Comment by sbarre | Hacker News Reader

Comment by sbarre | Hacker News Reader

Parent Full thread

> You don’t know if it will get better. Even if it does, you don’t know by how much or the time frame. You don’t know if it will ever improve enough to overcome the current limitations. You don’t know if it will take years.

You are _technically_ correct but if I base my assumptions on the fact that almost all worthwhile software and technology has gotten better over the years, I feel pretty confident in standing behind that assumption.

> In the meantime, while someone is sitting on their ass for years waiting for the uncertain future of the tool getting better, someone else is getting their hands dirty, learning the craft, improving, having fun, collaborating, creating.

This is a pretty cynical take. We all decide where we prioritize our efforts and spend our time in life, and very few of us have the luxury to freely choose where we want to focus our learning.

While I wait for technologies I enjoy but haven't mastered to get better, I am certainly not "sitting on my ass".. I am dedicating my time to other necessary things like making a living or supporting my family.

In this specific case I wish I could spend hours and hours getting good at Blender and 3D modelling or animation. Dog knows I tried when I was younger.. But it wasn't in the cards.

I'm allowed to be excited at the prospect that technology advancements will make this more accessible and interesting for me to explore and enjoy with less time investment. I also want to "get my hands dirty, learn, improve, have fun, create" but on my own terms and in my own time.

Any objection to that is shitty gatekeeping.

Since you used the term "shitty gatekeeping": to me, your comment reads like the most generic kind of optimism you see everywhere on the internet about everything ever. Shitty optimism.

No one told you weren't allowed to be excited, but you took it that way anyway.

Fair enough! My original comment was pretty generic and flippant. That's totally valid.

> almost all worthwhile software and technology has gotten better over the years

You only know if it was worthwhile in hindsight. We aren’t there yet. And “better” is definitely debatable. We certainly do more things with software these days, but it’s a hard sell to unambiguously say it is better. Subscriptions everywhere, required internet access, invasions of privacy left and right, automated rejections, lingering bugs which are never fixed…

Your exact argument was given by everyone selling every tech grift ever. Which is not to say this specific case is another grift, only that you cannot truly judge the long-term impact of something while it is being invented.

> Any objection to that is shitty gatekeeping.

If gatekeeping is what you took from my comment, you haven’t understood it. Which could certainly mean my explanation wasn’t thorough enough. My objection is to the hand-wavy “this will only improve” commentary which doesn’t truly say anything and never advances the discussion, yet is always there. See the “low-hanging fruit” section of my other comment.

https://news.ycombinator.com/item?id=44932526

> You only know if it was worthwhile in hindsight.

Yes that's because, so far, we haven't been able to see the future. Which is why we base predictions and assumptions on past performance and on lived experience. Sometimes we will be wrong.

You're also arguing in the abstract here while I am speaking about this specific topic of using LLMs to improve 3D modelling tooling.

Are you arguing that neither LLMs or 3D modelling tools are "worthwhile"?

Are you suggesting that improvements that make these tools more accessible, even incrementally, are a bad thing?

Or are you just challenging my right to make assumptions and speculate? I realize that we may not be even on the same page here.

You're also cherry-picking a limited number of examples where software isn't better, and I agree with all those (and never said that software universally gets better), but those examples are a tiny subset of what software does in today's world.

It's starting to feel like you just want to argue.

My opinion is that, broadly speaking, software advancements have improved the world, and will continue to improve the world, both in big and small ways. There will of course be exceptions.

The absolute nanosecond someone suggests using this over skilled labor, the “shitty gatekeeping” roles reverse. I personally care more about the skilled labor than your optimism.

I am in no way implying that these advancements should/would make me better than a skilled professional, nor do I believe they will ever replace true expertise or craft.

These kinds of advancements lower the bar for entry and reduce the effort required to achieve better results, and so make the tools more accessible to more people.

And that is a good thing. The scenario you are implying is laying the choices of people at the feet of the tools, rather than holding the people accountable.

Also: history has, so far, mostly proven that the end result of better tools is better experts, not less experts.

> history has, so far, mostly proven that the end result of better tools is better experts

I’m inclined to agree with the general sentiment. However, it is not a given that LLMs are better tools. You don’t really get better at them in the same sense as before, you just type something and pray. The exact same thing you typed may produce exactly what you wanted, and aberration, or something close with subtle mistakes that you don’t have the expertise to fix. Other tools made you better at the craft in general.

> You don’t really get better at them in the same sense as before, you just type something and pray.

I very much disagree with this. I've spent the last 18 months working with LLMs daily at my work (I'm not exaggerating here) and while the models themselves have certainly gotten better, I have personally learned how to extract better results from the tools and improved my proficiency at applying LLMs as part of my overall toolchain and workflow.

LLMs are very much like many other tools in some ways, that the more you learn how to use them, the better results you will get.

I do agree that their non-deterministic nature makes them less reliable in some contexts as well though, but that's a trade-off you work into your approach, just like other trade-offs.

Reply on news.ycombinator.com