I've been working in banking/finance IT for the past 12 years and yes! The screwups in hyper complex systems tend to become less frequent but far more serious in their impact due to concentration to a few data centres and a handful of hyperscalers. A third-party vendor of AWS deploys a wrong IP list - boom! Half the street in a local financial hub is unreachable, across mutliple companies and industries.
it's definitely vibe coding pushing half-backed llm code strait to prod without proper feature flags or automated rollbacks. Everyone is shipping faster but the guardrails just aren't there yet
> What prospective guardrails are being considered or used?
The parent mentioned two: (1) feature flags, (2) rollbacks. You can add tests in CI, etc. YMMV. The LLM will do that if you ask it. But someone needs to spent time on it, test, etc. And most engineers don't get paid to ship tests / feature flags / rollbacks ... they get paid to test features.
"It's all the same, only the names have changed..." - JBJ
Something I'm seeing is good organisations doing good things, bad organisations getting worse and average organisations mostly staying flat or trending down. This is happening across multiple domains, not just an IT-centric thing.
What I'm putting it down to is the general culture/strategy of the organisation as opposed to the tools. It just so happens that tech is enabling you to accelerate whatever trajectory you were already on.
Note: This is my observation, working mostly in supply chains, at the bottom end of the earth. YMMV.
If you push systems to failure by squeezing workers, continually rolling layoffs, and disregarding quality (because consumers are sticky and will put up with a lot, especially if there are no other options), systems eventually fail. Money is expensive now [1], labor supply will decline into the future [2], its a fight for profits and growth versus labor in an unfavorable macro. "Productivity."
(also seeing similar failures, both as a consumer and inside clients pushing their organizations as systems to failure, "how few people or inexpensive talent can we operate this business with?")
I think this makes up for an interesting question.
my interpretation is similar to mitchell's that there are influential people within tech (jobs,influencers and just about anything) who are genuinely and utterly convinced with AI psychosis.
If you create a non-deterministic tech that is so powerful to psychosis people (who well should know better) then, it creates for problems down the line and you are just a side-effect of that. Especially, lately it seems that most AI is directed towards investors and not consumers (There is HN post trending about it essentially)
So well, you are just side-product of a larger psychosis. I believe that the tech is cool and it has genuine use-cases and many things but there has to be nuanced opinions about it but we as humans are similar to computers in this one thing that we want either 1 and 0 and can't capture this particular nuance perhaps.
I don’t think we used to be this way. I think the binary aspect of modern discourse was designed. A lot of arguments are reduced to us v them. It is as if holding two simultaneous opinions is now a bag thing.
Politicians used to rely on facts. On evidence. Colbert, like him or not, hit the nail on the head over a decade ago: “We live in a post-truth world.”
Maybe it’s a side effect of modern media platforms. You either like or dislike. If you don’t do either… do you even exist?
The tribalism has been pushed to an incredible level. You can’t agree with part of someone’s discourse. You either fully agree or fully disagree with them.
I think that people still hold nuanced opinions. It's just that the algorithm doesn't reward that. So perhaps there is some person writing nuanced opinions but there would be no views of it.
At the end perhaps I will sound un-knowledgable and technically I am because there is just so much to learn in world! but my opinion on AI has started to faze into an: "I don't know man"
so... that doesn't meaningfully add to people's life anything. What they want is validation, that either they are right about using AI and they can print millions with it or that they are right about AI and its all a fad and how they are right for not using it and falling into the trap.
So both these views get algorithm'd to people in their own feeds, especially completely pro AI psychosis. They interact with the polarizing think-piece and they achieve their goal of monetization or what-not.
This is my opinion on social media, you can find nuanced opinions but its like finding needle in hay-stack.
I want to be honest here and say this, there is a difference in using AI as well. For example, I really like to use AI to automate other people's work if I see someone doing some work. I really just think of (any) technology which can help automate or do anything which can help them or just out of curiosity, and I also build things for fun and for my own use cases.
but at the same time, I am against the fact of using AI in software industry as I have found enough evidence that aside from prototyping, it can still fall short in numerous things. I have found anecdotal evidence where tech engineers are expected to get 10x and then burn out and personally called out if they are honest ie. honesty isn't favoured yet other employees are working till late night just to do the things that they have promised.
What the fuck is this. Management doesn't even probably know or if they know, they might as well leave it knowingly unknown because this allows them to get work easier... I wish to use stronger words like exploitation as well because that is what is happening to engineers or designers or many other people just about in general.
Basically, although the intelligence metric has skyrocketed and intelligence might've become commoditized. I am unsure what is left behind or the lateral expectations left behind if we are all expected to be less honest and be more 10x engineers while that is simply not possible if you are actually good after a point. There's so much nuance that gets lost in these discussions imo but I have tried to give some nuance that I have seen personally visible talking to people online and offline.
