dudeinhawaii

7 ms·

31.
by dudeinhawaii 7mo ago
I don't see what you're seeing, in any dimension. But here's a fair take. I wrote several very specialized benchmarks that I've used over time, that surface "model personalities" and their effects on decision m
32.
by dudeinhawaii 7mo ago
I don't see why we can't have AI powered reviews as a verification of truth and trust score modifier. Let me explain. 1. You layout policy stating that all code, especially AI code has to be written to a high quality level and hav
33.
by dudeinhawaii 8mo ago
This is marketing. The same way Apple cares about your privacy so long as they can wall you in their garden. Not a value judgment, just saying that the CEO of a company making a statement isn't worth anything. See Googles "don&#x2
34.
by dudeinhawaii 8mo ago
It usually refers to situations without access to the source code. I've always taken "clean room" to be the kind of manufacturing clean room (sealed/etc). You're given a device and told "make our version".
35.
by dudeinhawaii 8mo ago
True, but I think the implication (as I read it) is that AI may be providing more complex solutions than were needed for the problem and perhaps more complex than a human engineer would have provided.
36.
by dudeinhawaii 8mo ago
Somehow this article explains perfectly, visually, how AI generated code differs from human generated code as well. You see the exact same patterns. AI uses more code to accomplish the same thing, less efficiently. I'm not even an AI h
37.
by dudeinhawaii 8mo ago
After 2 days of giving it a go, I find that Gemini CLI is still considerably worse than both Codex and Claude Code. The model itself also has strange behaviors that seem like it gets randomly replaced with Gemini-3-Flash or something else.
38.
by dudeinhawaii 8mo ago
My take has been... Gemini 3.1 (and Gemini 3) are a lot smarter than Claude Opus 4.6 But... Gemini 3 series are both mediocre at best in agentic coding. Single shot question(s) about a code problem vs "build this feature autonomously&q
39.
by dudeinhawaii 8mo ago
My experience is that on large codebases that get tricky problems, you eventually get an answer quicker if you can send _all_ the context to a relevant large model to crunch on it for a long period of time. Last night I was happily coding a
40.
by dudeinhawaii 8mo ago
I just ran some numbers and it works out if you're a prolific user. Over 9 days I would have spent roughly $63 dollars on Codex with 11.5M input tokens plus 141M cached input tokens and 1.3M output tokens. That roughly mirrors the $100
41.
by dudeinhawaii 8mo ago
I think you've stated this in reverse. API limits are infinite but you'd blow through $20 of usage in a maybe 1 hours or less of intense Opus use. The subscription at $20/mo (or $200) allows for vastly more queries than $20 w
42.
by dudeinhawaii 8mo ago
This is something that I was thinking about today. We're at the point where anyone can vibe code a product that "appears" to work. There's going to be a glut of garbage. It used to be that getting to that point required
43.
by dudeinhawaii 8mo ago
The other element here is that the vibecoder hasn't done the interesting thing, they've pulled other people's interesting things. Let's see, how to say this less inflamatory.. (just did this) I sit here in a hotel and I
44.
by dudeinhawaii 8mo ago
If you visualize it as AI Agents throwing a rope to wrangle a problem, and then visualize a dozen of these agents throwing their ropes around a room, and at each other -- very quickly you'll also visualize the mess of code that a colle
45.
by dudeinhawaii 8mo ago
Right but, do you or the founder have actual responses to the story posted? It seemed to give RentAhuman the benefit of the doubt every step of the way. The site doesn't work as advertised, appears to be begging for hype, got a reporte
46.
by dudeinhawaii 8mo ago
So the exploiters have deprecated that version of spyware and moved on I see. This has been the case every other time. The state actors realize that there's too many fingers in the pie (every other nation has caught on), the exploit is
47.
by dudeinhawaii 8mo ago
I don't know how to say this but either you haven't written any complex code or your definition of complex and impossible is not the same as mine, or you are "ai hyper booster clickbaiting" (your words). It strains belie
48.
by dudeinhawaii 8mo ago
I'm not making a judgement on the product but why is everything about selling shovels? I want to see "real" software and new innovative ideas that are developed and maintained using AI and not just shovels 'advertised&#x
49.
by dudeinhawaii 9mo ago
I think for many/most programmers = 'speed + output' and webdev == "great coding". Not throwing shade anyone's way. I actually do prefer Claude for webdev (even if it does cringe things like generate custom CSS
50.
by dudeinhawaii 9mo ago
When I think of the effect of a single word on Agent behavior - I wonder if a 'compiler' for the human prompt isn't something that would benefit the engineer. I've had comical instances where asking an agent to "per
51.
by dudeinhawaii 9mo ago
> Nah I disagree. A charter plane with 200 Dutch tourists is lower risk than a flight coming out of Bolivia. What a weird and random thing to say. There's literally no data that can support for or against and neither have a history
52.
by dudeinhawaii 9mo ago
Curious, are you doing the same planning with Codex out-of-band or otherwise? In order to have the same measurable outcome you'd need to perhaps use Codex in a plan state (there's experimental settings - not recommended) or other
53.
by dudeinhawaii 9mo ago
I think you're underestimating the features used within Office. Offices isn't bloated because they wanted to add fluff. It's bloated because of the large number of customers that have differing but overlapping needs. The Engi
54.
by dudeinhawaii 9mo ago
I'm very PRO AI and a daily user but let's be real, where is the productivity gain? It's at the edges, it's in greenfield, we've seen a handful of "things written with AI" be launched successfully and the
55.
by dudeinhawaii 9mo ago
If a fan starts a $10/mo Patreon membership inside the iOS app, Apple's subscription terms imply $3/mo goes to Apple for the first year (then $1.50/mo after), and Patreon's platform fee still applies on top. Patreon
56.
by dudeinhawaii 9mo ago
To play devils advocate, why do we have to layout a simple task in PAINSTAKING DETAIL to an AI model which is "PHD LEVEL" and going to take our jobs in 6-12 months? Why am I still holding its hand like it has the intellect and exp
57.
by dudeinhawaii 9mo ago
I've noticed a degradation in Opus 4.5, also with Gemini-3-Pro. For me, it was a sudden rapid decline in adherence to specs in Claude Code. On an internal benchmark we developed, Gemini-3-Pro also dramatically declined. Going from bein
58.
by dudeinhawaii 9mo ago
On the one hand, I created vibe coded a large-ish (100k LOC) C#, Python, Powershell project over the holidays. The whole thing was more than I could ever complete on my own in the 5 days it took to vibe code using three agents. I wrote coun
59.
by dudeinhawaii 9mo ago
Funny, the last interview I watched with Karpathy he highlighted the way the AI/LLM was unable to think in a way that aligned with his codebase. He described vibe-coding a transition from Python to Rust but specifically called out that
60.
by dudeinhawaii 9mo ago
After reading the article (and watching the video), I think the author makes very clear points that comments here are skipping over. The opener is 100% true. Our current approach with AI code is "draft a design in 15mins" and have

More ›