Not a lot of people here work for Meta, which is why you had to lump in "Random SaaS" like that's remotely comparable. I doubt most people here are working on anything harmful, let alone a fraction of what Meta does.
Unless you think Todo list apps cause Ethnic Cleansing.
You say this right when your central banks went on a massive money printing spree, shot up inflation beyond their own baseline for years, and have created a frankenstenian K-shaped economy where everyone who is not in the top 10-20% is getting actively poorer
It doesn't take a lot to figure out that hard currencies with some supply cap tend to hold value better. BTC might not be it - it might be gold or silver - but you really can't ignore the inflation issues inherent in fiat currencies anymore.
Deflationary "currencies" like crypto are significantly worse (why spend when you can HODL). Regular crypto "currencies" will have the exact same effects.
You touch upon an important point: the entire foundation of our current economic system is built on high velocity of money. You are incentivized to spend rather than hold because without it, there's no mindless conspicuous consumption. A deflationary currency will grind this system to a halt
But the question you need to ask is: has all this mindless consumption really done us any real good? For the handful of genuinely useful things to come out of excessive consumption, there are a gazillion pieces of wasteful crap that clogs homes first, then landfills
A deflationary currency is bad for the current economic system. But that doesn't mean it's bad by itself.
Lol. All you have to do is think about five seconds without going into rants about over consumption.
Let's take Bitcoin. It has a predefined limited supply. That is, it already has an artificial deficit built-in. On top of that, obtaining bitcoin becomes exponentially more expensive over time (until it becomes impossible once all the bitcoin has been "mined"). Now you have to spend many magnitudes more resources to obtain even a tiny fraction of a bitcoin.
So even in the absence of any external world to compare to (e.g. without denominating Bitcoin in USD) why would anyone spend a limited resource they obtained? Especially considering that even today getting even a single bitcoin through traditional mining is nearly impossible except for the filthy rich miners running industrial-scale data centers.
So you are saying there would be a limited resource becoming exponentialy more expensive over time? I guess there might occur a demand for such a resource? And mayhaps in that case the holders might decide to exchange it for some other assets they desire?
(ofc thats not the 100% correct description of bitcoin which depreciated against most other assets ytd but the idea still stands)
> So you are saying there would be a limited resource becoming exponentialy more expensive over time?
I'm not saying it. It is literally what Bitcoin is, and other deflationary "currencies".
> And mayhaps in that case the holders might decide to exchange it for some other assets they desire?
Keyword is "some".
In 2010 when bitcoin was novelty and had no value at all, someone paid 10 000 (yes, that's ten thousand) bitcoins for two pizzas. Now imagine bitcoin becomes a currency, as some cryptobros still prophesize. You have a constant pool of money for all "assets", and no way to get more money.
There are massive P2P economies already built in most third world nations that work on crypto. They prefer stablecoins but will happily accept BTC as well. It's a serious enough problem that they've started using code words to refer to crypto currencies ("goat" for USDT, "chicken" for ETH, etc.)
If a corrupt third world government is scared of a piece of tech, it is, imo, good tech
the good news is that I'm personally on my last few years online. I don't think there's anything really worthwhile in this space to do as a contributor or even as a consumer
My brother who is nearly 50 and has worked in tech since the dot com boom, got laid off in January and couldn't find a job until last week. This job, too, was just a contractual position at his old Fortune 50 firm.
He has an engineering degree from one of the top 5 engineering colleges in India, a Master's from one of the top 5 engineering schools in the US. He built some of the systems that form the foundation of the entire call center industry.
And now he pivoted to GenAI and has dozens of very impressive public projects including some heavily starred open source repos
The dead comment replying to this said "Start your own company," which I hear a lot. The problem is, the #1 thing YC looks for in a founder is they're wealthy enough to absorb risk. You don't take on massive risk from a position of financial insecurity.
I think the margin isn't that large to be honest. If we compare available resources and data it is quite tiny and perhaps should be larger.
Also it doesn't feel solved to me at all. There is no general model, perhaps it cannot reasonably exist. I think these tests are benchmarks are smart, but they don't show the whole picture.
Domain specific image generation tasks still require a domain specific models. For art purposes SD1.5 with specialized and finely tuned checkpoints will still provide the best results by far. It is also limited, but I think it dampened the hype for new image generators significantly.
Does SD1.5 suffer from resolution / coherence / complexity issues?
I understand most outputs could be fine tuned for most domains, but still felt sd1.5 had a resolution ceiling, and a complexity ceiling no matter how good the fine tuning
Yeah SD 1.5 is mostly trained on datasets of resolution of 512x512. That's why you'd get crazy multi-limb goro abominations if you pushed checkpoints too much higher than 768x768 without either using a Hires Fix or Img2Img.
There's not much of a reason to use SD 1.5 over SDXL if image quality is paramount.
A lot of people (myself included) use a pipeline that involves using Flux to get the basic action / image correct, then SDXL as a refiner and finally a decent NMKD-based upscaler.
Yes, the toolchains around it can alleviate it, but only to a degree. You more or less dependent on a fine tune specifically trained for the things you want. But if you have that, the image quality is usually far better than from any generic model in my opinion, aside from resolution.
Merging any or all concepts is mostly beyond it, but I haven't seen any model being good at it yet. There are some that are significantly better, but often come with other disadvantages.
Overall what these models can do is quite impressive. But if you want a really high quality image, finding the fitting model is as difficult as finding the right prompt. And the general models tend to always fall back to some mean AI standard image.
Prompt understanding will only ever be as good as the language embeddings that are fed into the model’s input. Google’s hardware can host massive models that will never be run on your desktop GPU. By contrast, Flux and its kin have to make do with relatively tiny LLMs (Qwen Image uses a 7B-param LLM).
I asked Nano Banana Pro to generate a selfie that looks realistic (in terms of skin, lighting etc.). I feel the irises still give it away, but apart from that... https://imgur.com/a/hPcMybi
They always had the best talent, but with Brin at the helm, they also have someone with the organizational heft to drive them towards a single goal
reply