thank you so much @RedGrimnir i am language hopping for like 1-2 months still understand zero now ill focus on c++
Epic doesn’t care about any of us. They have gotten so big that they don’t need us any more. The Unreal Editor will no longer be a tool made for artists and game developers. It will be the slop generator of the future. It will be trained and fed on our work and our time until it has become capable of generating its own slop content without human interaction. This is what they want. To take the human part out of development and bury us under the AI trash heap.
If you want an immediate example of this, then look no further than the state that they have allowed the Fab store to become. I have advocated for the complete removal of stolen products and AI slop from Fab since day 1. They make empty promises but have yet to implement a single meaningful solution to fix anything. I am convinced that it will be the same when they remove blueprints and replace them with their AI “code”.
In order to make the removal of blueprints compatible they will also need to remove or replace everything in the editor that is dependant on blueprints, including the end users.
It’s Epic’s choice to push us away. Either we continue to allow ourselves to be used, or we take a stand and fight for our rights to develop what we want and how we want. Even if that means abandoning their engine and not looking back.
UE6 can’t come fast enough.
That’s it? No reason or explanation?
I have tried looking up this magical alternative some of you keep talking about and all I can find is a new way to train their AI. I don’t want to have to type my queries into a nonvisual builder and I do not consent to having my development used to train AI.
So unless you can point me to a comprehensive article explaining how their new system will be better, then I don’t see how anything good can come of their changes when there hasn’t been adequate information provided.
I just want better Mass integration and a production ready UAF. Everything else is just optional..
An animation framework would be cool but I really hope that they don’t just push bipedal humans without consideration for other rigging and animation models. They have been known to offer seemingly magical solutions like Nanite but the compatibility is another story and so is the pipeline to build optimized scenes with it.
I really don’t see why UAF would suddenly impose any limitation.
With the removal of blueprints it makes me skeptical on how they plan on integrating movement and animation systems done entirely with the control rig and sequencer.
They use so many random phrases and key words to describe this new process without actually giving any details about it. Including the word “Nativization”. Like what is that supposed to mean in this context? That Epic will magically have everyone using a brand new production language built entirely from their AI without the need for a human to build the actual blueprint and parameters will set themselves up automatically?
Also I’ve seen the verse language and holy batman what a major step backwards in building a visual interface for developers. It’s the opposite of visual and has a garbage looking syntax that looks dated and redundant.
there will be portions of the editor that will remain in visual scripting (similar to blueprints)
animations, PCG, Behavior Tree (for the positive and negative that data multiplication ends up being), Materials, and the positioning of objects relative to each other within an object.
the problem isn’t so much that my stuff will be used to train AI, its that even the best LLM-AI systems would still after almost 7 years of active distribution would fail 3rd grade math, and have reading competency of a 5th grader.
anyone that is seeing massive improvements in time and design by using LLM-AI is in 99% of cases working on something that already has a public solution to copy (and the problems accessioned with just copying someone elses solution when it might not actually fit your needs, or be correct/efficient), or solvable by brute force (and a Junior devs yearly salary spent in 1 week)
Nah. Chatgpt and anthropic have been fantastic with very advanced math for quite a while now. Nearly all computer sci problems are a rehash of a previously solved problem. LLMs let you explore many solutions quickly and choose the best for the situation. I use GitHub copilot daily with various LLMs and never go over $20 in tokens a month.
As humans we consider mathematics to be additive (in order to understand Calculus you need to understand algebra, to understand algebra you would need to understand arithmetic, to understand arithmetic you need to understand numbers), but that is on the conceptual level. the LLMs can “do” very advanced math because the harder mathematics are hard fast rules, and formulas, and then algebra, but most algebraic situations are proofs.
when I say “it would fail 3rd grade math” isn’t that it can’t do the arithmetic, but rather it fails to extract out numerical details and place them into equations with units successfully (the questions of given a birth date and a current date determine how old the person is, or simply counting letters in a word where some words had to be hard coded after ridicule)
I say “5th grade reading competency” because it lacks consistency in responses even within a short timespan to the same exact prompts (with cleared context), and as prompt size increases deviation between responses increases exponentially with some results giving unfounded results of interpretation. Almost like it is still based more on Random then logic.
for your success rate I could easily just point to my 3rd paragraph: many things in game systems design is not outright “hard”, it is balancing the interactions, and the costs that is the “hard part”
making an inventory, or a enhancement system is not the hard part, but rather ensuring that both systems know of each other as much as needed, and having those systems interoperate, and not be “too big” to slow down performance, or take up too much memory.
I am not impressed when an LLM spits out 500 LoC for a feature, and then needs 5 agent passes to make it actually function because its own mistakes.
I am impressed by the developer who can take that output, and reduce it down to 200 LoC, be more performant, and more correct to the problem state without the LLM needing to explain every line.
I don’t think it has to be even math or physics.
Define a gameplay problem that doesn’t have a solution (or a public visible solution) and watch it embarrass itself with popcorn.
I started with Blueprints back around 2013-2014. BP taught me how to structure systems and overall architecture, which made my transition to C++ much smoother at least conceptually/though process. I still use BP alongside C++ every day. As a visual learner, it’s been an incredible tool to have, and I will genuinely miss it if it goes. ![]()
I am almost certain that the first and most popular plugin for UE6 will be a reincarnation of Blueprints (to some extent) ![]()
hahha, the only correct answer! xD, but yeah might hold some value in verse or what they call the new stuff. But still BP is BP ![]()
I find EPIC’s approach as ridiculous as claiming—regardless of the context—that composition is always superior to inheritance; in the development of standard libraries, inheritance has always held the dominant position.
Programming is hard…
Think again on Ultra high
Unity is fantastic though.

