601–610 of 728 posts
Be careful, this model is worse than 03-25 in 10 of the 12 benchmarks (!) I bet they kept training on coding, made everything worse on the way, and tried to hide it under the rug because of the sunk costs.
where do you see that?
Old model card : https://storage.googleapis.com/model-cards/documents/gemini-...
They intentionally buried that information
My frustration with using these models for programming in the past has largely been around their tendency to hallucinate APIs that simply don't exist. The Gemini 2.5 models, both pro and flash, seem significantly less susceptible to this than any other model I've tried. There are still significant limitations, no amount of prompting will get current models to approach abstraction and architecture the way a person doe…
> no amount of prompting will get current models to approach abstraction and architecture the way a person does I find this sentiment increasingly worrisome. It's entirely clear that every last human will be beaten on code design in the upcoming years (I am not going to argue if it's 1 or 5 years away, who cares?) I wished people would just stop holding on to what amounts to nothing, and think and talk more about wha…
I don't know this sentiment would be considered worrisome. The situation itself seems more worrisome. If people do end up being beaten on code design next year, there's not much that could be done anyways. If LLMs reach such capability, the automation tools will be developed and if effective, they'll be deployed en masse.
If the situation you've described comes, pondering the miraculousness of the new world brought by AI would be a pretty fruitless endeavor for the average developer (besides startup founders perhaps). It would be much better to focus on achieving job security and accumulating savings for any layoff.
Quite frankly, I have a feeling that deglobalisation, disrupted supply chains, climate change, aging demographics, global conflict, mass migration, etc. will leave a much larger print on this new world than any advance in AI will.
Earlier quoted context omitted.
At that point why not just have LLMs generate bytecode in one shot? Plenty of training data to go on, I'd imagine.
The code would be un reviewable.
Earlier quoted context omitted.
[flagged]
[flagged]
Earlier quoted context omitted.
It seems like the fix is straightforward (check the output against a machine readable spec before providing it to the user), but perhaps I am a rube. This is no different than me clicking through a search result to the underlying page to verify the veracity of the search result surfaced.
Why coding agents et al don't make use of the AST through LSP is a question I've been asking myself since the first release of GitHub copilot. I assume that it's trickier than it seems as it hasn't happened yet.
While we have millions of LOCs to train models on, we don’t have that for ASTs. Also, except for LISP and some macro supporting languages, the AST is not usually stable at all (it’s an internal implementation detail). It’s also way too sparse because you need a pile of tokens for even simple operations. The Scala AST for 1 + 2 for example probably looks like this,
Apply(Select(scala, Select(math, Select(Int, Select(+)))), New(Literal(1)), Seq(This, New(Literal(2))) etc etc
which is way more tokens than 1 + 2. You could possibly use a token per AST operation but then you can’t train on human language anymore and you need a new LLM per PL, and you can’t solve problem X in language Y based on a solution from language Z.
My frustration with using these models for programming in the past has largely been around their tendency to hallucinate APIs that simply don't exist. The Gemini 2.5 models, both pro and flash, seem significantly less susceptible to this than any other model I've tried. There are still significant limitations, no amount of prompting will get current models to approach abstraction and architecture the way a person doe…
> no amount of prompting will get current models to approach abstraction and architecture the way a person does I find this sentiment increasingly worrisome. It's entirely clear that every last human will be beaten on code design in the upcoming years (I am not going to argue if it's 1 or 5 years away, who cares?) I wished people would just stop holding on to what amounts to nothing, and think and talk more about wha…
My frustration with using these models for programming in the past has largely been around their tendency to hallucinate APIs that simply don't exist. The Gemini 2.5 models, both pro and flash, seem significantly less susceptible to this than any other model I've tried. There are still significant limitations, no amount of prompting will get current models to approach abstraction and architecture the way a person doe…
> no amount of prompting will get current models to approach abstraction and architecture the way a person does I find this sentiment increasingly worrisome. It's entirely clear that every last human will be beaten on code design in the upcoming years (I am not going to argue if it's 1 or 5 years away, who cares?) I wished people would just stop holding on to what amounts to nothing, and think and talk more about wha…
Can't really prepare for that unless you switch to a different career... Ideally, with manual labor. As automation might be still too expensive :P
Earlier quoted context omitted.
> It's entirely clear that every last human will be beaten on code design in the upcoming years Citation needed. In fact, I think this pretty clearly hits the "extraordinary claims require extraordinary evidence" bar.
I would argue that what LLMs are capable of doing right now is already pretty extraordinary, and would fulfil your extraordinary evidence request. To turn it on its head - given the rather astonishing success of the recent LLM training approaches, what evidence do you have that these models are going to plateau short of your own abilities?
Theoretical limitations of multi-layer Transformer https://arxiv.org/abs/2412.02975