I’m curious if anyone has seen AI used effectively in their company.
We have been encouraged to try to incorporate AI into our processes, but almost all the improvements I’ve seen are using AI to summarize meetings, create clip art, and act as an improved search engine. None of these are BAD, but none of these uses has made a noticeable impact either. … well we did have to use more cloud storage for all the extra notes from meetings.
Anyone seen anything near the GAME-CHANGING promises made to the VC and investors?
LLMs have some limited use cases but the AI revolution that's been marketed seems like complete bs. I think most of the AI bubble pops sooner than later. I haven't seen anything indicating otherwise.
I work in visual effects. I’m using it on the side to speed myself up … if I have a complicated spreadsheet to put together or a GUI I need made around a script I wrote, it’s fantastic for that kind of stuff.
Haven’t seen it used yet in my more complex day to day
As a software developer, this is exactly my experience as well. It's great for small things that is tedious and has been done thousands of times by different people before. But it falls apart immediately if given any more complex task.
It simply can't keep track of things, and as far as I know that is by design of it being a stateless machine. You only grow the input tokes, and after a while there is so much garbage in the input, that it breaks and starts completely hallucinating, brings up obsolete code, etc.
Even with the small things, you have to babysit every output. They tend to make grave mistakes that can affect a project critically, even though the result is functionally correct. i.e. it switched matrix row/column order midway in some simple functions.
My rule of thumb is: Never ask AI for code that you couldn't write yourself.
I can absolutely say it’s made me a lot more efficient. I’ve seen it make a lot of people worse though because they don’t verify the information it gives.
My wife is a leasing consultant for a property management group. She works with four different properties. She gives tours for apartments, provides customer service, things like that. Her company shifted to using AI last year for tasks. AI is used to schedule tours, field maintenance requests, and create data for decision making.
Since using the programs, they’ve seen a huge increase in tours and maintenance requests. My wife spends less time scheduling, and providing customer service, and more time actually selling the apartments themselves. The skills needed for the job shifted from entry level to being more experienced. Since activity has increased, they’ve hired more maintenance techs and additional leasing consultants to keep up with volume and provide service.
So it’s been a really positive change, but for a while there it was a nightmare because they were not properly staffed to handle the volume at first.
This is a case if AI is being used properly, supporting continuing workers. However in most cases, its being used to replace. Did anyone in your wife's company get laid off due to the switch? If not then it wasn't being used to replace any jobs which is what a lot of worry about AI comes from
This all seems like something a CRM or something similar would have handled just fine, and is what is used for most situations like this. Why weren't they using something like that, and how were they taking requests before? If it's a small business (four properties doesn't sound like it's too small), I imagine an expensive CRM maybe wasn't cost efficient, but I don't see how AI actually replaced that type of tool at a more cost effective rate, since you'd still need to build a system to manage that type of stuff which would require at least a little application development knowledge, which would mean hiring a dev at least under contract for a minute to make sure it works even if it's just vibe coded, which would be more expensive than just getting a couple CRM licenses and connecting it to a web form and outlook. Maybe the scheduling part would be hard for the CRM, I've never looked into that, but outlook calendars have been around forever, so I imagine there's some sort of tool that allows users to pick times they can schedule for a tour based on availability. I've seen that at my doctor's.
For maintenance, how was AI more beneficial than something that allowed users to fill out forms online and get an automated response that someone would be in contact and then sending the request to a maintenance person's email? I work in an industry that provides leasing and mobile repairs for equipment, and we've had tools that did all of this stuff automatically for 20 years. AI chat bots in a CRM might provide a more personalized response, but the efficiency being touted is about reducing human labor, and it sounds like your wife's company's problem was simply solved by hiring more labor rather than using AI tool. If the AI convinced the execs to make that decision, then that's great for your wife, but not sure it's a great sign for AI.
To add, how does having more maintenance requests as a leasing company make business better? It's the opposite of what you would want. Hiring more maintenance staff because now people can complain about broken things easier sounds like they need to do better inspections rather than get an AI tool. You'd also want the emergency maintenance requests that can be solved with a phone call, or a troubleshooting guide by the user or tenant without a maintenance person going out to be solved that way, like check the breaker, check the pilot light, is it plugged in type stuff so you can avoid sending a tech out. Again, this is stuff a lot of CRM software has had available for more than a decade, so not sure why AI is the tool they chose other than it being cheap right now.
Kind of around the same benefits as you. We’ve used it to create product lifestyle image, but majority of imagery needs to be fixed. A lot of the images are also low resolution so we can’t even use them for print.
It’s a nightmare trying to get higher up to understand its limitations when they see random AI video/photos online.
I do a lot of mechanical and chemical engineering stuff. I have find chat gpt and copilot very helpful IF i know what I’m looking for and ask the proper questions. I think before internet andAI the edge was to be able to find information no one else had access to. Now the edge is being able to filter through information overload and ask the right questions.
My wife uses it to cross reference her diagnosis, summarize her reports and patient notes after the 30+ minute pre screening. In it's current form, it's the best "colleague" she's had to bounce ideas off of but needs to scrutinize it's answers due to hallucinating.
