I keep watching for the AI bubble to burst.
-
I keep watching for the AI bubble to burst. It's been so expensive for so little returns.
At work, a senior leader asked me to present on AI. As a manager, I'm expected to adopt it, whether I can use it or not.
Our usage is tracked, so I've had to do some stuff with it. I tried setting up a complex process, since the easy things aren't really time-savers. I presented on this testing, even though I doubt I'm actually going to use it.
Every day, the AI tools seem to interpret the exact same prompts differently. So when I validate the output, it has changed.
How can I add an AI process into my workflow when the output is different every day? How much extra time am I wasting in validating the output? And don't tell me that my prompt is bad. It works until suddenly it doesn't. The tool is bad. Tools shouldn't do that.
I can write programming logic to do the same things. I know the code isn't going to give me different output every time I open it.
The senior leaders are desperate to find use cases for AI. The companies have invested a lot of money and want to see returns, but it's not happening. When are they going to cut their losses?
@VeeRat I too have been forced to use it.
Current strategy is to get LLM to write scripts, then I get the scripts to do the actual work, so it is predictable day to day.
MCP (JSON-RPC underneath the trenchcoats) is fairly amenable to use in scripts.
Then to keep the AI usage stats happy I have the script run an LLM at the end to give a "summary", even though the scripts render their own output before that.
When the bubble breaks, at least I have the scripts (assuming MCP doesn't also die with it), and I can manually disable the final LLM step.
-
I keep watching for the AI bubble to burst. It's been so expensive for so little returns.
At work, a senior leader asked me to present on AI. As a manager, I'm expected to adopt it, whether I can use it or not.
Our usage is tracked, so I've had to do some stuff with it. I tried setting up a complex process, since the easy things aren't really time-savers. I presented on this testing, even though I doubt I'm actually going to use it.
Every day, the AI tools seem to interpret the exact same prompts differently. So when I validate the output, it has changed.
How can I add an AI process into my workflow when the output is different every day? How much extra time am I wasting in validating the output? And don't tell me that my prompt is bad. It works until suddenly it doesn't. The tool is bad. Tools shouldn't do that.
I can write programming logic to do the same things. I know the code isn't going to give me different output every time I open it.
The senior leaders are desperate to find use cases for AI. The companies have invested a lot of money and want to see returns, but it's not happening. When are they going to cut their losses?
@VeeRat its not a tool.
-
I keep watching for the AI bubble to burst. It's been so expensive for so little returns.
At work, a senior leader asked me to present on AI. As a manager, I'm expected to adopt it, whether I can use it or not.
Our usage is tracked, so I've had to do some stuff with it. I tried setting up a complex process, since the easy things aren't really time-savers. I presented on this testing, even though I doubt I'm actually going to use it.
Every day, the AI tools seem to interpret the exact same prompts differently. So when I validate the output, it has changed.
How can I add an AI process into my workflow when the output is different every day? How much extra time am I wasting in validating the output? And don't tell me that my prompt is bad. It works until suddenly it doesn't. The tool is bad. Tools shouldn't do that.
I can write programming logic to do the same things. I know the code isn't going to give me different output every time I open it.
The senior leaders are desperate to find use cases for AI. The companies have invested a lot of money and want to see returns, but it's not happening. When are they going to cut their losses?
Are you tracked by token use?
The best answer I've heard for this is to set it up on a pointless loop and go do your work the old way.
-
@VeeRat its not a tool.
True. It's a cult.
But our demented leaders pretend its a tool ("look, those sacrifices of virgins on the altar of the sun god --- it's not something we want to do, personally. It's just a tool to ensure a good rain and harvest for the next year).
So, when we talk about how unreasonable this is, what the leadership expects, it makes sense to use "tool" in sentences like "this is not a suitable tool"...
-
@n_dimension @VeeRat The problem to be solved in this story is upper management wanting staff to use AI.
-
@assimilateborg Nah I’m not going to replace my coding team’s jobs with slop.
