Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
padolsey
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
6 ms
·
61.
▲
by
padolsey
11mo ago
> What would the IoCs even be? Prompts.
62.
▲
by
padolsey
11mo ago
> PoC || GTFO I agree so much with this. And am so sick of AI labs, who genuinely do have access to some really great engineers, putting stuff out that just doesn't pass the smell test. GPT-5's system card was pathetic. Big-tal
63.
▲
by
padolsey
11mo ago
We've done this! https://weval.org/analysis/visual__pelican/f141a8500de7f37f/...
64.
▲
by
padolsey
11mo ago
Yeh I still don't think there's a fixed definition of what a world model is or in what modality it will emerge. I'm unconvinced it will emerge as a satisfying 3d game-like first-person walkthrough.
65.
▲
by
padolsey
11mo ago
>What happens when you prompt one of these kind of models with de_dust? Presumably de_dust2
66.
▲
by
padolsey
11mo ago
I am looking forward to this being available on taobao/temu for around $5 !
67.
▲
by
padolsey
11mo ago
To this day I hate captchas. Back when it was genuinely helping to improve OCR for old books, I loved that in the same way I loved folding@home, but now I just see these widgets as a fundamentally exclusionary and ableist blocker. People wi
68.
▲
by
padolsey
11mo ago
Some randoms I remember : ) - Thinking in Systems - by Donella Meadows - Dead Aid - Dambisa Moyo - Day of Empire - Amy Chua - Mistakes Were Made (not by me) - Caroll Tavris - Brotopia - Emily Chang
69.
▲
by
padolsey
11mo ago
Ideas matter, yes. And distribution. If you want good coverage, read books from a more diverse crowd. If you want narrower ideas, do the opposite.
70.
▲
by
padolsey
11mo ago
Of the 25 authors listed there are: 18 americans. 24 people in the global north. 24 men. EDIT: people reaalllllyy hate when this is pointed out. See child comment for some alt recommendations.
71.
▲
by
padolsey
11mo ago
I dunno. A competent human can hold the a mental image and work through it. Not too hard with experience. What I generally mean tho is: I don't think we can state the supreme capabilities of AI (which people love to do with grate fervo
72.
▲
by
padolsey
11mo ago
I help run an eval platform and thought it fun to try a bunch of models on this challenge [1]. There's some fun little ones in there. I've not idea what Llama 405B is doing. Qwen 30B A3B is the only one that cutely starts on the l
73.
▲
by
padolsey
11mo ago
I mean, almost anything would be better. But here's my swing: > I'm so sorry about this. We definitely screwed up here and want to > fix things. We want to chat to you in a call if you're able? > We will stop
74.
▲
by
padolsey
1y ago
I agree, and this reminds me: I really wish there was better URL (and DNS) literacy amongst the mainstream 'digitally literate'. It would help reduce risk of phishing attacks, allow people to observe and control state meaningful t
75.
▲
by
padolsey
1y ago
For something a bit simpler I made redoku a while ago: https://padolsey.github.io/redoku/
76.
▲
by
padolsey
1y ago
Thanks for sharing! I feel the fear of another attack with epilepsy too. It is terrifying. The doom and the walking on thin ice constantly hoping you're not gonna over-step or do the wrong thing. And all that at the same time as trying
77.
▲
by
padolsey
1y ago
> when I got dragged into a project that was sufficiently interesting that I started overworking This is what bites. I have some really narrow interest areas that I can end up being obsessive about, to my own detriment. We have to be car
78.
▲
Tips for stroke-surviving software engineers
(blog.j11y.io)
505 points
by
padolsey
1y ago
|
189 comments
79.
▲
by
padolsey
1y ago
What concerns me most in the era of gen AI irt photography is journalism. We need truth, most especially when limited-means citizen journalism is the only reliable source of that truth. But I feel like the only way to accomplish fool-proof
80.
▲
by
padolsey
1y ago
I wish they'd release some data or evaluation methodology alongside such claims. It just seems like empty words otherwise. If they did 'extensive safety testing' and don't release material, I'm gonna say with 90% ce
81.
▲
by
padolsey
1y ago
I mean, HN is no stranger to cults of personality. Post a paulg essay that says something a teenager could write and it'll fly up to position 1.
82.
▲
by
padolsey
1y ago
I worked on Maps when the re-brand to Ovi was happening. I remember piles of nokia models for us to test it on. Supporting that diversity of devices was kind of a tall order and pretty unprecedented. It was a different era.
83.
▲
by
padolsey
1y ago
Beautiful images in some ways, but so raw and stomach-churning in others. I am not American but feel sick to this day at the thoughts of these events. On the day of 9/11, I was 11 and went to my IT class that afternoon in Sussex, UK. O
84.
▲
by
padolsey
1y ago
OOC did you ask them with or without 'web search' enabled?
85.
▲
by
padolsey
1y ago
There is a famous case from a few years ago where a laywer using ChatGPT accidentally referenced a fictitious case of Varghese v. China Southern Airlines Co. [0] This is completely hallucinated case that never occurred, yet seemingly every
86.
▲
by
padolsey
1y ago
> Do any career software engineers here see more than a 10% boost to their coding productivity with LLMs? I know it'll be touted as rhetoric but I have seen an order of magnitude of difference in my ability to ship things. Thankfull
87.
▲
by
padolsey
1y ago
Interesting. Curious to reproduce across models, I made a comprehensive eval based on your post and ran it against 30 models, each tasked with recalling specific data from 500 rows in different tabular formats. Have a look at the results he
88.
▲
by
padolsey
1y ago
Thanks for your work on this! It's a very legit domain of problem for LLMs to optimize for. I've produced a comprehensive eval based on your post and run it against 30 models, each tasked with recalling specific data from 500 rows
89.
▲
by
padolsey
1y ago
Because it's fun, I'd like to pose the contrary position that AI will actually make us more different. Perhaps dangerously so. Many people don't understand the nature of LLMs nor how rabbit-hole-y a long context will necessar
90.
▲
Sorry, We Deprecated Your Friend
(blog.j11y.io)
3 points
by
padolsey
1y ago
|
2 comments
More ›