Edit: Also forgot it helped her in research with Stats
We use it a little bit to start off new scripts occasionally. It basically does formatting and we write everything else and do it properly. Saves like 20 minutes a fortnight or so.
Apparently that's enough for my company to justify laying off almost 50% of its workforce last year, about 1800 people.
Competent professionals use (or will use) LLMs to increase productivity without reducing of quality but this is not the case for incompetent ones. LLLs also can replace workers in many error-tolerant tasks like first level of customer support, for example. LLMs can be useful where there are repeating patterns in decisions but they can be less competitive than other machine reasoning technologies. Thus, they are useful and will be even more useful in the future but they have exactly zero chances to become general AI. In my opinion, the current AI boom is a regular investor hype, as many other before.
I'm in IT and we've been encouraged to integrate it into our work flow (very large company):
Creation of Knowledge Base Articles - For this it will basically repurpose your notes on a case into a KB. If your notes are crap, the KB is crap, which is usually the case.
Searching of Knowledge Base/Documents - This seems no better than any previous methods of searching. It will hone in on very specific keywords even when inappropriate, so kind of no better/no worse.
Meeting Notes/Recording - This seems useful although I don't use it personally. We already had a way to record but the highlights can be useful for people who's entire role is to be in meetings.
Image Recognition - This is on the customer side so I don't deal with it directly. There are many issues with it, same for other LLMs where they're doing facial recognition and so on.
That's all we get out of it on my side of the house. I don't know how much we've spent on AI, probably a lot from the Executive stuff I see.
IT is an inherently complex domain for AI IMO. A lot of what you do is knowledge based, not text and task based it seems. At least not the most critical stuff like... flipping the giant UPDATE THE NETWORK SECURITY SOFTWARE switch lol
Very true, I think it will be one of the least impacted fields. That runs counter to what most people would assume but then again there's a lot of assumptions about IT and almost all of them are wrong. :P
The real threat is LLM gobbling up the lower support roles, which is how someone like me came up in the ranks. That is a lot of following processes and doing simple tasks, then as you go up things become more complex.
I've certainly seen some of that when working with 3rd Party companies. IT has always been a "net loss" on the balance sheet, not in reality, but balance sheets seem to win.
Doubt it. It's an algorithm that is great at summarizing and doing the boring repeated work that people hate. It is definitely not for what is marketed. Ironically it can effectively replace the low to middle management, who just create spreadsheets, write emails and implement orders from above.
I am sometimes still impressed with LLMs. Recently I ordered a package into a box inside a huge shopping center. Google maps didn't really help much. Before I left home, I wanted to look it up and typed the thing into google, but as it sometimes does, instead of the browser, it fed my query into the LLM. To my surprise, it was able to tell me exactly where to find the box and how to get there, where google maps showed the other side of the center and would make me completely lost. That's a summary again tho, that relies on having data to summarize, data from people. When it doesn't have them, it hallucinates an answer.
So, it does have uses. It can help researchers with pattern seeking, it can shorten repeated, boring tasks, like writing polite emails, but it can't really replace workers. Oh, and what it is also amazing at is translating. Paste a text in any language and it will do a much better job than google translate. But there is no revolution really. It is not and will not be AI as we imagine. It is an algorithm, an impressive one for sure, but still. It's able to pattern seek which words are associated with which other words, so it appears like it "understands," what the word represents, but it doesn't.
In IT, super useful for things I could solve by googling anyway, but it can be quicker.
Notable example is "I want to do X in framework Y that I'm unfamiliar with" and it's all right at scouring docs to give some alternatives. Next step is always reading the real docs though.
Also for small ad-hoc scripts in languages I will never bother memorizing (this means bash). I feel very content in my decision to not bother memorizing Linux commands for super niche use cases, AI just covers that.
And dumb repetitive things I used to use Vim macros for, like one time formatting of some text file to some other format.
How much does this increase my productivity? Eh, not a lot honestly. Definitely not 20%, but it's nice and saves me some mental load for simple things.
Same boat here but I get the sense this is common in system/software admin jobs like ours. Even though my title is "software engineer", I spend most of my time managing/updating/monitoring/fixing several computer networks onsite and in the cloud so I feel you.
Yes. AI is like having an army of junior level thinkers and digitally native servants.
The impact available is really only limited by the employee’s discernment and curiosity. Or is more likely limited by the organizations throttling of those qualities in employees.
No, so far AI is bullshit. CEOs are using AI implementation for PR buzz and to justify downsizing. There are even calculators for implementing it in small ways knowing that the actual work will fall on the remaining employees and how many employees can be fired before you have mass quitting or demands for raises.
This is one reason companies try to make sure employees do not communicate with each other outside of the job. They are terrified that since it has been shown a no office system works, the employees will get together to compete with them as a human run company versus an AI "implementation" company.
30 years ago this happened with phone systems. Having a live human answering a phone every time versus a phone system to answer and redirect calls or delay callers. People preferred human interaction back then and would call the companies with live humans over listening to a machine.
