wtetzner

8 ms·

151.
by wtetzner 8mo ago
> The compiler metaphor is simply incorrect If an LLM was analogous to a compiler, then we would be committing prompts to source control, not the output of the LLM (the "machine code").
152.
by wtetzner 8mo ago
What I don't understand about these arguments is that the input to the LLMs is natural language, which is inherently ambiguous. At which point, what does it even mean for an LLM to be reliable? And if you start feeding an unambiguous,
153.
by wtetzner 8mo ago
And at the point that your detailed specification language is deterministic, why do you need AI in the middle?
154.
by wtetzner 8mo ago
I don't think the argument is that AI isn't useful. I think the argument is that it is qualitatively different from a compiler.
155.
by wtetzner 8mo ago
You don't ever make the prototype public. Also, there's a certain point where you can't avoid management sabotaging things.
156.
by wtetzner 8mo ago
> A lot of software people like to jump in and I see them portray the planning people as trying to figure everything out first. My approach, especially for a project with a lot of unknowns, is usually to jump in right away and try to bui
157.
by wtetzner 8mo ago
Apple News is a paid subscription. Facebook and Google are not. Apple is supposedly the premium brand that provides a curated experience (isn't that their reasoning behind the closed nature of the App Store?).
158.
by wtetzner 8mo ago
I feel like I'm still not understanding something. How does making the output from the LLM lower level help?
159.
by wtetzner 8mo ago
Am I understanding what you're saying correctly? * We need a deterministic input language * The LLM generates machine code Isn't that just a compiler? Why do we need the LLM at that point?
160.
by wtetzner 8mo ago
I buy 3-in-1 shampoo, conditioner and body wash.
161.
by wtetzner 8mo ago
The analogy to assembly is wrong. Even in a high level language, you can read the code and reason about what it does. What's the equivalent for an LLM? The string of prompts that non-deterministically generates code? Also, if LLM outpu
162.
by wtetzner 8mo ago
This has largely been my experience. Just reading and understanding the code, and writing the change myself ends up actually being faster.
163.
by wtetzner 8mo ago
Is reviewing outputs really more efficient than writing the code? Especially if it's a code base you haven't written code in?
164.
by wtetzner 8mo ago
Maybe. But I don't have trauma that causes me to tie my self worth to physical exercise. Yet when I can get myself to actually do it, I feel better. I'm not sure deep thinking is that different.
165.
by wtetzner 8mo ago
Just because a problem takes more than a few hours doesn't mean you'll never get past it. And those are the problems that really grow your abilities.
166.
by wtetzner 8mo ago
> We don't really need more software Yeah, we need better software.
167.
by wtetzner 8mo ago
On the other hand, those situations can be valuable learning experiences, and (even if subconscious) can help you learn to see things from more angles in the future. Unfortunately, predicting which situations are worth it and which aren&#x2
168.
by wtetzner 8mo ago
> If I can get a solution that is “close enough” in a fraction of the time and effort, it is irrational not to take the AI route. I think this is context dependent. Over time, within a single code base or ecosystem, incompatibilities bet
169.
by wtetzner 8mo ago
Another way in which their different is that, because they are non-deterministic, we check in the output, not the input. It's the equivalent of checking machine code into our source control instead of the input language. It's not
170.
by wtetzner 8mo ago
And because of that, we check in the generated code, not the high-level abstraction. So to understand your program, you have to read the output, not the input.
171.
by wtetzner 8mo ago
> you're probably not going to learn a ton with clay pot #10,001 Why not just use a library at that point? We already have support for abstractions in programming.
172.
by wtetzner 8mo ago
I don't think this comparison holds up. With a higher-level language, the material you're building with is a formal description of the software, which can be fed back into a compiler to get a deterministic outcome. With an LLM, yo
173.
by wtetzner 8mo ago
> plugging them together like LEGOs Aren't Legos known for their ability to enable creativity and endless possibilities? It doesn't feel that different from the clay analogy, except a bit coarser grained.
174.
by wtetzner 8mo ago
If you're only building things that have been built before, then sure, though I'd argue we already had solutions for that before LLMs.
175.
by wtetzner 8mo ago
> It's the finished product, not the process, that matters. I think the point is that the finished product depends on the process.
176.
by wtetzner 8mo ago
Yeah, it's why I don't like trying to write up a comprehensive design before coding in the first place. You don't know what you've gotten wrong until the rubber meets the road. I try to get a prototype/v1 of whateve
177.
by wtetzner 8mo ago
> You can now do things you couldn't even have contemplated before. You've been handed a near-godlike power, and all you can do is complain about it? This seems to be a common narrative, but TBH I don't really see it. Wher
178.
by wtetzner 8mo ago
Because this is Hacker News, not a marketing website. People may not only be interested in the tool itself, but also in the implementation.
179.
by wtetzner 8mo ago
> If you paint a picture of Sonic the Hedgehog in your living room, you are technically creating an unauthorized derivative work Is this even true? It might violate a trademark, but I don't think it would violate copyright law unles
180.
by wtetzner 8mo ago
> You could write a test that makes sure the output of someCall is passed directly to printeven without being modified. But why would anyone ever do that? There's nothing incorrect about the code, it's just less efficient than

More ›