mchusma

7 ms·

91.
by mchusma 3mo ago
Good article. I will only add that I think the problems outlined in the article are exacerbated by the regulatory apparatus. If this were a debate related to truth seeking alone, it would be good. But trials drive banning/unbanning of
92.
by mchusma 3mo ago
I was hoping they would be discussing some path to improving things faster and cheaper. But in this post it looks like they offer quantized version for the same price as full version, and a fast version at much higher cost.
93.
by mchusma 3mo ago
I do think if you are claiming something is not permanent you should basically have to make some specific claims about the timeliness of it (eg you can use it for at least 10 years or your money back).
94.
by mchusma 3mo ago
I really feel that many of the issues with mRNA vaccines and health studies in general are generalizations like “safe and effective”. Everything has statistical risks and benefits, and we should just share those front and center with people
95.
by mchusma 3mo ago
Really if they wanted a standout model that would really take the wind out of GLM's sails, they should have made this the new Haiku, priced at Haiku levels with this performance.
96.
by mchusma 3mo ago
I agree with this assessment, IMO my takeaway from this is "Generally run Sonnet on low, otherwise use Opus". It's kind of like an "extra low" setting of Opus. (depends on the application for sure).
97.
by mchusma 3mo ago
This is much more interesting of a model at $2/$10 (their launch pricing) than at full price. There are many competing models at around this level of performance. I also like that the difference between low, medium, high, xhigh seems m
98.
by mchusma 4mo ago
I don't see this as that different. Anthropic was the first one to get involved in the "AI models must be approved" regime. OpenAI just has the advantage of being second. (To be clear: I do not like this new paradigm)
99.
by mchusma 4mo ago
I've struggled with this. You definitely can have great cheap models. There are many of them open source and served profitably by neo-clouds. The big labs have basically given up on cheap models, and it is frustrating. It means applica
100.
by mchusma 4mo ago
This is also the type of thing that makes space based data centers more viable. I was previously more skeptical on the concept but have come around. I do think ground based centers will have better economics when they can be built though, a
101.
by mchusma 4mo ago
Following up on the Snapchat Specs launch: Its almost like this is a spite product. Like the product is designed specifically to demolish Snapchat Specs. I know its not. Obviously Meta has been involved here a long time, but this is basical
102.
by mchusma 4mo ago
I guess AI images only for me from now on. Why open yourself up for the hassle?
103.
by mchusma 4mo ago
For those not following the debate of X, there have been a surge of doctors who are saying that full body scans are net negative. The argument is that current full body scans often have false positives, and our treatment of false positives
104.
by mchusma 4mo ago
Incredible. I wonder if they can make progress on survivability of regular drowning.
105.
by mchusma 4mo ago
Bravo for this vision. I wish them well and hope they succeed. I look forward to the first real technical reports.
106.
by mchusma 4mo ago
I do love the Frontier carbon removal effort, but I think Anthropic is in a fight with the Trump administration and probably shouldn't be doing things like this right now. I am not sure of Trump's position though, and he could fav
107.
Anthropic: First AI startup in Frontier carbon removal coalition(techcrunch.com)
2 points by mchusma 4mo ago|1 comments
108.
by mchusma 4mo ago
The reason desalination hasn't happened in california is because of political entities like the coastal commission, which blocked a desalination plant that would serve most of LA's water.
109.
by mchusma 4mo ago
You are right. These outcomes also skew heavily towards the easy stuff for LLMs to get. So tickets that take a human 1 min to respond to now cost you $0.99 ($60+/hour) and you are stuck only doing the hard tickets.
110.
by mchusma 4mo ago
I agree with you 100%. Fin and products like this simply do NOT solve the hard part of providing support in 2026. Basically, the hard parts are (1) coming up with the tools for agents to use, like searching for data, making updates, etc. (2
111.
by mchusma 4mo ago
I’ll just say that AI companies need to be pounding the table more about the necessity of AI. The US (and most other countries) have zero idea how to pay for its deficit spending. The only hope is massive GDP based growth and the only idea
112.
by mchusma 4mo ago
I like "Proverb" as smaller than Haiku too, Aphorism is also good. But seriously I want Anthropic to up its small model game. Haiku is not competitive, Deepseek v4 flash outperforms my uses for about $0.10 / $0.20. Whereas Ha
113.
by mchusma 4mo ago
It’s on openrouter. We just noticed performance was worse in a specific agentic app usecase. It’s possible we made an implementation mistake, my main point though is Google is really silly not hosting their own models.
114.
by mchusma 4mo ago
Gemma 4 31b outperformed Gemini 3.1 Flash-Lite in our app benchmarks (agentic tool use via api in our application as a part of various workflows). But google won't let you pay to use Gemma models, you have to go elsewhere, I think this
115.
by mchusma 4mo ago
I think its even more puzzling because you can't even run Gemma 31b on google cloud, they only let you test it with a rate limit. No way (I can find) to actually pay them to use it. We saw great results in our usecase using google dire
116.
by mchusma 5mo ago
If we define product market fit as profitable with a trillion dollar valuation, I think the term has lost its helpfulness. I do agree with the author that these companies seem much stronger financially recently though.
117.
by mchusma 5mo ago
I can’t read the full article, but the snippet (6%gdp/$2T) seems not that expensive? And you could read that either way ”cheap so we should do it” or “if we end up needing to do it, we can do it”.
118.
by mchusma 5mo ago
I am pretty convinced that friendly fraud is about 90% of chargebacks. I have seen some genuine fraud, but dwarfed by friendly fraud over time across 3 companies.
119.
by mchusma 5mo ago
I think it does bear that out in general, although it is slightly more complicated. What seems to happen 1. Low-wage workers, as a collective group, experience an increase in earnings (Dube & Zipperer, 2024). 2. Total job losses do take
120.
by mchusma 5mo ago
There are also home warranties or tech solutions/concierges like tidy.com. The issue historically is that these concierge things are expensive (should be solved by tech/ai) and the warranties create their own class of problems (cl

More ›