Love your work, love those who love you..live an adventure! Entrepreneurial. GovTech. Software, ideas, strategy, investing. AI Driven Development is the future.
@davepl1968 I would actually guess the number is way higher than 30%. Unless you were a school kid in most of the DC metro area that day because it was a snow day, so you would have had to have been watching at home.
We've had around 10% inflation over that period of time. Presumably there was some population growth too.
And that's really the problem, Amendment 3 took the wrong approach - it didn't deal with spending. It should of been focused on keeping local spending limited to a factor based on inflation & population growth.
Never forget, Anthropic wants you to perceive high value from the plan. You are looking at retail API pricing which is a different level of guaranteed performance and, obviously, doesn't reflect the actual inference cost to Ant.
Also, you didn't output 16.3b tokens...15.9b were cache reads which, on Opus 5.5, have a retail cost of $.20 but the actual cost to Ant is way way less. Overall, looking at your actual token usage and assuming it's all on Opus 5.5, we are talking about $7k of lifetime usage (not $100k).
My guess is that these plans work financially because they can limit performance when they need the GPUs for API paying customers and they need the conversations to train/tune/improve models and harnesses.
That depends on 1) do you think the GOP has a viable shot w/ a better candidate and 2) will the Dems go center-left moderate or hard left?
If you think any GOP candidate will struggle in 2028, maybe you want to save you good candidates, let Vance get trounced, and put a stake in Trump/MAGAism ahead of 2030 midterms and 2032.
I wouldn't advocate this, but I can see some political people thinking this way.
I agree with Demis that anything you understand about biology can be explained computationally. The part I'm doubtful about (and would love to be proven wrong) is that you can go from computation to biology.
The model I use for science is that biology is explained by chemistry -> chemistry is explained by physics -> physics is explained by math.
But discovery is not the same an explaining. And it is much easier - although not easy - to explain math than it is biology. In fact, all math solutions are explanations but in biology we discover stuff all the time that we can't explain.
So I don't think biology will be solved like math. The more likely scenario is that AI helps us take biology discoveries and explain them and that those explanations then unlock interesting research paths that lead to more discoveries.
That said, I don't expect that biology discoveries - especially the kind that materially change disease outcomes - will look and feel anything like how AI as "solved" math.
The problem here is that Kennedy is looking for an actual understanding of Originalist thinking. Scalia wrote about this. More recently Barrett has wrote about this. This is a well known challenge to originalism (ear cutting, whipping, branding vs 8th amendment.)
The expectation is that someone seeking a Federal judgeship saying they are an originalist has spent enough time thinking about this to understand different approaches/arguments and be able to provide a coherent explanation of how they would approach such a problem.
The fact that stare decisis means he wouldn’t have to make a decision here, plain reading of precedent makes that clear, makes it a good question to understand his thinking.
And he just utterly failed.
You can’t get a 30% price drop without a lot of forced selling. And there isn’t a big enough piece of the market - at least that I can see in the data - to get significant forced selling.
You basically need a GFC type crisis, but that requires very specific conditions in multiple markets that don’t exist today.
@krassenstein This story isn’t actually about religion - and anyone with a good understanding of history would know this. This story is about changing global demographics and what values are likely to have sway over the next century.
This only works if you remove the requirement for insurers to spend a certain percent of premiums on healthcare. That rule, part of the ACA, makes it so the insurers are disincentivized to negotiate hard on rates b/c their profit is essentially a fixed percentage of premiums. Higher premiums = more profit $ but reducing healthcare pricing = less profit $.
The thing becomes that so many things can be reduced to programming tasks. They may not feel to the average person doing them that they are programmable tasks but frequently they are. So, in a lot of cases, if you can describe what you want done a model will develop a one off script and complete the task.
I watch how Claude got really good at PowerPoint. It didn’t get better at editing a slide like a human would through the UI, it got really good at writing python scripts it can run to generate the PowerPoint deck.
Idk, if you don't use the models for coding or data-oriented tasks, you might think this. And I don't think super-skilled coding = super-intelligence. But if you do, it's obvious the models are pretty fucking smart and rapidly improving.
I would argue trust in any modern classically liberal society (aka the west) the progressives will always look like the winners, when conservatives win we stop more progressive policies (as Buckley described it “standing on a wall yelling stop.”) So a conservative win is to not change and a progressive win is to change, and only one is visible. What really matters is are the changes slow and moderate enough to avoid societal damage - that’s what conservatives ultimately need to measure our work against.
@33MolonLabe@chrislhayes What’s the difference? Whatever they cut is someone’s pain. That’s always the reality of reducing spending. You just have to be honest about the reality.
@Jonbarros84@Nads_Shariff Why would you setup your incentive pay structure so most don’t get a bonus? Sounds like you are running a loser of a business.