3,186 karma · joined February 6, 2011
From where I'm sitting, it's just turning python functions into boxes and instead of write the function yourself, you drag from the output of one box to the input of another. For 2 or 3 boxes, this is cool, but I opened up a professional workflow and was taken into a view with 100s of boxes and wires all over the place. Uhh, ok?
For myself, I'd rather just create my own python environment, write some quick pytorch or mlx calls, wire up some cli to it and share that in a GitHub.
Rust is what you use when you want to maximize on the runtime performance. It will build software that requires significantly less memory and maximize the CPU that will save time and energy. It's a system language for writing software meant to access the hardware.
It was only in the last 10-20k years or so where farming emerged.
I don't know, but I think it's incredible how little time has passed relatively compared to our long history, of us being domesticated.
Apple and Meta are the only two big companies really pushing the frontier of this space, and everyone else is trying to figure out what to do with the technology. World Labs had the best marketing demos and showed the world what you could do with it, but it's all been vaporware from where I've been sitting.
So, if you're pooping out code, and committing it because tests still pass, and that's all you know, you're in for a treat. When an executive wants to know why a b0rked feature lost their department millions of dollars, guess who will have to answer for it, and its not the LLM.
My advice is to find ways to keep on top of how it all works, and if you're the only one who cares, well, then, that makes you even more valuable, not less.
Community flyers have been notoriously cringe for a very long time. In the 1990s they were all Comic Sans with cringe Microsoft Word clipart. Then, the design world decided that Comic Sans was the bar for bad taste.
Even if these examples look lazy, someone still had to do the work to prompt the thing into existence and they, at the very least, found it acceptable. So, either way, it had the same amount of effort it always had.
Instead of quickly iterating to find how to take voice to the next level, both projects quickly locked in on whatever that was and became too costly to pivot.
If I take an analogy of two kinds of people who visit Yosemite National Park, person A would refuse to see Glacier Point unless they embark on the long all day hike themselves, while person B would happily drive to the parking lot, get out, take photos with their families and admire the view.
Either person cannot understand the other's, and yet both people will go out of their way to make the other one feel guilty for their approach if pushed.
When they built the parking lot and the roads to get there, it probably upset a lot of people at the time (only speculating). Yet, the park invests and maintains the hiking trails just as well. So if anything, the park takes the very human opinion that you should take the option that works for you, because the view is worth it.
The people who are investing in machine learning are OBSESSED with efficiency. Where I would say, v1 of Co-pilot was entirely conceived as an AI system meant to help ME as a software engineer, claude code is entirely conceived and built to replace ME as a software engineer.
Where co-pilot was a choice that I made, and willingly enabled in my workflow, claude code was forced on me by my employer. In my entire long career, we were always given autonomy to decide what tools and systems we needed to get our jobs done. This was the first time it felt like a few invisible power brokers at the top decided for me.
Anyway, my point in all this ranting, is that, unlike national parks that go out of their way to be accessible for everyone, generative AI has firmly sided with one kind of person's life philosophy at the expense of the others, yet we all still have to work together. Change is hard for sure.
I agree though. My issue is the cost for using AI video models is way too high for anyone not building anything serious with them, at the same time they are too restricted for actually using professionally. Prompting them with text to get something generated is cute, but then you just end up creating slop that everyone hates, ultimately devaluing the power of these things.
I definitely remember a moment of endless leaked dick pic stories that came out back in 2010 or 2011.
As a millennial looking outward, gen-z seems to have embraced a full attachment to technology in a very open way. They almost act like their value as a human being is determined by the visibility of their online presence.
I know that Google just launched a go version a couple months ago, but I haven't really spent any time with it, because well, gemini isn't really useful to me right now.
The thing I constantly ran into with using non-platform (ie. claude code or codex) agents is that even though their tool call API seems flexible, they've finetuned these "agentic" models around their specific agents. Opus is the worst I've seen about this. It really wants to use specific bash tools and if you restrict or hide it, it'll start writing python code to call into bash to do it.
If you were to clone yourself, and that clone could maintain 100% of whatever state your brain is in so that all your experiences and memories were intact. From that point forward, as far as we know, you and your clone would diverge. Then, from my tiny-brained vantage point, the fact you remain your original self, your clone is technically a new person. You can't see through your clone's eyes, or process any internal information from your clone. However, something new powers that clone, and the only word I can think of for that is "soul." I don't say that in religious context, but unless there's a more precise word that can detangle it from sounding mystical, for now I'm happy to use it, since people mostly understand what I'm talking about.
