According to that article, the data they analyzed was API prices from LLM providers, not their actual cost to perform the inference. From that perspective, it's entirely possible to make "the cost of inference" appear to decline by simply subsidizing it more. The authors even hint at the same possibility in the overview:
> Note that while the data insight provides some commentary on what factors drive these price drops, we did not explicitly model these factors. Reduced profit margins may explain some of the drops in price, but we didn’t find clear evidence for this.
What in the world would the profit motive be to “make it appear” that inference cost is declining? Any investors would have access to the real data. End users don’t care. Why would you do the work for an elaborate deception?
Sent an email out just a bit ago! Please let me know if it doesn't arrive for whatever reason.
EDIT: Looks like my comment was detached from its thread (https://news.ycombinator.com/item?id=46857847) without a comment explaining why. It's not breaking any of the stated rules at the top, and it seems like it would be common sense to let someone know they should expect a message (especially considering how temperamental email filtering can be these days). If this _is_ against the rules, I'd appreciate it if future hiring threads stated as such so that people like myself that have been trying to find a job for 7+ months can stop wasting our time here. I understand how it can be challenging to moderate a forum like HN, but the moderation priorities seem pretty disappointing, unfortunately.
> I applied and got rejected without feedback. They stated that there was a high volume of strong applicants (if there were, why are they posting again?)
This is the case for the vast majority of companies posting jobs here and elsewhere (e.g., LinkedIn) right now. I've been looking for the past 7 months after a lay-off and that exact reply is the most common response I've seen (because it's an automated response). Well, most common after not hearing back at all, of course.
> Plus, it seems some of y’all love to hate the very industry which puts a roof over your head. You’re hoping and praying that it all burns down—yet where will that leave you? How do you feel about becoming a plumber—-until the robots take that job?
This probably isn't a line of argument you want to go down. I've been unemployed for 7 months, in part due to how difficult it is to get so much as an intro call because so many people have totally automated the process of spamming every open job posting with as many resumes (many of which were likely LLM-generated as well) as possible.
I didn't get as far as experiencing their billing system; I decided to finally look into Claude recently and discovered that their pricing page's content area was completely blank. It loads on my phone if I'm off WiFi, so I assume my PiHole is blocking something there. That being said, if anyone at Anthropic is reading this, I'm primarily an Android dev but if you'd like to hire me to write a simple static HTML page just like I taught myself back when I was in elementary school, I think it could significantly improve your conversion rates or whatever, and I've been unable to even get an interview anywhere for the past 7 months so I really think we'd be helping each other out.
(Suffice to say I passed on Claude/Claude Code for the time being.)
On Android it more or less just uses the accessibility APIs to grab the actual text, you can do it without using Google Assistant even by selecting text inside an app's thumbnail window from the Recent Apps screen.
> Note that while the data insight provides some commentary on what factors drive these price drops, we did not explicitly model these factors. Reduced profit margins may explain some of the drops in price, but we didn’t find clear evidence for this.