camdenreslink
Joined 564 karma
Software Developer in Buffalo, NY.
Personal Dev Blog: https://code.camdenreslink.com/
LinkedIn: https://www.linkedin.com/in/camden-reslink/
- camdenreslinkI'm not sure I'd feel comfortable making large changes to a complex legacy code base with just an LLM. But I think an LLM could be a big help to try to understand a complex legacy code base. As long as you are verifying what it is saying (which you would naturally kind of have to for that use case).
- That’s not really what I think of when I think of “banner ads”. Clickbait title imo.
- I thought the title was pretty clear.
- I use them every day for coding and Gemini 3 pro, Opus 4.5, and GPT 5.1 (haven’t tried 5.2 yet) are basically identical in terms of ability. Opus 4.5 has a slight edge in my personal experience so far.
- Sometimes you do need to (as a human) break down a complex thing into smaller simple things, and then ask the LLM to do those simple things. I find it still saves some time.
- If 3 or 4 competitors can all provide a mostly identical product, isn't that a commodity? That is essentially the case right now, with the different companies playing around with UI, integrations and business model.
- Surely there is an order of magnitude more training data on plain CSS than tailwind, right?
- My hypothesis is people will get burned out on this unguided learning via LLMs and still want some sort of curated/guided learning experience through material to understand some subject.
There is the problem of "I don't know what I don't know" that a course can solve for you. An LLM can sort of do that, but you have to take its word for it, and it does it pretty much strictly worse at the moment (but is much more flexible).
- It works well, and had first mover advantage. It also is a fork of VSCode, not just an extension/plugin.
- As a personal anecdote, I had a fairly involved application that built up a context with a lot of custom prompting and created a ~1000 word output. I could run my application over and over again to inspect the results. It was fairly reproducible.
I was having really nice results with the o4-mini model with high thinking. A little while after GPT-5 came out I revisited my application and tried to continue. The o4-mini results were unusable, while the GPT-5 results were similar to what I had before. I'm not sure what happened to the model in those ~4-5 months I set it down, but there was real degradation.
- Sounds a lot like LINQ in .NET (which is usually compatible with ORMs actually querying tables).
- If you live in a city with multiple colleges/universities check out the cars in the staff parking lot. Many will have parking passes/stickers for multiple colleges on there. These are adjuncts that have to drive all over town to cobble together a full time job as an instructor.
- That SVG is impressive, but wouldn’t be usable in a real product as-is.
- But if they spend 10x on infrastructure, and capabilities only improve 10%, then that still can be a bubble even if infrastructure is a bottleneck.
- It seems the benchmarks that had a big jump had to do with visual capabilities. I wonder how that will translate to improvements to the workloads LLMs are currently used for (or maybe it will introduce new workloads).
- It was generated by an LLM like everything else these days.
- I think that form of argument is called "whataboutism". Whether flights waste energy or are environmentally unfriendly is really a separate issue. Both things can be bad.
- That is all well and good, but I think it's a fair question in terms of valuation. What is their moat other than momentum?
- They had already won the search engine wars well before any of those additional products existed.
- If OpenAI were to shut down today, would anything in society really change? It seems all valuations are based on future integration into society and our daily lives. I don't think it has really happened yet.