Apart from the fact, that the time comes where this code would need to be extended or updated just for maintenance reasons (like: "Please use the next version of the library, this has security problems)
And then it mutates.
-
@n_dimension @VeeRat The problem to be solved in this story is upper management wanting staff to use AI.
-
@n_dimension @VeeRat Experience would suggest not.
-
True. It's a cult.
But our demented leaders pretend its a tool ("look, those sacrifices of virgins on the altar of the sun god --- it's not something we want to do, personally. It's just a tool to ensure a good rain and harvest for the next year).
So, when we talk about how unreasonable this is, what the leadership expects, it makes sense to use "tool" in sentences like "this is not a suitable tool"...
@glitzersachen @VeeRat i don't think "suitable" helps there at all, they will just keep looking for a purpose then.
AI/llm is a tool as much as a sophisticated die. Fully refuting that a certain outcome can be achieved reliably IS the only way.
-
I keep watching for the AI bubble to burst. It's been so expensive for so little returns.
At work, a senior leader asked me to present on AI. As a manager, I'm expected to adopt it, whether I can use it or not.
Our usage is tracked, so I've had to do some stuff with it. I tried setting up a complex process, since the easy things aren't really time-savers. I presented on this testing, even though I doubt I'm actually going to use it.
Every day, the AI tools seem to interpret the exact same prompts differently. So when I validate the output, it has changed.
How can I add an AI process into my workflow when the output is different every day? How much extra time am I wasting in validating the output? And don't tell me that my prompt is bad. It works until suddenly it doesn't. The tool is bad. Tools shouldn't do that.
I can write programming logic to do the same things. I know the code isn't going to give me different output every time I open it.
The senior leaders are desperate to find use cases for AI. The companies have invested a lot of money and want to see returns, but it's not happening. When are they going to cut their losses?
@VeeRat sounds like you are trying to solve the wrong problem. different outputs would not necessarily be bad for certain use-cases. let multiple people analyse or summarise the same document document, or even have the same person do it multiple times, and you will also get different outputs. what you are doing wrong is that you are trying to use something that is a tool (in a weird sense) to replace something that can only come from an intelligent designer. your keyboard autocomplete will never write a novel for you.
-
@VeeRat Have you seen the Big Short? The bit where the guy keeps expecting everything to fall, but a bunch of rich assholes keep propping it up while they raid the piggy bank?
the same guy who the big short was based on has already shorted on ai companies
"AI valuation concerns took hold in the US as well after it was revealed the trader who inspired The Big Short has bet $1.1bn (£840m) on a fall in prices for AI-related stocks Nvidia and Palantir."
Trader who inspired The Big Short bets against AI as tech shares fall
https://www.bbc.co.uk/news/articles/c867vyn2evlo -
I keep watching for the AI bubble to burst. It's been so expensive for so little returns.
At work, a senior leader asked me to present on AI. As a manager, I'm expected to adopt it, whether I can use it or not.
Our usage is tracked, so I've had to do some stuff with it. I tried setting up a complex process, since the easy things aren't really time-savers. I presented on this testing, even though I doubt I'm actually going to use it.
Every day, the AI tools seem to interpret the exact same prompts differently. So when I validate the output, it has changed.
How can I add an AI process into my workflow when the output is different every day? How much extra time am I wasting in validating the output? And don't tell me that my prompt is bad. It works until suddenly it doesn't. The tool is bad. Tools shouldn't do that.
I can write programming logic to do the same things. I know the code isn't going to give me different output every time I open it.
The senior leaders are desperate to find use cases for AI. The companies have invested a lot of money and want to see returns, but it's not happening. When are they going to cut their losses?
@VeeRat Same as me.
-
@VeeRat
> When are they going to cut their losses?They can't. It is easy to fool a man, getting them to ack they were fooled is next to impossible.
https://www.groundbrkr.com/p/the-second-derivative-why-no-one
-
the same guy who the big short was based on has already shorted on ai companies
"AI valuation concerns took hold in the US as well after it was revealed the trader who inspired The Big Short has bet $1.1bn (£840m) on a fall in prices for AI-related stocks Nvidia and Palantir."
