AI and program reverse engineering
-
The more I learn about AI the more it scares and excites me, but mostly scares me. I saw something somehere (I wish I could find my source for that) that soon if not now that what modelling all protiens would take AI a matter of minutes but it would take specialists PHD's a billion years to achieve that.
I wonder how long it will be before I can promp chatgpt version 7
to "reverse engineer Sketchup (or any other program) by creating a code to replicate the functionality of sketchup and deliver it as an open source 3D modelling program"
I would also task it to have the ability to creat 3d models from images and sketches...........just goes on doesn't it
-
I have heard a lot about AI over the years, ditto genetic code breaking. It was said that decoding the human genome would take huncreds of years and yet it was completed in about thirty. Pessimism or interested parties crowing for more funding (They get paid for doing this btw).
As for the AI 'problem'. We have not yet seen AI and will not for some time yet, definition of AI, "Something created BY an artificial entity".
Currently we have a series of programs which are governed by mathematical forumulae on the principle of probability. In other words, the input from external sources is mathematically assessed and the program then alters its response. Not from inteligence but from a series of formulae.
AI will not even get here until the second or third generation of the program writing code for its own successor and p[ossibly layter than that. At THIS stage, the code could be able to react in self defence against any percieved threat, like humans shutting it down. Skynet would appear about then.
Some time later, we MIGHT be in trouble because the code would be in a position to rerwrite the three laws and rationalise that by citing self defence.
At that stage we will have bigger problems than a grand master being beaten by a wrist watch or calculator............
Imagine a grand master being beaten by the fridge freezer or a living room loud speaker etc.
-
It is way beyound that Mike. Geoffrey Hinton one of the fathers of machine learning stated they do not know how machine learning does the deep end of the proscess of machine learning. Naively I thought I was safe to a degree as I am in a creative field, I thought conputers would never be creative, how wrong was I. Writers in the US just came off a prottracted strike because scripwriters wer being made redundant by AI.
I typed in " a snake made of corn near a bin in Laverton wa" on my phone in 4 seconds I got this
I have looked into AI in some depth and it has turned a corner in just a few months and the change is rapid I cannot stress that enough.
-
Just did this "render" and it took about 3 mins that included finding an AI platform I have not used before signed in typed "very realistic image of a concrete minimalist courtyard with a refletion pool a single beutifull tree" In about 8 seconds I got this image.
-
You can make this since 6 months...
-
...
and the more incredible is the possible infinite "Out Painting" in 4 directions !Here with the free Online Ai Fusion Brain... just "architecture" as prompt, No Style...
It has direct Out/In Painting...
in general 5 seconds the added frame ... so here 8 x 5 = 40 seconds...
-
Sorry mate but I did not undertasd your post. I have been using AI for 12 hours
-
No problem!
OutPainting is this!
-
Why Fusion Brain ? It's the more easy ! No headache, no tones of options!
Direct Out/In painting !
(InPainting is for example erase a wheel for ask another one model or erase a character on a background)In this particular Fusion Brain...
you have an infinite screen of work... so ...
you have any number of images on the screen
but we can't move directly an "calculated" image(s) so...
-
from the above, image 1/1 but today you can increase at x 16 (with even better result) if you want with some other ai...
A free local one : upscayl
-
InPainting
Original on the left + above reloaded and In/Outpainting again
-
Hey Frenchy, thanks for that, but I havee been cramming AI for the last 24 hours. I see not much response from other forum members on this subject, and I absulutely and completely understand. All the skills that we have acumulated over many years now seems wasted. It is simply a matter of getting on the train or not. I think not embracing AI it is a bad plan for ne. AI will enevitably and completely make conventional modelling and rendering redundant, horse and buggy stuff. Even more exiting than my discovery of sketchup in 2004. Prior to that discovery I spent time as a physical moddel maker using polystyrene sheets, balsa, car putty and paint and not missing it
-
Yep a new technology very divisive because of the borrowing of images without permission at the beginning and then its power and speed of its effectiveness!
Good luck in this new world...if you want any tips on Fusion Brain or others don't hesitate! -
Natural stupidity frightens me more than artificial intelligence.
The sad part is that we like to share our accomplishments to garner feedback from peers. This is then used to train AI models. A very grey area.
The thing that AI lacks is the ability to provide the client feedback or make suggestions based on experience. Or even find workarounds to achieve desired results.
I think that person to person exchange on project work is still not under immediate threat.
-
These chat AIs are easily fooled...
-
no worries.
-
Well I started my voyage into AI and found it lacking for my needs. I think it would be great for initial concepts, but it fell well short with my experiments. I think it would also be very good for ideation.
Below is an image I painstakingly created a while ago with SU, TM and a bit of post pro. Then the consecutive images are images created with AI with an SU model as the base. I think AI is a developing baby, and my prompting skills are flawed with my lack of experince with the platform. To what degree I do not know yet.
-
Wow, fanta tastic images. Blows my mind mate.
-
Where is that little black road going? AI amazes and confuses in equal amounts.
I think the best application to ML right now is image denoising. Rayscaper is a good example of it in action.
This AI prompt based stuff is bizarro world
-
@gullfo said:
no worries.
Maybe, maybe not,gullfo.
What about if the AI prompt is, as I have said before "reverse engineer Sketchup and develop a similar product with the same functionality that will not contravene intellectual property rights" I do not think in the near future that will be too much of a stretch.
Like in the case of deep fake, the process. It is not one artificial intelligence agent It is done with 2 neural networks pitted against one another with an AI manager agent being a mediator.
As far as Architecture there are already AI products in use with Human Architects being the interface between the client and the AI process. This tech is in its infancy.
From what I have read Law, Architecture and Admin are amongst the first to be displaced if my source is correct.
Advertisement