

[computah]

I am a computer science engineering graduate (not an expert though)
So i have been thinking to dedicate a lot time starting from now on this idea (obviously without compromising on my job and will study for this and team up with others too)
I want to know if is it worth working on..or should i let it go..haha
Idea: I want to create an application or ai tech that helps to solve any kind of problem.
The users have to write the problem in detail. The ai technology in the app will comprehend it...and will turn the problem into a picture/pattern/some abstact puzzle...(with the help of pre programed programs)..and these puzzles will be solved by high IQ people..and once the puzzle is solved. .it will convert the puzzle solution into a solution for the user..using ai..as it knows the jargon/terms i that field (the software also uses heruestics and a lot of new thinking models)...if the solution works for the user...it is great...if not...we repeat the process..but this time ai analyse what went wrong last time..so that there is a better chance of getting the solution the next time...we repeat the process till we solve the problem..
(+ the ai suggestes solution that has lot of open ends .users have to think again to get the solution..this ai only gives that eureka factor)
I want to make it a powerful tool and also only licensed professionals like scientists ,psychologists etc who are doing research or business men who need not solve busines problems can use it a high price...high price to avoid misuse

business men
🙄

😀

Maybe their penises fell off and they weren’t able to do their big business problem-solving any more. Hence needing AI…?

?

Or the 'high IQ people' could solve the actual problem, with no need for the 'turning it into a puzzle' step.

I've just worked out the world ends. People just bore each other to death.

<Wonders whether I should have charged a high price for my answer, to prevent it being misused.>

Sorry, what exactly do you mean?

The first trick is in The users have to write the problem in detail.

You've not met many users huh... The whole reason some people are computer programmers/analysts and some aren't is that 'writing a problem in detail' is bloody difficult to do.

"Being able to write a description of the problem in terms that a computer will understand it" is basically a job description for "computer programmer" though, isn't it?

If AI means that perhaps one doesn't need to get the grammar and punctuation completely perfect as we do in current programming languages then that will free up our time for the bigger-picture issues, but fundamentally the job of comprehending a human scale issue and formulating what needs to be done for a computer to do the necessary computing to produce a solution is already the job we do and AI will only change what the interface is like for that process. It won't give non-programmers the skills to use that interface successfully. (Someone who learns how to use it successfully will de facto be a programmer)

No, because you didn’t turn it into a puzzle en route to replying…?

I’m now puzzled about the puzzle diversion.

OP, do you know many scientists? Generally their actual job involves solving their own problems.

This is already a pattern that is in use, although in much more restricted use cases. In essence what you are doing is ‘gameifying’ ai model training. It’s not as outrageous as it sounds really but I am not sure ai is there yet wrt applying this to more general use cases.

If I want to know how an expert in their field solves a problem i type the problem into YouTube.

I've been in tech/SW for 25 years how would you make money from it?
I don't see that from what you said.

I don't see this working out. If I were to give you a problem to solve, surely the system would need to know all of my company-specific policies and SOPs. I would also need to share commercially sensitive information in order to give you enough detail to solve the problem which would be a huge no-no. Overall, it would be better to seek insight from more senior members of staff within my organisation who would use their experience to help resolve the problem.

I mean this in the kindest way possible: now that you have graduated, go and get a 9-5 job as that will help you to understand the real problems that are faced in the workplace, and maybe then work on a machine learning project to solve one of these. You can't use one machine learning algorithm to solve literally every problem anyone could possibly come across ever.

Putting aside the issue of actually making it into a sustainable system, do you have any idea how you'd fund such an endeavour? A system capable of doing what you're asking it to would be an order of magnitude more complex than ChatGPT - and just one machine out of the thousands running the training process alone is buy-a-decent-house money.

On top of that, in order to get as far as raising money like that, you need to be able to demonstrate a fully-working prototype, not just an idea in a notebook (or even a fully-costed business plan). If you're not an expert...who's going to build it? AI isn't your standard development tool, it requires real (and deep) mathematical understanding, as well as constant maintenance and tweaking for a service like that.

And...finally, your concept doesn't even work if you don't have a sufficiently large pool of experts backing it. That's gonna be your main problem, I think; that's a population of high-net-worth individuals, for whom time is incredibly valuable, and you're going to have to somehow persuade them to spend that time on your puzzles. I would guess that's going to cost you as much as the compute resources will. And don't forget someone's going to have to check the final solutions returned to the users, because - as we all know - the #1 problem with AI answers is that when they get it wrong, they're absolutely 100% confident that they're right.

It's an interesting concept, but it strikes me as the sort of thing that's destined to fail on financial, conceptual and engineering grounds.

If you really think you're on to something, though, then you need to prove the basics of the concept. Act as the AI yourself - take a real-world set of requirements, and convert them into a puzzle for a set of friendly experts. Get them to solve those puzzles, then translate the solutions into a single response to the original requirements.

If you can't do that - whether it's because you can't come up with a way to create the puzzles, or a way to translate them back, or even if you can't find the friendly experts - then you're tilting at windmills.

If you can, spend another year writing a white paper on it and come up with a business plan.

YABU. if you dedicate 5-10 years of your life to this, at the end you will have a mishmash of code that doesn't achieve what you want.

writing a computer program that can solve general problems (i.e. general AI) is a very difficult challenge that a lot of big companies (google / apple etc) are trying to solve but many people in the field believe it inherently can't be solved with current approaches.

and I don't see what the bit about translating problems into puzzles will achieve.

if you want to put your coding skills to good use, find one small niche problem and fix that. you could sign up to something like https://socialcoder.org/Home/Index and find projects that will actually make a difference to people.

This has to be a windup. Surely OP you don't think Google et al are already on it?

Most successful businesses work because they bring a simple solution to a huge customer base. It is called leveraging. Solutions that cook rice perfectly every time or prevent socks from disappearing in the washing machine or get you through to a real person at HMRC within 10 minutes will make you lots of money.

The market for what you are offering is very small but probably premium priced. There are already legions of experts and consultants in every sector. I suspect what you are really talking about is inventing a holistic solution to any large scale project. In other words "Solve congestion in Nottingham" will find a solution that deals with financing, public health, budgeting costs, materials, labour, compensation, environmental net-zero, etc. If that is where you are heading you need access to a lot of data. It will be an expensive invention and you will need well into 8 figures of capital to fund this, probably 9. But go for it. If you have partial success one of the bigger tech firms will always buy you out (to suppress your work).

42