fwipsy

8 ms·

151.
by fwipsy 4mo ago
I cracked up when I got to the marathon example. When I ran a half marathon I realized about 80% of the way through that I was on track to finish under 2:30:00 and pushed myself to make it happen. I should have guessed that sort of behavior
152.
by fwipsy 4mo ago
Not hard to be a fast follower. Lots of companies are ~6-9 months behind. Reaching the actual bleeding edge is much harder.
153.
by fwipsy 4mo ago
First impression: Third-party benchmarks or gtfo. Personally, I've never heard of either of these companies before. We're just supposed to take their word that they've matched the best models on the market? Sakana describes t
154.
by fwipsy 4mo ago
I wonder how many people buy a "living room PC" just because they don't know about these fiber optic cables?
155.
by fwipsy 4mo ago
Feels like OpenAI and Anthropic are both more interested in B2B. OpenAI I'm more suspicious of after GPT-4o. I mostly use Claude and I haven't noticed anything that feels like an intentional dark pattern. It's constantly remi
156.
by fwipsy 4mo ago
POV OpenAI, early 2021. You have a pretty good next-token predictor called GPT-3. You noticed sometimes it can do useful things if you write out the start of a task and let it predict the next step. However, sometimes it's very difficu
157.
by fwipsy 4mo ago
Economic growth and short-term job loss are both results of automation. Anthropic seems to have been pretty honest about that to me?
158.
by fwipsy 4mo ago
My understanding is that datacenter water consumption is not really that big of a deal compared to other industries, and it's mostly inflated by people who are looking for a reason to criticize AI/datacenters. If Nvidia is hoping
159.
by fwipsy 4mo ago
Hm! In this context, introducing ID verification may have been a significant silver lining to the order to take down Fable for Anthropic. This also sheds a very different light on people saying that competitive open-source models are underm
160.
by fwipsy 4mo ago
https://www.slate.auto/en they have a size comparison if you scroll down about 3 pages.
161.
by fwipsy 4mo ago
You're right, I was imprecise. Additionally, I wasn't aware that some multiplayer games are free. I'm obviously not a PS5 user.
162.
by fwipsy 4mo ago
Neither of those options is good value. PS5 charges more for games and you have to pay extra to connect to the Internet. The value proposition of PC gaming is openness. You can play on whatever you want, on anything that can run it; you&#x2
163.
by fwipsy 4mo ago
I think this is predicted? Part of the story is how they were able to preserve core reasoning ability while cutting knowledge like "pelicans have wings." > these findings motivate the Parametric Compression-Coverage Hypothesis,
164.
by fwipsy 4mo ago
Every reply to my comment has been complaining about the single line quoting Claude. People don't want a human response either, they just want to complain about LLMs.
165.
by fwipsy 4mo ago
From the article I guess Qontour reproduced the entire text verbatim. > it also includes the entire text of the book, from its opening 800-word foreword to a complete archive of all 311 neologisms... all penned by Koenig. So it doesn&#x2
166.
by fwipsy 4mo ago
And in this case, they didn't work. Perhaps Qontour, as a web-native dev firm, has figured out a blind spot in Google's DMCA takedown process?
167.
by fwipsy 4mo ago
In context, not being able to juggle car engines.
168.
by fwipsy 4mo ago
My half-baked sense is that America is good at making the best things for the highest cost, but tends to be meh at making good or adequate things at a competitive price. Our military seems like the strongest example, but I feel like it also
169.
by fwipsy 4mo ago
Healthy people are wired to respect each other. No, I don't really believe it, I'm not a solipsist. But I do not think it has been logically disproven.
170.
by fwipsy 4mo ago
Yes! As long as everyone else does too.
171.
by fwipsy 4mo ago
Most people at my company default to Opus 4.6. I personally use Sonnet for a lot of stuff.
172.
by fwipsy 4mo ago
It's missing a "nt:" https://www.anthropic.com/institute/recursive-self-improveme...
173.
by fwipsy 4mo ago
"Are AIs conscious?" is not currently answerable because "Are humans conscious?" is not currently answerable. I know I'm conscious because I perceive (it) directly. Everyone else's consciousness basically depen
174.
by fwipsy 4mo ago
Anthropic is valued a lot because they are building AI. If they stop building AI, they are worth zero. Then they're just activists - why don't you ask Yud how many AI companies he's bought out and shut down? Even if they coul
175.
by fwipsy 4mo ago
Anthropic has called for a coordinated pause: https://www.reuters.com/business/anthropic-says-ai-labs-need... I characterize the culture of companies based on who works there. Anthropic is founded by people who left Op
176.
by fwipsy 4mo ago
I'm not taking that stance. It's possible to have a moderate position between "Don't use AI" and "let AI do everything." I am doing the same job, more or less, just a little faster/more efficiently. H
177.
by fwipsy 4mo ago
So do you believe that individual income should be public? Or do you believe that the government should not take income into account for taxation or distribution of benefits?
178.
by fwipsy 4mo ago
They don't stick working for sama, they split off and found Anthropic.
179.
by fwipsy 4mo ago
SVG generation is a good test because it's extremely easy to subjectively assess with visual reasoning where humans are strong. However, pelican on a bike specifically may be overused at this point.
180.
by fwipsy 5mo ago
"People shouldn't need to use AI unless they're overworked, Google is good enough." But why should I work harder than necessary to do the same job? Why wouldn't I want to use the best tools available?

More ›