Once we start talking about resetting this clone at-will, then, with consciousness in the mix, you will have to confront the ethics of that clone being enslaved by whatever. All very messy and I hope we don't have to answer this question in my lifetime.
Post-production is very much still in California though.
This is entirely because of deepfakes.
Apparently Seedance has a version without guardrails if you are a licensed production company. I suspect Google Omni would too, but I don't know anyone who uses that model.
Post pandemic, in my little bubble at least, I almost never see people drinking beer anymore.
I see two forces working against this that proprietary models will always have over an open source model.
1. The biggest is content licensing. Content is quickly becoming gated by systems at the front of their load balancers, completely changing the social contract of the Internet. What used to be a quick google search for recent facts that lead me to places like reddit or twitter, is now completely walled off if you're not physically at your browser and using an IP address from a last-mile provider.
LLMs have pre-trained on the bulk of the information up to 2024/2025, but over time that will be more and more out of date.
Anthropic, OpenAI and Google will all have to pay for access to a lot of this content refresh going forward, and it does make a material difference in the output you get.
2. Liability is the other. A corporation can look at a contract for model access and see one that provides uptime guarentees, content infringement promises and model safety, and pick the contract that shields the corporation from the most liability. A 3rd party hosting platform like fireworks.ai that hosts open weights models won't provide any of that at all. They will simply bill you for time spent on their hardware and make promises that they won't log or inspect corporate traffic.
1. an artifact of agent tools only doing generic greps based on keyword searches, limiting the scope to a few lines 2. RLHF from Anthropic and OpenAI that bake in catastrophic edit avoidance
I used to have a CLAUDE.md rule that forbid inline comments and only allowed public API doc comments. That used to work perfectly back in 2025 era models. Now these models refuse to honor it and write worse comment slop.
This has become the "voice" of Opus 4.8 and I just roll my eyes when someone puts up a PR that includes more comments than code changes telling a story about all the random things it found during exploration.
These comments do cost us real dollars to endure, both from the model generation cost and the human cost of having to read them. I was hopeful Fable would have been better, but aye, it's worse and writes the same comment slop in multiple places now.
Still, I don't agree. I think this machine is meant to use local models. You just have to wear pants if you want to keep it directly on your lap. I rarely use it that way anyway. I prefer it plugged into an external display and comfortably sitting on a laptop stand.
1: right now, it's basically 90% beautiful hyper-realistic woman in some scene.
That is SO incredibly boring. These models are able to generate all kinds of never before seen mixed together content. Yet, everyone just wants to generate photos of things you could have already searched for.
2: creative people should be the biggest and loudest voices celebrating it but the vision researchers building the tools have only been interested in solving the math side of it and haven't bothered to ask creatives what they even wanted. It's fine in private R&D, at the hypothesis stage, but then these corporate heads went and shipped as finished products and charge us serious money for it. So, creatives rightly feel uneasy about the models that legitimately "stole" their content without permission, while also minting new millionaires at these companies by selling it back to them.
Once companies license the content and artists can make royalties, and the tools to interface with the models become sophisticated so that you're not one-shot prompting into existence a random thing, more creative people will come back. (i.e. ComfyUI)
---
Plus, these models need actual fully staffed product teams behind them to solve and own the ethical issues around things like character identity locking that can be used to deceive. Or being too easy to say "in the style of" that could almost perfectly replicate another artist's entire voice. Then, there is the nudity issue. I don't want generative models to have morals, especially when it's art. Art is meant to be provocative and boundary pushing. I realize that's an entire deep-hole of shit in there waiting to get through, but that's why we need people spending their work day thinking about it more than just blocking anything that shows too much skin.
There are at least 4 platforms they would need to support: Win, Mac, iPhone, and Android.
That's 4 different software engineers at least, just for the frontend.
Then, there's various backend engineers, who could be shared, yes, but not always. Android's weird runtime requirements are bespoke enough that just because the database is written in C++, doesn't mean it's the same C++ database as what the Windows backend would use.
Finally, there's the designers, who end up consolidating all the unique things about each native platform into a common design language so they can have a shared vision on all of the platforms. So engineers end up building UI that works identically on all 4 platforms, and you're basically building a bespoke "browser" at that point.
The entire western world had been shifting towards neoliberalism as a direct response to the eastern world shifting towards communism since WW2.
Trump also isn't the embodiment of anything other than the guy who didn't take it seriously and suddenly ended up with the job because the voters in the country decided it couldn't be any worse under him than whatever the current situation they were living with was.