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

Why is AI so slow to spread?

economist.com

Post reply on HN
Print view View original

151–160 of 189 posts

Re: Why is AI so slow to spread?

#151

Post by spacemadness »

ike2792 wrote: post #149
crinkly wrote: post #33

Among non-technical friends, after the initial wow factor, even limited expectations were not met. Then it was tainted by the fact that everyone is promoting it as a human replacement technology which is then a tangible threat to their existence. That leads to not just lack of adoption but active sabotage. And then there’s the large body of people who just haven’t noticed it at all because they don’t give a shit. Stu…

I would add that no one has shown any solid business benefit or productivity gain due to AI despite there being massive incentives for someone to do so. We've tried to use AI for multiple projects within my company and none of them have panned out.

The two groups I see speaking to massive productivity gains are CEOs of AI related companies, for obvious reasons, and software engineers and influencers that want a piece of the pie or to show their bosses they’re learning the technology. I think it’s important to understand what productivity means as well. Auto generating a document isn’t a productivity gain if it contains a bunch of falsehoods or ignores data that an expert in the field might catch.

Re: Why is AI so slow to spread?

#152

Post by dwaltrip »

logicchains wrote: post #148
surgical_fire wrote: post #116

Earlier quoted context omitted.

> I find this perspective so hard to relate to. LLMs have completely changed my workflow; the majority of my coding has been replaced by writing a detailed textual description of the change I want, letting an LLM make the change and add tests, then just reviewing the code it wrote and fixing anything stupid it did. I use it in much the same way you describe, but I find that it doesn't save me that much time. It may s…

>Reviewing the code it writes to fix the inevitable mistakes and making adjustments takes time too, and it will always be a required step due to the nature of LLMs. Yep but this is much less time than writing the code, compiling it, fixing compiler errors, writing tests, fixing the code, fixing the compilation, all that busy-work. LLMs make mistakes but with Gemini 2.5 Pro at least most of these are due to under-spec…

Do you have any example prompts of the level of specificity and task difficulty you usually do? I oscillate between finding them useful and finding it annoying to get output that is actually good enough.

How many iterations does it normally take to get a feature correctly implemented? How much manual code cleanup do you do?

Re: Why is AI so slow to spread?

#154

Post by spacemadness »

surgical_fire wrote: post #118
specproc wrote: post #90

Earlier quoted context omitted.

I wouldn't underestimate the anxiety it's causing. Most of my social circle are non-technicial. A lot of people have had a difficult time with work recently, for various reasons. The global economic climate feels very precarious, politics is ugly, people feel powerless and afraid. AI tends to come up in the "state of the world" conversation. It's destroying friends' decade old businesses in translation, copywriting a…

> Corporate enthusiasm for AI is seen for what it actually is, a chance to cut head count I mean, it is the reason why the usual suspects push it so aggressively. The underclasses much always be pushed down. It's mostly bullshit, on most areas LLMs cannot reliably replace humans. But they embrace it because the chance that it might undermine labor is very seductive.

Instead of taking time to understand how it may be effective, a lot of leadership decided immediately that people could take on double the work they were doing or more right out the gate as long as they use AI. I’ve seen people driven to tears in this environment right now from stress due to overwork. Everyone in my org was either burnt out or soon to be. And that overwork is from managers demanding more and more without any data to back up exactly how much more is reasonable. Then you get these types on HN calling people luddites for having strong opinions and anxieties as if it’s only ever about the technology itself and not the effect it has on actual people in a cutthroat capitalist system. That’s exactly the sort of thing that brought the term “tech bro” into the limelight.

Re: Why is AI so slow to spread?

#155

Post by rurp »

logicchains wrote: post #100
sbt wrote: post #54

I have been using it for coding for some time, but I don't think I'm getting much value out of it. It's useful for some boilerplate generation, but for more complex stuff I find that it's more tedious to explain to the AI what I'm trying to do. The issue, I think, is lack of big picture context in a large codebase. It's not useless, but I wouldn't trade it for say access to StackOverflow. My non-technical friends are…

>I have been using it for coding for some time, but I don't think I'm getting much value out of it. I find this perspective so hard to relate to. LLMs have completely changed my workflow; the majority of my coding has been replaced by writing a detailed textual description of the change I want, letting an LLM make the change and add tests, then just reviewing the code it wrote and fixing anything stupid it did. This…

If you ever end up working on large complicated code bases you'll likely have an easier time relating to the sentiment. LLMs are vastly better at small greenfield coding than for working on large projects. I think 100% of the people I've heard rave about AI coding are using them for small isolated projects. Among people who work on large projects sentiment seems to range from mildly useful to providing negative value.

Re: Why is AI so slow to spread?

