Viewing profile — joegibbs
joegibbs
HN member- Joined
- Tue, Mar 10, 2020, 9:49 PM UTC
- HN karma
- 1,851
- Public activity
- 671 items
- HN profile
- View on Hacker News ↗
About joegibbs
https://jgibbs.dev joe@jgibbs.dev
Recent public activity
-
comment
Comment #49205755
“The noun verbs.” is such an incredibly Claudeish sentence. The other models don’t do that stuff. I’d say it’s actually biggest tell for Claude writing.
-
comment
Comment #49130095
More people are probably using stuff like Claude Code, Codex etc from the CLI or desktop apps, meaning they’re actually using more AI and skipping Zed altogether
-
comment
Comment #49107530
They were trained on a lot of garbage human-written slop originally, they have to post-train the models so the responses are good. Luckily if you’re an AI company you probably have…
-
comment
Comment #49092526
With hundreds of millions of people, what are the odds that more die from scurvy than the outbreak?
-
comment
Comment #49032506
It does, but I don’t think the Ukrainians will get much flak for it. Bombing consumer goods distributors doesn’t worry people as much as bombing food distributors or chemical warfa…
-
comment
Comment #49017811
I don’t think the slippery slope is a fallacy. If a group makes a change in one direction is it not more likely that they will continue making changes in that direction than if the…
-
comment
Comment #49015759
GPT-Image has that horrible deep-fried look to it. Can't draw anything without looking like it's got grit thrown over the whole image. Nano Banana is much, much better here, it's o…
-
comment
Comment #49004100
That legalistic "I can do whatever I want as long as it's not breaking the law" argumnet lasts until the first time they ban wrongfully ban someone for it and get huge negative PR
-
comment
Comment #48987698
I like the moving cars. They're copies of the stopped ones right? Adds a good effect.
-
comment
Comment #48986510
I think that's literally about 3-4 requests. Fable is ridiculous expensive on API pricing.
-
comment
Comment #48986490
Doesn't it say something if 5% of a group are wacko lunatics if, for instance, a random sample of the rest of the population has only 0.5% wackos?
-
comment
Comment #48986428
The problem is the whole idea of working from first principles to try to discover the actual truth, rather than just going along with other ideas that other people have created thr…
-
comment
Comment #48967441
What about an armadillo playing a piano? There are so many potential combinations It would say something if the pelican looked great but the armadillo looked terrible
-
comment
Comment #48942357
What does it say if you ask it in Chinese? I imagine English training data would be more critical of China than Chinese
-
comment
Comment #48930778
But who are you to decide whether suburban Americans are living the right kind of life? Surely if they wanted to they could move to an apartment in a dense inner-city area. It woul…
-
comment
Comment #48841085
A timer tool with a callback feature would take like a couple of hours to implement, a model which has a native internal ability to know the time would take ages to make
-
comment
Comment #48810932
Why would the rows be gigabyte size? You’d just save the text.
-
comment
Comment #48800156
But there must be a ton of generic questions that people ask. Stuff like "What's the capital of country X?" - it's probably at least 10% of queries. Memories, custom instructions e…
-
comment
Comment #48683100
The coat of arms of the city has been using SPQR since the medieval period
-
comment
Comment #48681897
Interesting how Anthropic is popularly considered the left-wing AI company and OpenAI the right-wing one, but ChatGPT is much more left-wing while Claude is centrist
-
comment
Comment #48669228
What would be the point? It would need to be six double-side Blurays to fit the actual game on there. You could have one but you're still going to have to download the majority of …
-
comment
Comment #48628892
He made a few mistakes but hardly any more than previous PMs. Burnham doesn't seem to have any grand new plans, it's just business as usual. Three months and his approval is going …
-
story
Show HN: AdvertBench, ranking the ability of LLMs to create image ads
Experiment that I've made. The models get access to an E2B sandbox and are instructed to create an ad according to the specifications (they can choose whatever tools they want to u…
-
comment
Comment #48594528
As an Australian, why are European train prices so high? Obviously it's due to a lack of subsidies, but why are they not subsidised? For instance, a train from London to Edinburgh …
-
comment
Comment #48577964
AI is growing much faster than the other components of MS and Alphabet's business, and OAI is 100% dedicated to AI while the other two only have small portions on AI