Fable 5 was literally the only model on earth that did not produce slop, OMG! You guys have no idea how good the model is. Its currently banned in less than a week but when it returns i reckon you all use it 
What would take me weeks of manual coding and assited coding with models like Opus fable 5 could one shot clean code full intergrations with 0 slop. It was a complete monster and UI design skills were on another level. This is coming from the model thats said to be able to hack darn near everything.
No need to be hyperbolic Zody. Yes it is a good model. No it's not ‘the only’ model (I use sonnet for most of my coding in fact for token burn purposes) There are VERY capable open weight models capable of good. It really depends on how you groung and tool and inform the model. Is it (Fable) an amazing long context complex coding model that can one shot like a monster? Unconditionally… Yes. It is. 1000%. See my filecab 6 post in the party. It one-shotted it with a few bug fixes after.
Is it the best that will ever exist. Not at all. Is it the ONLY? Nope. Does it still make mistakes? yep. Is it still subject to the skill of the prompting user like it's predecessors. Yes. It makes less ‘common sense’ mistakes.
Is it the end? Not by a long shot. To avoid certain conversations. I'll just say I believe people are over inflating certain things. Most haven't been paying enough attention to other aspects of this conversation about to take the stage in the tech.
This is only chapter 1 in what is essentially the end of un-REGULATED models. Heck it's just the Prologue…
Just let it play out.
4 Likes
People are rightfully concerned that AI will take away human jobs; it won't, because you'll still need people to code the prompts and maintain the data centers. They're also rightfully concerned about the impact that AI will have on data security and copyright; I wonder how much of the data that makes up most of the models we use is actually stuff that people didn't know would end up in such things when they posted it on social media. And, I would rather have the Federal Government in the US regulating it, as much as I don't trust them as having individual US States doing so and creating a patchwork of conflicting rules.
I also wholeheartedly disagree with ya on this one, buddy. It absolutely will. Entire sectors are going to be wiped the heck out.
But
We also don’t employ very many typewriter repairmen in 2026 either. I don’t see mass production of vaccum tubes. Tungsten filament incandescent, mercury vapor lamps… All of it fades. People move on and they take new jobs. Or they fade away. It sounds harsh but it’s true.
The DIFFERENCE this time is the speed. We need to do the same things to prepare people for a generational shift and… MAYBE 1/10th the time to accomplish something in the order of Henry Ford and the Industrial Revolution.
(personal soapbox - you don’t have to agree with me here but…)
Its incumbent upon us to think through the effects and teach people how to cope and/or change. It will take every single teacher on the planet to do it. Yet, in most modern developed countries we’re still arguing about IF we should teach it or ban it.
I got news for ya buddy. That horse left the barn two years ago. It’s teaching time.
Its not doom and gloom but I'm also not going to pretend it’s going to be easy. (not joking on that one) People WILL (not May, WILL) have to adapt and some will have to get outside thier comfort zones. Either we learn how to work With em or bow to em in a few hundred years…
I kinda prefer C3po to Demerzel. Way more stats, way less violence.
As crazy as it may sound, I'd rather have any of the Droids from Star Wars than Skynet, which is what a lot of folks think of when you mention homicidal AI; even HK-48 (or whatever his designation was) from Knights of the Old Republic was straightforward to deal with.
1 Like
Fair point. I was definitely being hyperbolic when I said “the only.” There are other strong models, no argument there.
What caught me off guard with Fable 5 was how little babysitting it needed. Most models still need constant steering, patching logic holes, or cleaning up bloated outputs. Fable felt different. It understood architecture better, held context cleaner, and built with far less handholding.
That’s what impressed me most.
I agree it’s not the endgame and prompting skill still matters. I just think it marked a bigger jump than people realize, especially for complex Home Assistant integrations where context and structure matter more than raw code output.
1 Like
I think AI should replace plenty of human labor where the demand exposes real inefficiencies, bottlenecks, and poor labor conditions. A lot of the systems we treat as untouchable are already failing people and have been for a long time. Preserving inefficiency just for the sake of preserving jobs is not always progress. But that opens a bigger political conversation.
On regulation, I think federal oversight this early is premature. The federal government has a long track record of struggling to regulate fast-moving technology effectively, and AI is evolving at a pace that makes static legislation risky. Poorly written laws could easily create more harm than protection by freezing innovation or consolidating power into the hands of a few major players.
Even companies like Anthropic have learned that engaging too aggressively in the policy arena can backfire, especially under volatile political administrations. Good intentions do not guarantee good outcomes when politics enters the equation.
Right now, targeted accountability around misuse, privacy, and transparency makes far more sense than broad regulatory frameworks.
I bumped on this some time ago, found it an interesting view: We Need To Talk About AI... and we're already seeing what good & bad things LLM's can do even if you question Amazon's meddle in Mythos.
Purely because anyone from the star wars universe has bad aim
2 Likes
History teaches us that any development that increases productivity only leads to more production and consumption, not to fewer jobs. One thing I am concerned about, is that it may further inflate the cost* of jobs that are harder to accelerate in productivity, but still important. I'm talking about things like taking care of the sick or educating children.
*by inflation I don't mean these people will get more money, but that their work will be valued less.
1 Like
IDK about your take as the very fuel for these big tech companies is to replace human labor where they can. I couldn't tell you the last time a human replied to a csr email XD because they are cutting back on customer service only a few humans and a huge ai. things like taxi, food delivery, trucking, airplanes are on radar to be fully controlled by ai as well as finance and medical.
It's been an interesting journey so far for me using AI tools and HA + ESPHome. I initially wired up Cursor AI to HA and had a paid subscription, but now I have Deepseek wired into Cursor AI through a proxy at 1/50th the cost of Opus and Claude and haven't written a line of ESPHome code or built a dashboard manually since. Define an ESPHome skill, create a long lived token and let it go. Beautiful UI's on LVGL screens, custom dashboards, energy configuration, all using simple, straightforward prompts. It even wrote a complex custom component for a chip Im using where nothing existed. I wouldn't have even attempted this in the past.
It's been a somewhat steep learning journey for sure but the more I dig into it, the more impressed I am. The only thing that Opus has excelled at vs Deepseek V4 is schematic design. Opus got it right on the first try. DS never did get it done. I'm sure folks are working on a KiCad skill that will soon close that gap.
1 Like
I don't think anyone seriously believes AI will not replace jobs that exist now. But I do think that just like when factories replaced manual labor in the 19th/early 20th century, or computers and robots in the 20th/early 21st, humankind will come up with new jobs that didn't exist before.
I found it very interesting when I first saw an exhibition about a group of artists and City planners in the early 1900s that was planning lots of ways for people to spend all the free time they would have now that machines took over the work. That never happened and I don't think it will this time, unless people will decide they don't actually need more than they have now.
We have similar experiences, and i been trying to learn advanced circuitry to manufacture my own smart home gadgets been letting ai help me but it sucks there as your saying.
Im sure the Fed are thinking of how to turn us into a communist country as we speak to offset allot of whats about to happen.
1 Like