Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Malleable Software in the Age of LLMs (2023)

geoffreylitt.com

Post reply on HN
Print view View original

31–39 of 39 posts

Re: Malleable Software in the Age of LLMs (2023)

#31

Post by sublinear »

itishappy wrote: post #27
surfingdino wrote: post #9

Earlier quoted context omitted.

I have yet to meet a non-dev who wants to write or modify software themselves. Quite a few people still don't know that it's the software that makes appliances and devices do things and that it is written. They just don't know what software is so how would they know where to begin writing or modifying code?

You've never met a non-dev who complains about the UX of Outlook, wants to adjust the contents of their Twitter/Facebook feed, or wishes to tweak the styles of their blog?

Funny that you mention Outlook.

A lot of Microsoft Office and adjacent programs have customizable UIs almost to the point of absurdity for decades yet it's uncommon to see anyone actually change the default layouts, adjust panes, or view modes.

Maybe it's because everyone hates toolbars and ribbons, but why? Perhaps it's that the number of options is overwhelming. I am fairly certain most humans have seen code and their naive impression is that it's "cluttered"/"noisy". That impression is not because it's text or a ton of buttons staring back, but the cognitive overhead of actually understanding what you're looking at. LLMs can create or suggest things, but the user still has to comprehend something to continue interacting. That's where the interaction falls apart.

What I can see is LLMs taking natural language input and curating catalogs of software or plugins. As it stands searching for software by feature is a terrible experience even for devs.

Re: Malleable Software in the Age of LLMs (2023)

#32

Post by surfingdino »

itishappy wrote: post #27
surfingdino wrote: post #9

Earlier quoted context omitted.

I have yet to meet a non-dev who wants to write or modify software themselves. Quite a few people still don't know that it's the software that makes appliances and devices do things and that it is written. They just don't know what software is so how would they know where to begin writing or modifying code?

You've never met a non-dev who complains about the UX of Outlook, wants to adjust the contents of their Twitter/Facebook feed, or wishes to tweak the styles of their blog?

Outside of our IT bubble? No. People will complain that something doesn't work, but they have no desire to change it themselves.

Re: Malleable Software in the Age of LLMs (2023)

#33

Post by hdarshane »

ickelbawd wrote: post #20

I’m quite sure I don’t want a different bespoke UI for everything. And given the non-deterministic nature of these models it’s going to be different even for the same task. Now instead of hunting for the right buttons after a software “update” with some genius redesign: I’ll be hunting every time I use it? Sounds like a nightmare. I do agree with the author though as to the deficiencies with a purely chat based inter…

Non-deterministic nature of these models is definitely a concern. Interesting to see how companies implementing generative interface circumvent this. Maybe they create a set of fixed generic UI elements or write a system prompt describing general design guidelines. Still wouldn't fix it 100%.

Re: Malleable Software in the Age of LLMs (2023)

#34

Post by itishappy »

surfingdino wrote: post #32
itishappy wrote: post #27

Earlier quoted context omitted.

You've never met a non-dev who complains about the UX of Outlook, wants to adjust the contents of their Twitter/Facebook feed, or wishes to tweak the styles of their blog?

Outside of our IT bubble? No. People will complain that something doesn't work, but they have no desire to change it themselves.

Desire or ability?

Currently if I desire to make the "new mail" button always available, some technical individual would have to devote hours of their time to figuring out the required changes.

Replace a technical individual with an LLM and the story becomes completely different. People love complaining! If that's all that's required to make changes then I bet we'd see a lot more of it.

Re: Malleable Software in the Age of LLMs (2023)

#35

Post by itishappy »

sublinear wrote: post #31
itishappy wrote: post #27

Earlier quoted context omitted.

You've never met a non-dev who complains about the UX of Outlook, wants to adjust the contents of their Twitter/Facebook feed, or wishes to tweak the styles of their blog?

Funny that you mention Outlook. A lot of Microsoft Office and adjacent programs have customizable UIs almost to the point of absurdity for decades yet it's uncommon to see anyone actually change the default layouts, adjust panes, or view modes. Maybe it's because everyone hates toolbars and ribbons, but why? Perhaps it's that the number of options is overwhelming. I am fairly certain most humans have seen code and th…

I completely agree that the UX for toolbar customization currently sucks, but that's exactly what LLMs are being proposed as a replacement for.

When the UX is not a barrier, people love customizing their devices! My collogues of all abilities use different Outlook themes, and everybody's mother and their dog have changed their backgrounds and ringtones.