I've been working in banking/finance IT for the past 12 years and yes! The screwups in hyper complex systems tend to become less frequent but far more serious in their impact due to concentration to a few data centres and a handful of hyperscalers. A third-party vendor of AWS deploys a wrong IP list - boom! Half the street in a local financial hub is unreachable, across mutliple companies and industries.
it's definitely vibe coding pushing half-backed llm code strait to prod without proper feature flags or automated rollbacks. Everyone is shipping faster but the guardrails just aren't there yet
I'm just dipping my toes into AI. What prospective guardrails are being considered or used?
> What prospective guardrails are being considered or used?
The parent mentioned two: (1) feature flags, (2) rollbacks. You can add tests in CI, etc. YMMV. The LLM will do that if you ask it. But someone needs to spent time on it, test, etc. And most engineers don't get paid to ship tests / feature flags / rollbacks ... they get paid to test features.
"It's all the same, only the names have changed..." - JBJ
> What prospective guardrails are being considered or used?
The most important guardrail we are using at the moment is: "Make no mistake"
/s
Something I'm seeing is good organisations doing good things, bad organisations getting worse and average organisations mostly staying flat or trending down. This is happening across multiple domains, not just an IT-centric thing.
What I'm putting it down to is the general culture/strategy of the organisation as opposed to the tools. It just so happens that tech is enabling you to accelerate whatever trajectory you were already on.
Note: This is my observation, working mostly in supply chains, at the bottom end of the earth. YMMV.
If you push systems to failure by squeezing workers, continually rolling layoffs, and disregarding quality (because consumers are sticky and will put up with a lot, especially if there are no other options), systems eventually fail. Money is expensive now [1], labor supply will decline into the future [2], its a fight for profits and growth versus labor in an unfavorable macro. "Productivity."
(also seeing similar failures, both as a consumer and inside clients pushing their organizations as systems to failure, "how few people or inexpensive talent can we operate this business with?")
[1] https://fred.stlouisfed.org/series/FEDFUNDS
[2] https://news.ycombinator.com/item?id=47680794 (citations)
I think this makes up for an interesting question.
my interpretation is similar to mitchell's that there are influential people within tech (jobs,influencers and just about anything) who are genuinely and utterly convinced with AI psychosis.
If you create a non-deterministic tech that is so powerful to psychosis people (who well should know better) then, it creates for problems down the line and you are just a side-effect of that. Especially, lately it seems that most AI is directed towards investors and not consumers (There is HN post trending about it essentially)
So well, you are just side-product of a larger psychosis. I believe that the tech is cool and it has genuine use-cases and many things but there has to be nuanced opinions about it but we as humans are similar to computers in this one thing that we want either 1 and 0 and can't capture this particular nuance perhaps.
I don’t think we used to be this way. I think the binary aspect of modern discourse was designed. A lot of arguments are reduced to us v them. It is as if holding two simultaneous opinions is now a bag thing.
Politicians used to rely on facts. On evidence. Colbert, like him or not, hit the nail on the head over a decade ago: “We live in a post-truth world.”
Maybe it’s a side effect of modern media platforms. You either like or dislike. If you don’t do either… do you even exist?
The tribalism has been pushed to an incredible level. You can’t agree with part of someone’s discourse. You either fully agree or fully disagree with them.
I think that people still hold nuanced opinions. It's just that the algorithm doesn't reward that. So perhaps there is some person writing nuanced opinions but there would be no views of it.
At the end perhaps I will sound un-knowledgable and technically I am because there is just so much to learn in world! but my opinion on AI has started to faze into an: "I don't know man"
so... that doesn't meaningfully add to people's life anything. What they want is validation, that either they are right about using AI and they can print millions with it or that they are right about AI and its all a fad and how they are right for not using it and falling into the trap.
So both these views get algorithm'd to people in their own feeds, especially completely pro AI psychosis. They interact with the polarizing think-piece and they achieve their goal of monetization or what-not.
This is my opinion on social media, you can find nuanced opinions but its like finding needle in hay-stack.
I want to be honest here and say this, there is a difference in using AI as well. For example, I really like to use AI to automate other people's work if I see someone doing some work. I really just think of (any) technology which can help automate or do anything which can help them or just out of curiosity, and I also build things for fun and for my own use cases.
but at the same time, I am against the fact of using AI in software industry as I have found enough evidence that aside from prototyping, it can still fall short in numerous things. I have found anecdotal evidence where tech engineers are expected to get 10x and then burn out and personally called out if they are honest ie. honesty isn't favoured yet other employees are working till late night just to do the things that they have promised.
What the fuck is this. Management doesn't even probably know or if they know, they might as well leave it knowingly unknown because this allows them to get work easier... I wish to use stronger words like exploitation as well because that is what is happening to engineers or designers or many other people just about in general.
Basically, although the intelligence metric has skyrocketed and intelligence might've become commoditized. I am unsure what is left behind or the lateral expectations left behind if we are all expected to be less honest and be more 10x engineers while that is simply not possible if you are actually good after a point. There's so much nuance that gets lost in these discussions imo but I have tried to give some nuance that I have seen personally visible talking to people online and offline.
yes