Trader who inspired The Big Short bets against AI as tech shares fall
https://www.bbc.co.uk/news/articles/c867vyn2evlo@Captain_Jack_Sparrow @VeeRat Seems like he has worked on his timing. I'll guess he wins on this one. It feels like we are at that time.
-
@VeeRat sounds like you are trying to solve the wrong problem. different outputs would not necessarily be bad for certain use-cases. let multiple people analyse or summarise the same document document, or even have the same person do it multiple times, and you will also get different outputs. what you are doing wrong is that you are trying to use something that is a tool (in a weird sense) to replace something that can only come from an intelligent designer. your keyboard autocomplete will never write a novel for you.
-
I keep watching for the AI bubble to burst. It's been so expensive for so little returns.
At work, a senior leader asked me to present on AI. As a manager, I'm expected to adopt it, whether I can use it or not.
Our usage is tracked, so I've had to do some stuff with it. I tried setting up a complex process, since the easy things aren't really time-savers. I presented on this testing, even though I doubt I'm actually going to use it.
Every day, the AI tools seem to interpret the exact same prompts differently. So when I validate the output, it has changed.
How can I add an AI process into my workflow when the output is different every day? How much extra time am I wasting in validating the output? And don't tell me that my prompt is bad. It works until suddenly it doesn't. The tool is bad. Tools shouldn't do that.
I can write programming logic to do the same things. I know the code isn't going to give me different output every time I open it.
The senior leaders are desperate to find use cases for AI. The companies have invested a lot of money and want to see returns, but it's not happening. When are they going to cut their losses?
Yeah, they're the behavior of a non deterministic toy not a tool.
-
-
I keep watching for the AI bubble to burst. It's been so expensive for so little returns.
At work, a senior leader asked me to present on AI. As a manager, I'm expected to adopt it, whether I can use it or not.
Our usage is tracked, so I've had to do some stuff with it. I tried setting up a complex process, since the easy things aren't really time-savers. I presented on this testing, even though I doubt I'm actually going to use it.
Every day, the AI tools seem to interpret the exact same prompts differently. So when I validate the output, it has changed.
How can I add an AI process into my workflow when the output is different every day? How much extra time am I wasting in validating the output? And don't tell me that my prompt is bad. It works until suddenly it doesn't. The tool is bad. Tools shouldn't do that.
I can write programming logic to do the same things. I know the code isn't going to give me different output every time I open it.
The senior leaders are desperate to find use cases for AI. The companies have invested a lot of money and want to see returns, but it's not happening. When are they going to cut their losses?
@VeeRat
Sunk cost fallacy on their part? -
Are you tracked by token use?
The best answer I've heard for this is to set it up on a pointless loop and go do your work the old way.
-
I keep watching for the AI bubble to burst. It's been so expensive for so little returns.
At work, a senior leader asked me to present on AI. As a manager, I'm expected to adopt it, whether I can use it or not.
Our usage is tracked, so I've had to do some stuff with it. I tried setting up a complex process, since the easy things aren't really time-savers. I presented on this testing, even though I doubt I'm actually going to use it.
Every day, the AI tools seem to interpret the exact same prompts differently. So when I validate the output, it has changed.
How can I add an AI process into my workflow when the output is different every day? How much extra time am I wasting in validating the output? And don't tell me that my prompt is bad. It works until suddenly it doesn't. The tool is bad. Tools shouldn't do that.
I can write programming logic to do the same things. I know the code isn't going to give me different output every time I open it.
The senior leaders are desperate to find use cases for AI. The companies have invested a lot of money and want to see returns, but it's not happening. When are they going to cut their losses?
@VeeRat I mean... give the presentation with this conclusion?
"I tried to use it, it is too unreliable, anything I try to do with it could be better done with just an old fashioned program, it would be better to cut our losses and not waste more money on this." is a perfectly reasonable conclusion.
