ewoodrich

7 ms·

31.
by ewoodrich 9mo ago
The sneaky move that I hate most is when Claude (and does seem to mostly be a Claude-ism I haven’t encountered on GPT Codex or GLM) is when dealing with an external data source (API, locally polling hardware, etc) as a “helpful” fallback on
32.
by ewoodrich 9mo ago
> prompts where we'd ask for e.g. 15 sections, would only do 10 sections and then ask "Would you like me to continue?". I can’t speak to any kind of identifiable pattern but man that behavior drives me up the wall when it
33.
by ewoodrich 9mo ago
That seems a bit dramatic. What I learned from all this is that OpenAI is willing to offer a service compatible with my preferred workflow/method of billing and Anthropic clearly is not. That's fine but disappointing, I'm kee
34.
by ewoodrich 9mo ago
I recently mentioned in another comment that Fedora 43 on my Ideapad is the first “just works” experience I’ve had with my multi monitor setup(s) on anything other than Windows 11 (including MacOS where I needed to pay for Better Display to
35.
by ewoodrich 9mo ago
The first time I’ve had my multi-monitor setup(s) “just work” on Linux is recently installing Fedora 43 on my Ideapad. (After becoming exhausted trying to tweak Linux Mint to get tolerable sizing across all the screens). Wayland per-monitor
36.
by ewoodrich 9mo ago
FYI, nearly all of that UI/app garbage can be removed (or re-enabled like the Start/context menu) in <5 minutes with: https://github.com/Raphire/Win11Debloat It persists across updates, can be customized w
37.
by ewoodrich 9mo ago
Ugh, well at least this was the nudge I needed to cancel my Claude Pro subscription... I've already had a bad taste in my mouth watching the rate limits on the plan get worse and worse since I first subscribed and I have a few other su
38.
by ewoodrich 9mo ago
Exactly, if a comment just feels a little off but you're unsure, do a quick scan of the profile, takes 15-30 seconds at most to get sufficient signal. If it's actually AI, the pattern becomes extremely obvious reading them bac
39.
Alphabet's market cap surpasses Apple's for first time since 2019(cnbc.com)
2 points by ewoodrich 9mo ago|0 comments
40.
by ewoodrich 9mo ago
I’m well aware, as I said I am regularly using CC/Codex/OC in a variety of projects, and I certainly didn’t claim that can’t be used productively in a large code base. But that different challenges become apparent that aren’t addr
41.
by ewoodrich 9mo ago
Fully agreed! That’s the exact kind of thing I was hoping to find when I read the article title, but unfortunately it was really just another “normal AI agent experience” I’ve seen (and built) many examples of before.
42.
by ewoodrich 9mo ago
That happy path works okay for a while but provides very little margin for when the battery inevitably starts to degrade. I’m a few years in and now every few days it’s started to die at around 8pm (yet claims the battery health is still j
43.
by ewoodrich 9mo ago
I literally only use Siri on my Apple Watch, I’ve only triggered it accidentally on my iPhone and have the hot word disabled on all my other devices. Of course, all I ever use it for is setting timers and alarms on the watch, but still…
44.
by ewoodrich 9mo ago
Right, I am a daily user of agentic LLM tools and have this exact problem in one large project that has complex business logic externally dictated by real world requirements out of my control, and let's say, variable quality of legac
45.
by ewoodrich 9mo ago
And for the Elder Scrolls games: https://en.uesp.net/wiki/Main_Page
46.
by ewoodrich 9mo ago
That may keep them preoccupied for a while. But eventually they'll try to upload their post-relativity recursive quantum intelligence unification theory magnum opus to arXiv as a neatly LaTeX-formatted paper so they can spam university
47.
by ewoodrich 9mo ago
Yes, I use it with a regular Claude Pro subscription. It also supports using GitHub Copilot subscriptions as a backend.
48.
by ewoodrich 9mo ago
Exactly. There was a stretch of 6 months or so right after ChatGPT was released where approximately 50% of front page posts at any given time were related to LLMs. And these days every other Show HN is some kind of agentic dev tool and Anth
49.
by ewoodrich 10mo ago
https://xcancel.com/karpathy/status/2004607146781278521 (FYI: you can easily bypass the awful logged out view by replacing x.com with xcancel.com, I use a URL Autoredirector rule to do it automatically in Chromium
50.
by ewoodrich 10mo ago
That's a dishonest framing of their argument. There's nothing logically inconsistent in believing wide adoption of AI tools causes developers' skills to atrophy and that the tools also fail to deliver on the hype/promise
51.
by ewoodrich 10mo ago
As an Android user since the T-Mobile G1, I tried switching to an iPhone 15 Pro as my primary phone from a carrier deal as the hardware looked nice plus Android/iOS has converged so much over the years with all the same apps available
52.
by ewoodrich 10mo ago
So does the family of four driving the speed limit that gets killed in a head-on collision.
53.
by ewoodrich 10mo ago
Hah, I was listening to a similar conversation that began with family members working in the school system complaining about AI slop that began (relatively) harmlessly in day-to-day email conversations padded with time wasting filler but no
54.
by ewoodrich 10mo ago
I've been able to do WiFi-WiFi connection sharing on my last four Samsung phones as well
55.
by ewoodrich 10mo ago
I bookmarked to take a closer look later, but I'm a little unclear on the premise, could you explain what you mean by "slow"/how it is filtered/curated?
56.
by ewoodrich 10mo ago
I mean, we typically architect systems depending on humans around an assumption of human fallibility. But when it comes to automation , randomly still doing the exact opposite even if somewhat rare is problematic and limits where and at wh
57.
by ewoodrich 10mo ago
I've gone down that route already with Roo/Kilo Code and then OpenCode, but OpenCode with the z.ai backend and/or the CC z.ai Anthropic compatible endpoint although I've been moving to OC in general more and more over ti
58.
by ewoodrich 10mo ago
It's a perfectly serviceable fallback when Claude Code kicks me off in the middle of an edit on the Pro plan (which happens constantly to me now) and I just want to finish tweaking some CSS styles or whatever to wrap up. If you have a
59.
by ewoodrich 10mo ago
Not OP, and haven't had it flat out rm the entire .git, but I have had Claude get flustered and pull a "Wait, no! what was I thinking? that idea doesn't work at all here, I need to revert that attempt and try something else
60.
by ewoodrich 10mo ago
I recently ran into two baffling, what felt like GPT 3.5 era completely backwards misinterpretations of an unambiguous sentence once each in Codex and CC/Sonnet a few days apart in completely different scenarios (both very early in the

More ›