The AI is taking our jerbs thread

mid_gen said:
Have you been living under a rock? :p Yes these are problems that a lot of people are thinking about.

The ROI on automating a process that was previously done by a human is so clear and obvious that the market forces behind this transition will be unstoppable.

I'm not worried about 60% for now, what I'd be concerned about is the fact that I don't think our society will deal with a 10-20% reduction in jobs over the next decade.
Click to expand...
The ROI won't be there if no one can afford the services that ai have built for us...


See that nice app on your mobile phone that u pay per month that was made by ai.. Nope, going to cancel that as I need that money to buy food instead.

Need some legal advice from a legal broker run by ai? Na can't afford that either...

List goes on. The roi is only there if most people are working and earning a wage
 
jonneymendoza said:
The ROI won't be there if no one can afford the services that ai have built for us...


See that nice app on your mobile phone that u pay per month that was made by ai.. Nope, going to cancel that as I need that money to buy food instead.

Need some legal advice from a legal broker run by ai? Na can't afford that either...
Click to expand...
In this fantasy scenario you have created where most people don't have a job....it doesn't matter if the services are driven by AI or not.

jonneymendoza said:
List goes on. The roi is only there if most people are working and earning a wage
Click to expand...
Your fantasy world where nobody works has no impact whatsoever on the ROI proposition of replacing people with AI automation. It is extremely attractive, and *is* happening, right now, in reality.
 
mid_gen said:
In this fantasy scenario you have created where most people don't have a job....it doesn't matter if the services are driven by AI or not.


Your fantasy world where nobody works has no impact whatsoever on the ROI proposition of replacing people with AI automation. It is extremely attractive, and *is* happening, right now, in reality.
Click to expand...
Again. Where does the money come from? Where do the big corporates get there money form if we don't give them any due to being jobless?

Of course I am fanatsisis and exaturate a bit but it is to discuss the situation that could happen if ai takes over most people's jobs
 
jonneymendoza said:
Again. Where does the money come from? Where do the big corporates get there money form if we don't give them any due to being jobless?
Click to expand...
"Where will big corporations get their money" is not the big problem you think it is, or what people are concerned about.

What most people are concerned about is what happens when big corporates *continue* to rake in vast amounts of money while increasingly automating tasks and reducing the demand for workers, leaving the state to provide a living for an increasingly large portion of the population.

Ultimately, that answer is either complete societal breakdown, or the large corporation siphoning off ever more money will have to start contributing more to the state.
 
mid_gen said:
"Where will big corporations get their money" is not the big problem you think it is, or what people are concerned about.

What most people are concerned about is what happens when big corporates *continue* to rake in vast amounts of money while increasingly automating tasks and reducing the demand for workers, leaving the state to provide a living for an increasingly large portion of the population.

Ultimately, that answer is either complete societal breakdown, or the large corporation siphoning off ever more money will have to start contributing more to the state.
Click to expand...
Basically we are ******..

That was my overall point.
 
The motor vehicle put farriers out of business but it didn't stop them becoming a delivery driver or mechanic

E: reliable clocks put people out of work
 
Last edited:
A machine-learning algorithm walks into a bar. The bartender asks: “What will you have?” The algorithm says: “What’s everyone else having?”

Copied from the guardian unfortunately (but at least by a human).
 
d_brennen said:
The motor vehicle put farriers out of business but it didn't stop them becoming a delivery driver or mechanic

E: reliable clocks put people out of work
Click to expand...

If your work is in the technology (shaping it, or operating it) then AI will break the role.

The problem that AI is attempting to solve is payroll costs.
 
Taken from LI..
Models perform better when matching with higher confidence - higher probability - and worse when matching with lower confidence. The fewer examples it's trained on, the lower the confidence, and the less accurate the predictions. It "guesses" more.
Click to expand...

Well no **** sherlock.
 
mid_gen said:
Learn how to best leverage LLMs to enhance your productivity, basically.

Understand what value you bring to the software development pipeline that LLMs will struggle to replicate, position yourself in those niches.
Click to expand...

Eventually you won't be able to learn development faster than an AI can. All the niches will be gone.

However my whole career I've been frequently blocked from automating things by the stubborn and the stupid. I expect AI will have the same issues.
 
NickK said:
Taken from LI..


Well no **** sherlock.
Click to expand...


But it is important problem for LLMs and why at least senior developers are safe for a long time. Since LLMs and even LRMs don't reason and merely do approximate pattern matching and sequence generation, if there are unseen, rare or new token sequences or requests then they fail badly and hallucinate.

You see this when doing AI coding in that Co-pilot or Cursor whatever hallucinates a new API, or incorrectly calls an API because the name is semantically similar to a required function.
 
D.P. said:
But it is important problem for LLMs and why at least senior developers are safe for a long time. Since LLMs and even LRMs don't reason and merely do approximate pattern matching and sequence generation, if there are unseen, rare or new token sequences or requests then they fail badly and hallucinate.

You see this when doing AI coding in that Co-pilot or Cursor whatever hallucinates a new API, or incorrectly calls an API because the name is semantically similar to a required function.
Click to expand...

Agreed. However the amount of rubbish being spat out is unbelievable.
 
NickK said:
Agreed. However the amount of rubbish being spat out is unbelievable.
Click to expand...


The problem is also most of the training data for coding is public domain like stack overflow posts. The LLMs learn solutions to simple coding problems , but in these environments people provide working solutions, not production level vode that is optimized, maintainable, handles edge cases, is secure etc. Production code looks very different to hacked together github project someone chucked together in a hackday. And with more LLM generated code appearing on the internet, the quality of the training data is only reducing. And the thing is as a senior software developer you can take the suggested code completion and using experience, context of the project etc, improve the code, make it secure and handle every edge case or simply throw it away and code by hand. Junior developers tend to rely heavily on the raw output because "it works", and even has Co-pilot generated UTs "to prove it"
 
Models like Anthropic’s aren’t just scraping GitHub and StackOverflow, there are all kinds of strategies employed including various adversarial and reinforcement learning techniques to generate high quality training data.

They are quite capable of producing high quality code once you take the time to properly adapt to an LLM workflow. Claude is very good once you have good context setup and discuss the implementation plan with it before committing to execution. Using the various degrees of ‘thinking’ keywords etc.

Sure seniors that are switched on and learn the tools aren’t going to be short of work in the short term, but it’s still going to have a massive impact on the industry and number of roles available.
 
mid_gen said:
Models like Anthropic’s aren’t just scraping GitHub and StackOverflow, there are all kinds of strategies employed including various adversarial and reinforcement learning techniques to generate high quality training data.

They are quite capable of producing high quality code once you take the time to properly adapt to an LLM workflow. Claude is very good once you have good context setup and discuss the implementation plan with it before committing to execution. Using the various degrees of ‘thinking’ keywords etc.

Sure seniors that are switched on and learn the tools aren’t going to be short of work in the short term, but it’s still going to have a massive impact on the industry and number of roles available.
Click to expand...

What've noted is that LLMs are good if you have a full specification first - then run through once to produce code.

I suspect the attention mechanisms fail to maintain the specification once it becomes an iterative definition. This latter limitation results in poor architecture and systems integrations from what I've seen so far.

Do you use more persona based rule sets in the context? For example - adding the security analyst, the operational monitoring, the financial cost management (for both cost visibility and savings etc)?
 
You must log in or register to reply here.
Back
Top Bottom