gordonhart

6 ms·

61.
by gordonhart 6mo ago
I'm not as optimistic as you are that AI automating only high-value employment paths is a good thing. It swings the power balance even further towards capital and away from labor.
62.
by gordonhart 6mo ago
Sure, but I would likewise appreciate if you made the assumption that my question wasn't snark and that I actually wanted to learn the answer.
63.
by gordonhart 6mo ago
I see, thanks. Strange times for internet censorship in the west.
64.
by gordonhart 6mo ago
Why post an archive link for a static site with no ads or subscriptionware?
65.
by gordonhart 6mo ago
The big difference is the number of 9s you get from LLM APIs versus your CPA office computers. Building a business around something with 0-2 nines is a nightmare.
66.
by gordonhart 6mo ago
I won't use it, but I'm excited to see it for the same reason why I'm excited to see a near-frontier open-source release: more competition pushes prices down and reduces monopoly/cartel risk. I won't use Muse or Gro
67.
by gordonhart 6mo ago
A new model comparable (ish) to the Claude/Gemini/GPT flagships is a big deal for the industry and for Meta even if it doesn't set the new frontier.
68.
by gordonhart 6mo ago
The biggest bill I've gotten from Bunny was like $10 when my app ( https://atlasof.space ) briefly went viral and got 100k+ views in a month. Bunny CDN is so reasonably priced and the realistic visitor ceiling for my projects
69.
by gordonhart 6mo ago
For personal projects, I'd rather just pay $2/month and not think about it than get hit with a random bill and scramble to migrate before the next month's bill. Bunny is perfect for this use case where you have a handful of p
70.
by gordonhart 7mo ago
They're primarily a Twitter phenomenon and get circulated quite widely within the tech sphere there.
71.
by gordonhart 7mo ago
The individual task scores are all on public tasks, they still held out a hundred or so private tasks that presumably GPT-5.4 did well on to get its leaderboard position.
72.
by gordonhart 7mo ago
My broad vibe is that Gemini 3.1 Pro is the best at visual/spatial tasks and oneshotting while Opus 4.6 is the best at path planning. This task leans heavily on both but maybe a little more towards planning so I'm not too shocked
73.
by gordonhart 7mo ago
The main frontier models are all up on https://arcprize.org/tasks Barely any of them break 0% on any of the demo tasks, with Claude Opus 4.6 coming out on top with a few <3% scores, Gemini 3.1 Pro getting two nonzero sco
74.
by gordonhart 7mo ago
The point is still to test frontier models at the limit of their capabilities, regardless of how it's branded. If we're still capable of doing so in 2057 I'll upvote the ARC-AGI-26 launch post!
75.
by gordonhart 7mo ago
At least according to the Head of Product at X, Sora was by far the most widely used tool to create fake war videos[0] aiming to push various false narratives. Given how popular fake content is at Meta I can only imagine what they see there
76.
by gordonhart 7mo ago
Not so different from the raise seed -> ICO -> dump bags pipeline that thrived in the early 2020s cryptoverse.
77.
by gordonhart 8mo ago
This comment was almost certainly written by an LLM.
78.
by gordonhart 8mo ago
Easy to forget but there was a ton of industry+investor excitement around computer vision from ~2015-2021, to the extent that the "MLops" niche sprung up around it. This was called AI at the time, and mostly went out the window wh
79.
by gordonhart 8mo ago
We’re reaching a saturation threshold where older models are good enough for many tasks, certainly at 100x faster inference speeds. Llama3.1 8B might be a little too old to be directly useful for e.g. coding but it certainly gets the gears
80.
by gordonhart 8mo ago
Yes, proof that you are a human. With so much AI slop on the internet you have to be defensive about your attention to avoid wasting it on low-effort LLM outputs.
81.
by gordonhart 8mo ago
Audiobooks?
82.
by gordonhart 8mo ago
> Due to our concerns about malicious applications of the technology, we are not releasing the trained model. As an experiment in responsible disclosure, we are instead releasing a much smaller model for researchers to experiment with, a
83.
by gordonhart 8mo ago
Remember when GPT-2 was “too dangerous to release” in 2019? That could have still been the state in 2026 if they didn’t YOLO it and ship ChatGPT to kick off this whole race.
84.
by gordonhart 8mo ago
I'm also looking for "proof of pulse" somewhere and can't find it on the linked site, on Amazon, in his HN bio, or with a Google search. Unfortunately a necessity in the AI era.
85.
by gordonhart 8mo ago
It's also more than a little misleading to compare to the 2022 peak. Anybody who was hiring software engineers in 2020-2022 or being hired as one knows that was a wild and unsustainable period.
86.
by gordonhart 8mo ago
Fair, I was hoping there was a feature that I was missing. Minor papercut to have to include harness-specific files/symlinks in your repo but it's probably a temporary state until the tools and usage patterns are more settled.
87.
by gordonhart 8mo ago
> This is supported out of the box by any agent worth its salt, including OpenCode and CC. I thought Claude Code didn't support AGENTS.md? At least according to this open issue[0], it's still unsupported and has to be symlinked
88.
by gordonhart 8mo ago
Exactly, it's the same documentation any contributor would need, just actually up-to-date and pared down to the essentials because it's "tested" continuously. If I were starting out on a new codebase, AGENTS.md is the fi
89.
by gordonhart 8mo ago
Kind of looks like vibecoding is doing to SaaS what Chinese mass manufacturing did to physical products two decades ago. Only the marketing and distribution matter in a world where it's very easy for others to clone something and sell
90.
by gordonhart 8mo ago
Every day the software world feels more and more like a casino.

More ›