Marker

hot take: ai having ads is the dumbest thing because they can be ran locally. unless openai finds a way to make chatgpt addictive like netflix/streaming did with content, self-hosting/competitors will almost always be better. AI models will be like browsers in the future.

@ReviewsPossum this is the lefty version of keeping politics out of sports isn't it

The atheists of Reddit are mad that the Artemis II astronauts mentioned God.

@reddit_lies Is this the liberal version of keep politics out of sports

Atheists are furious today

@Life_truthway ig oriental orthodox doesn't exist

Family Tree of Christianity. Comment if you are there in this family tree.

@BarryWhite84113 @JCSMITH420 @GuntherEagleman I don't understand this, have you been to ohio or penn or the dakotas? Yeah maybe the north east is like this but the Midwest pulls a lot more weight and guns in this scenario. This is like saying stereotypes about dc for 2.

@JCSMITH420 @GuntherEagleman 3 can't even figure out which bathroom to choose just saying lol

2 and it wouldn’t be close.

@TheDankBoi69 @AtheistTakes what. are you saying cultural traditions from religion aren't valid? Wtf is this response, you got a historical answer explaining cultural traditions and you're like "fuck culture and history you're wrong"

@MarkerDaSharker @AtheistTakes Well, whatever this is, it got out of hand It was a rhetorical question, of course it doesn't have to do anything with it

Easter is Christ's

@TheDankBoi69 @AtheistTakes eggs were used to break the fast as they weren't allowed to be eaten during lent, medieval Europe thought hares laid eggs, so fun childrens tales were told during Easter the hares would hide eggs for children to find to break their fast.

@AtheistTakes What do hidden eggs have to do with the resurrection of Christ

@esjesjesj they've done one an official one for diwali btw

Google doesn’t do religious holidays dumbass

@Atomixion2 @AtheistTakes that's... the point of the account? it's a meme account this happens on both sides to laugh at dumb responses. not every conversation needs to be a theological debate. big "sir this is a Wendy's" moment over here.

@AtheistTakes I see you still have nothing but low IQ strawmen snd no answers.

If God real why Earth big

@DrHassdenteufel @CattardSlim They still have the 3/5ths compromise

@CattardSlim And that she doesn’t understand math. Ten voters is incapable of producing 53%.

Top Trump advisor, Laura Loomer, confirms that CPAC is a total bust.

@DesignSpecter @TerribleMaps Pulling someone over going 79 in the left most lane of a 4 lane highway with the flow of the rest of traffic and staying there for 30+ minutes not switching lanes isn't keeping y'all safe and is def playing around

@MarkerDaSharker @TerribleMaps Our Officer's don't play around. And the the state highway patrol.... they are definitely in a league of their own.

Places you’re most likely to get a speeding ticket in

@VnSafeguard @kmcnam1 the point they're making is hes gaming and not learning or working

@kmcnam1 That's only 2k max. Your dad is poor for other reasons.

@creatrix_ttv when I clicked this I didn't expect a gold standard truther of all things

Blaming women working for macro outcomes is the laziest possible model of the economy. Here’s why 98% of this is misleading at best and absolute bullshit at worst. The premise leans on numbers that aren’t even true. It was not “only 5% of mothers working prior to the 1970s.” By 1970, the labor force participation rate for married women (husband present) with children ages 6–17 was 49.2%. If you want to tell a “something spooky happened in the 1970s” story, monetary policy is a much stronger candidate than women getting jobs. In 1971 the US severed the dollar’s convertibility to gold. From that point forward, we moved fully into a discretionary fiat monetary regime where money and credit could expand without constraint. Inflation isn’t caused by “too many workers." Sustained price inflation comes from monetary expansion outpacing real output. Adding workers increases the economy’s capacity to produce goods and services, it’s literally the thing that should make stuff cheaper over time if markets are competitive and money is stable. In competitive markets, higher output can actually LOWER prices and raise living standards. If goods get cheaper, real wages improve even if nominal wages don’t rise. So if you’re claiming women entering the workforce “must” have crushed living standards, you’re implicitly claiming the economy got less capable at producing... which is backwards. The reason the “more workers should have lowered prices” dynamic didn’t show up is because the supply-side gain from more labor is operating inside a system where money is elastic and intervention distorted key sectors. If anything, women entering the workforce should have been disinflationary in a sound money, high-competition environment. The fact that the biggest pain points since the 1970s have been asset-heavy, subsidy-heavy, regulation-heavy sectors with restricted supply and weak competition (housing, healthcare, tuition) should point you toward monetary and regulatory distortions, not toward women entering the workforce.

@SidKhurana3607 i feel like the washington bubble makes me wanna say hindi or some other indian lanauge, but nevada is throwing me off. is it hindi?

Guess the language?

i dont think that graph says what you think it does. thats saying for benchmarks we've trained better models over the years, not that the price/compute itself is decreasing. the price of inference is realitively the same, theres just better models. do you think its really fair to compare gpt-4, which is 1.8 trillion parameters, to claude haiku, which is 20 billion parameters directly? theres your 90x difference (though parameters are not all equal, as different experts can be shut off in MoE models, which is partly contributing to the 9x now that wont continue unless we find a better architectural paradigm). my previous post was saying that when we get closer to the saturation in information in the model, hardware costs and optimizations matter more, which are way harder and take more time to design, and face high initial costs (especially with ram prices getting higher). flops per dollar according to your sources, is only improving by 1.3x per year, not 9x. the inital 9x is only because of them not really knowing how to price it yet, and ramping up scale. were closer to scale, we have better techniques for scaling laws, and even more consistent intuition of what design works better on hardware (sparse MoEs/distilling, which, if we look at price per token per parameter, is captured by deepseek released almost 2 years ago now). why do you think that 9x trend is going to continue when there are neural scaling laws that say otherwise?

@MarkerDaSharker @valigo @jpnyunja The literal most conservative assumptions possible still put us at 9x decrease per year

Do I understand it right that when VCs stop subsidizing tokens you'll have to be a 10x dev just to break even on costs?

@suchnerve I am not a bot but my previous account was hacked.

You can safely assume that if an account: 1. Was created in 2023 or later 2. Has under 100 followers 3. Is saying something needlessly rude It’s just a bot. 🤷🏻‍♀️ ChatGPT launched near the end of 2022, and it’s easy to use it to troll, but AI text slop rarely attracts followers.

@compliantvc C teir ragebait, too wrong of numbers to be taken as actual bait, and not as funnily ironic to be an interesting read. take a lesson from @eurofounder if you arent the same account owner.

My friend got a job offer in the United States It would be a 5% pay increase "Will you take it?" I asked Nope! He explained how he'd actually be losing money by moving to the US: - $2k to lease a car (required in the US) - $1,500 a year tipping waiters - $37k extra for health care All of those expenses are zero in Europe Moving to the US just doesn't make sense anymore

@OSherikar @Amank1412 @OpenAI @Google uhh, Google acquired DeepMind in 2014 and internally started Google Brain in 2011. OpenAI was started in 2015, past both of these landmarks.

@Amank1412 True, @OpenAI started it's AI journey long back than @Google so yeah @OpenAI will definately win

If Google competes with OpenAI purely on AI, I’m sure OpenAI would win in the AI domain. But Google has a complete ecosystem, which gives it a strong advantage.

@Amank1412 google was the one that started this with "attention is all you need" and they have more data, and have had consistent research output in deepmind. openai has a product that went viral yet cant monetize it. $BYND shows how this plays out for poorly ran enthusiast brands.