Re: Malleable Software in the Age of LLMs (2023)

#36

Post by Vegenoid »

chiasson wrote: post #2

> Until now, that vision has been bottlenecked on turning fuzzy informal intent into formal, executable code Software has been and always will be fundamentally a problem of communication. What many don't realize is that the true challenge of communication is not one of transmission, but clarity of thought and understanding. Any tool, software language, or AI coder will still be limited by the clarity and completeness…

> Any tool, software language, or AI coder will still be limited by the clarity and completeness of the specs presented.

> Nothing will ever turn your fuzzy intent into your clear best interest.

I've seen a lot of SWEs saying this, and while it's true to an extent, it misses a lot. Good engineers don't simply turn a given spec into code, indeed there are somewhat deprecating terms for these types of positions, like 'code monkey'.

A good engineer does not require the spec they receive to be absolutely precise. They will recognize the intent, and make good judgements about what the requester most probably wants. They will ask clarifying questions when there is important information missing, or a decision that needs to be made where it isn't clear what the requester wants.

LLMs can't do this very well right now, but it doesn't seem like a stretch to say that they will be able to. Will they be able to turn half-baked, very underspecified requests into exactly what the requestor is looking for with a press of a button? No. But I think they could get, and often already are, quite good at filling in the blanks. Seems like current LLMs have a way to go before they can recognize deficiencies in a prompt and ask for more info, but it's somewhere in the future.

Re: Malleable Software in the Age of LLMs (2023)

#37

Post by sublinear »

itishappy wrote: post #35
sublinear wrote: post #31

Earlier quoted context omitted.

Funny that you mention Outlook. A lot of Microsoft Office and adjacent programs have customizable UIs almost to the point of absurdity for decades yet it's uncommon to see anyone actually change the default layouts, adjust panes, or view modes. Maybe it's because everyone hates toolbars and ribbons, but why? Perhaps it's that the number of options is overwhelming. I am fairly certain most humans have seen code and th…

I completely agree that the UX for toolbar customization currently sucks, but that's exactly what LLMs are being proposed as a replacement for. When the UX is not a barrier, people love customizing their devices! My collogues of all abilities use different Outlook themes, and everybody's mother and their dog have changed their backgrounds and ringtones.

Yeah, but do they create folders and rules for their inbox?

Backgrounds and ringtones don't require any meaningful comprehension. It makes a lot of sense to assume the user would get an itch to stop looking at the same pictures and hearing the same ringtones all the time. That's as far as their desire for customization would go. It's similar to why cars have stereos and paint colors. Anyone can pay money for more customization, but few do.

Inbox rules are just about the simplest actually useful thing someone might do, but probably doesn't. That's a pretty clear signal to me they don't care. That end user cognition is the actual barrier.

In so many words, the user is dumb and complacent. Computers aren't going to think for them no matter how advanced the AI. They also don't want the computer thinking for them.

They want to get things done with stability. Computers are not their focus in life. Imagine an office full of people constantly changing how they work and forcing and politicking their bullshit onto other people (like devs do all the time). That's a normal person's nightmare.

Re: Malleable Software in the Age of LLMs (2023)

#38

Post by sublinear »

itishappy wrote: post #34
surfingdino wrote: post #32

Earlier quoted context omitted.

Outside of our IT bubble? No. People will complain that something doesn't work, but they have no desire to change it themselves.

Desire or ability? Currently if I desire to make the "new mail" button always available, some technical individual would have to devote hours of their time to figuring out the required changes. Replace a technical individual with an LLM and the story becomes completely different. People love complaining! If that's all that's required to make changes then I bet we'd see a lot more of it.

Complaining to, praying to, or bullying an AI god is not going to solve anything. I expect this god to bring upon apocalypse. In the world of software, by its very definition, the righteous ones can only be the developers.

Pleasing everyone was never a people problem or a logic puzzle. It's fundamentally intractable and trivial to see how quickly contradictions would arise in any meaningful set of rules.

Re: Malleable Software in the Age of LLMs (2023)

#39

Post by sublinear »

philipswood wrote: post #8
sublinear wrote: post #5

Earlier quoted context omitted.

The same points you made make me wonder if end users even want to modify or write a program. Even if it's possible, it's still a lot of work. I just don't see that as the typical consumer behavior.

Software often doesn't do the thing you want to do. It almost does it. Or it can just-just do it with contortions and a lot of repetitive toil on your side. I think writing little plugins and drivers to do the thing you want with an LLM is something that could be built into a lot of software. I don't think LLM can architect and build whole systems yet, but this niche is something that can be done.

> I think writing little plugins and drivers to do the thing you want with an LLM is something that could be built into a lot of software.

Containing the craziness within low level drivers sounds like the perfect way to do several great things: fuzz the OS for vulnerabilities, finally relieve embedded engineers of what they clearly hate so much, save good tech from the scrap heap. So many wins!

Post reply on HN