#156

Post by ryandrake »

spacemadness wrote: post #151
ike2792 wrote: post #149

Earlier quoted context omitted.

I would add that no one has shown any solid business benefit or productivity gain due to AI despite there being massive incentives for someone to do so. We've tried to use AI for multiple projects within my company and none of them have panned out.

The two groups I see speaking to massive productivity gains are CEOs of AI related companies, for obvious reasons, and software engineers and influencers that want a piece of the pie or to show their bosses they’re learning the technology. I think it’s important to understand what productivity means as well. Auto generating a document isn’t a productivity gain if it contains a bunch of falsehoods or ignores data that…

> software engineers and influencers that want a piece of the pie or to show their bosses they’re learning the technology.

This breathlessly-positive performance art shows up all the time on HN, whether or not the article is even about AI. I can believe it when someone points out specific tasks that the tool is good at and why, but when someone shows up and extolls AI as the greatest human achievement since the steam engine and how it's going to change life as we know it, it just feels like they are performing for their manager or something.

Re: Why is AI so slow to spread?

#157

Post by saratogacx »

markbao wrote: post #19

There was an article on HN a few days back on how it’s very hard to convey all the context in your head about a codebase to solve a problem, and that’s partly why it’s actually hard to use AI for non-trivial implementations. That’s not just limited to code. I don’t use AI for most of my product work because it doesn’t know any of the nuances of our product, and just like doing code review for AI is boring and tedious…

I think architecture needs to evolve as well. Up until now the primary audiences were human comprehension and build tools. We may need to have more context embedded into software architecture and make systematic segmentation easier for an LLM to sus out as part of the concerns involved in our designs.

LLM's won't make it on their own because there isn't enough data with their concerns in mind for them to learn from, just a few happy accidents where an LLM can excel in that specific code base.

Re: Why is AI so slow to spread?

#158

Post by ryandrake »

acdha wrote: post #134
csa wrote: post #66

Earlier quoted context omitted.

> Among non-technical friends, after the initial wow factor, even limited expectations were not met. Indeed. I think that current AI tech needs quite a bit of scaffolding in order for the full benefits to be felt by non-tech people. > Then it was tainted by the fact that everyone is promoting it as a human replacement technology Yeah. This is a bad move. AI is a human force multiplier (exponentializer?). > which is t…

I wouldn’t be so dismissive: “force multiplier” means job loss unless there’s a large amount of work which isn’t currently being done. As you live in a society, it really is important to think about what happens if we get the mass layoffs almost all of the executive-class are promising. There are some new jobs around the tech itself, but that doesn’t help those people unless they can land one of the new jobs – and ev…

> I wouldn’t be so dismissive: “force multiplier” means job loss unless there’s a large amount of work which isn’t currently being done.

I think there is a massive amount of work that's currently not being done, industry-wide. Everywhere I've worked has had something like 4-10X more work to do than staff to do it. The feature and bug backlogs just endlessly grow because there is no capacity to keep up. Companies should be able to adopt a force multiplier without losing staff: It could bring that factor down to 2-5X. The fact that layoffs are happening industry-wide shows that leadership might not even know what their workers could be doing.

Re: Why is AI so slow to spread?

#159

Post by tim333 »

The Economist doesn't seem to consider that it just doesn't work very well yet and the executives who "rhapsodise about all the wonderful ways in which their business is using artificial intelligence" are trying to keep up appearances.

I've found it works quite well as a variant of a google search but trying to exchange a shock absorber that had been ordered incorrectly by dealing with an AI assistant was quite a mediocre experience. It was able to say do you want to return your order but not understand I had a left one instead of a right one. The hype seems to have run ahead of the reality.

Re: Why is AI so slow to spread?

#160

Post by SAI_Peregrinus »

skeletal88 wrote: post #94
kaoD wrote: post #53

Earlier quoted context omitted.

They mean slow to actually be used by people. Doesn't matter that it's shoved everywhere if in the end there are 0 users clicking the magic wand icon.

This is it. I don't want to use some kind of AI search in fb messenger, I want to find someone to send them a message. I don't need AI search in google, I don't need it in gmail. What I need is search that actually works, there is no need to replace existing search functionality with "AI". It is shoved everywhere but nobody really needs it.

My employer has Glean AI search. It's better than Atlassian's search for Jira/Confluence, and can simultaneously search Slack, Google Drive, & more tools where we've haphazardly scattered important documentation like sqwrls with ADHD.

Being better than Atlassian's search isn't a difficult hurdle to clear, but it's one of the only areas where I've seen a noticeable improvement caused by an "AI" product. Not sure the search actually uses AI or just happens to be a decent cross-product search tool, there's a separate interface to chat with it & have it summarize things so the search might not really be using the AI.

Post reply on HN