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

It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)

publicznyprofil.github.io

Post reply on HN
Print view View original

91–100 of 143 posts

Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)

#91

Post by jrflo »

sambapa wrote: post #90
jrflo wrote: post #81

Earlier quoted context omitted.

I think this is a good example of the moving goalposts. The Turing test is not "AI is indistinguishable from human writing 100% of the time", it's whether it is possible at all for a system can be designed where a person can be fooled into thinking they're talking to a human when they're actually talking to a computer in a turn-based text exchange. It is a "there exits..." problem rather than a "for all..." problem.

That's not a Turing test

It is the original Turing test: https://en.wikipedia.org/wiki/Turing_test#Versions

Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)

#92

Post by ofjcihen »

Ukv wrote: post #84
ofjcihen wrote: post #70

Earlier quoted context omitted.

Have outputs from engineers traditionally been measured in cost and speed? Remember, we aren’t just talking about the product you create. While you would measure deliverables by cost and speed are we ignoring something else? Something that could potentially be more important than either of those metrics?

> Have outputs from engineers traditionally been measured in cost and speed? Yes. How long it'll take and how much it'll cost are going to be among pretty much any customer's first questions. They're not the only considerations, and could potentially be outweighed by other concerns even when quality is the same, but I think they are the main drives of AI adoption in industry. If error rate is the same, a $1/hr (amort…

Oh machine learning has been useful for measuring deterministic and non deterministic outputs for a long time.

But that’s not the argument here, is it?

So the question still stands.

Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)

#93

Post by overgard »

AndrewKemendo wrote: post #27
ofjcihen wrote: post #25

Earlier quoted context omitted.

This argument style is always humorous. The intention is something like “so humans are as bad as AI” when the original question boils down to something like “why would I replace humans with AI?”.

The entire purpose of automation is to remove a capacity limited human from a continuous workflow because the workflow is more capably achieved with fewer errors than the human See: traffic lights

An AI traffic light sounds like an excellent way to kill a lot of people

Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)

#94

Post by analognoise »

Remember: people used to say playing Chess was “AI”. There were dedicated Chess chips, and it was a massive AI undertaking.

We’re treating LLMs like we have every other tool. It’s going to become like spellcheck eventually, not something you think about, and certainly not worth hundreds of billions.

Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)

#95

Post by overgard »

cwbuilds wrote: post #26
solumunus wrote: post #19

Earlier quoted context omitted.

It’s not meaningless because AGI = genuinely being able to replace humans in almost any job. Sure, AI is getting better at stuff that humans direct it to do, but the fact that a human is required in the loop is super meaningful.

Maybe I'm wrong, but I think it will always be used as a tool however smart it gets. In software, it hasn't replaced all the software engineers (even though it's better and faster than us), it's just meant that we now have more leverage. We can write more code and work on more projects than ever. I don't see why we would suddenly stop using it like that?

"AI babysitter" sounds more like a minimum wage job, not a career path. Outsourcing your primary skill to a machine creates less leverage. I'm convinced this is why CEO's love it -- even though you can't replace software devs, you can hold an axe over their head with the threat.

Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)

#96

Post by topgrain2 »

sanderjd wrote: post #77
topgrain2 wrote: post #54

As my company experiments with AI-heavy programming, I don’t feel like my job’s going away, but as a “full stack” programmer who’s also good at and comfortable managing a task board, walking “stakeholders” and SMEs through requirements-gathering, et c., I am increasingly wondering why our “teams” still have a couple non-programmers doing that stuff. Dedicated QA, for that matter—all that test-writing and test-data-ge…

In my experience, those roles never existed to begin with in well run teams and companies. Some companies drive product managers into this role, but that was always a bad use of their time. Even with AI tools, I still don't feel like I have the skills to do the job of a good product manager, which requires the kind of vision and business acumen that I am just not as good at as the best PMs I've worked with.

Yeah right after I posted I thought about editing in that a 3-4 all-programmer team with one person (maybe half-time) acting as basically team secretary was already the best unit for software writing, but bigcos seemed to have trouble forming them without adding a bunch of overhead and making their programmers often blocked or idle for some reason.

It’s just that now if I have an LLM that can talk to my business comms and reporting tools (jira was already mostly for managers to generate reports, not to help the people doing the work, that’s why it sucks so much as a tool for getting actual work done) and my first-pass at programming looks a lot like just writing a programmerly-flavored feature or bug ticket, having other people making useless first drafts of tickets for me to rewrite (and probably have to go clarify them with an SME or stakeholder anyway) and someone dedicated to poking around in Jira in general, is gonna start looking funny even to bigcos, I think.

Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)

#97

Post by AnotherGoodName »

suddenlybananas wrote: post #86
jrflo wrote: post #81

Earlier quoted context omitted.

I think this is a good example of the moving goalposts. The Turing test is not "AI is indistinguishable from human writing 100% of the time", it's whether it is possible at all for a system can be designed where a person can be fooled into thinking they're talking to a human when they're actually talking to a computer in a turn-based text exchange. It is a "there exits..." problem rather than a "for all..." problem.

If it's a there exists problem, then it was solved by ELIZA in the 60s.

I think your post is a very good example of goal post moving.

We currently have debates around whether content online is human or ai generated and instead of acknowledging the milesetone we have a post essentially claiming "the turing test was passed long ago, it's worthless".

Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)

#99

Post by Ukv »

ofjcihen wrote: post #92
Ukv wrote: post #84

Earlier quoted context omitted.

> Have outputs from engineers traditionally been measured in cost and speed? Yes. How long it'll take and how much it'll cost are going to be among pretty much any customer's first questions. They're not the only considerations, and could potentially be outweighed by other concerns even when quality is the same, but I think they are the main drives of AI adoption in industry. If error rate is the same, a $1/hr (amort…

Oh machine learning has been useful for measuring deterministic and non deterministic outputs for a long time. But that’s not the argument here, is it? So the question still stands.

My understanding of your argument is (paraphrasing):

> > People try to excuse AI issues/failure modes by saying humans have them too, but even if they're equally bad then what would be the whole point of replacing a human worker with AI?

To which my response is that speed and cost are also important factors, which can often give AI the edge in considerations when quality/error rate is equal.

If you meant something other than that, you may have to specify.

Re: It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)

#100

Post by cwbuilds »

overgard wrote: post #89
cwbuilds wrote: post #74

Earlier quoted context omitted.

I'm able to control my coding agent even though it writes code better and faster than I can.

The ultimate goal of this technology is to replace people -- if it's a tool and corporations still need to employ people to use the tool, then the valuations make no sense. So ultimately, you're going to need to hand the car keys to these agents more and more: here's access to production (we fired the devops people), here's access to all our internal messages and IP, here's all our source code (we fired the engineers…

It already has created jobs where people just act as reviewers and gatekeepers though. There are people who vibecoded products and are now running decent-sized businesses.
Post reply on HN