My team was heavily encouraged to begin using AI in our tasks. I tried it out for a bit and then stopped. There was basically no change in efficiency for me. At least not in any noticeable way.
learn how to use it before you get left behind. if your job mostly involves things that can be input into a computer, then an llm can do it 100x better given the right instructions.
"What would the economic impact be if the same investments were made into healthcare? Or affordable housing? Or bringing factories back to America? The debate over AI is complex, and it will take years of using it to known if it's a net positive. In the process we have people being laid off, the abuses of engines like Grok, even military uses. There are so many other investments that we so desperately need."
"When are states like Virginia going to stop subsidizing these data centers by passing energy costs to consumers?"
"This article completely misses the main question. Debt driven spending on AI is propping up the stock market, no big surprise there. The real question is will AI actually justify this debt. Will it actually transform companies, the way the true believers hope. A few articles in other magazines have said, so far… no. That question isn’t even asked here."
There is also a strong sentiment that AI investments are enriching a few at the expense of broader economic stability.
Funny about that AI will make housing worse. AI can be used to oppose or heavily scrutinize new development by analyzing complex, voluminous land-use data, zoning laws, and environmental reports to identify regulatory non-compliance, potential environmental risks, and inconsistencies in proposals. It can analyze planning histories to uncover potential issues.
before now if you want to block the new apartment complex coming up down the street you need to pay a lawyer a fuckload of money....but now you can just get a pro account and let it run.
Funnily enough it's red states that are moving far faster at ripping zoning/permit control out of the hands of municipalities. So you'll probably see an even larger divergence of housing affordability between red v blue states. Sure california has that slew of bills, but those bills only provide work arounds legally costly work arounds and they are loaded with poison pills. While some red states just straight up bash their muni zoning powers straight into oblivion.
This article is being misinterpreted all over the internet.
This article is saying that the idea that all of the spending on AI is boosting US GDP is wrong.
(This article is not about AI productivity).
It it simply saying that the spending US tech companies are doing on AI is actually boosting GDP in Taiwan and Korea.
The analyst from Goldman here did not set out to comment on AIs impact so far on productivity in workplaces. That is a different question. This is about investment in AI and where that boost is being seen in GDP.
The U.S. is essentially acting as a pass-through for capital that is actually boosting the GDP of East Asian manufacturing hubs.
Welcome to the dystopian future where humans have to provide a business case for access to water that, as a scarce resource, could be "better" allocated to data centers that... do... umm... *checks notes* ... ahem... a lot of computations.
Interesting article, though not quite as cut and dry as the title makes it sound. AI being built with foreign chips and foreign servers means that the infrastructure spends aren't really benefiting the US too much. Sounds about right though from this perspective, the chips themselves are agnostic to what they run, it's US models and US companies using the compute from those datacenters. What none of these articles seem to bring up is demand though is there's a reason why datacenters are popping up everywhere, and that's because of demand for compute that is AI driven. So where's the demand coming from if it's a useless technology (per the doomer headlines)? They wouldn't be building them if the demand wasn't there.
Business enterprise AI tooling is taking off gangbusters, all those 'useless AI technologies' are now getting purpose built tools utilizing agents and agentic swarms and are no longer just legacy software with a chatbox bolted on.
Because it’s not in general use… because like all new technology there is a learning curve. Most people are using it like a search engine, to answer questions, or write a paper, resume, email. That’s surface level stuff.
But when connected to additional tools and plugins it can do tasks that replace the need for additional staff, hiring a freelancer or contractor. Computer programming, web design, financial advisor, graphic artist, commercial music, training videos, tutor in any subject. The list goes on and on.
How well it does these things depends on learning to query and develop a plan or path to the desired outcome. And it can even explain to you how to do that.
I actually hate AI because our way of life has allowed people to earn a living in all these fields. But here it is, it’s up to us to learn how to use it or be like the generation before us that struggled to use a PC and mouse.
I had a comment from a client, AI can resolve 90% of the issues customers face but not the 10% remaining, which a human cannot solve who has not acquired practice addressing the 90%, so it is actually less efficient more costly.
Not even close. We have people using it to make slide decks but then manually updating trackers in excel or repeating activities for analysis. They don't even use it to proof documents so typos persist
Goldman Sachs didn't get their infamous sharkiest sharks on wall street reputation by being nice guys who give you good advice.
These are the same mofos who bought into EM early in the 2000 after the tech bubble they helped create popped. And then got an academic/economist to invent the BRICs term and write a research that they used to sell EM investments to their own INVESTMENT CLIENTS who were PAYING them for research/info.
But here's the kicker: Goldman had already bought in PRIOR to that and basically offloaded their bags onto their clients the next decade plus.
While the economic paper's talk about the massive population, resource, and GDP growth potential weren't lies.... the investment returns on the BRIC counties have been dismal since 2000 with China/Brazil peaking in 2007, Russia doing poorly before getting wiped out in 2022, and India lagging for decades.
Now if Goldman were willing to do this to the folks who have their assets with Goldman AND were paying them for research. Then how the fuck do you believe they aren't putting FREE information to max benefit themselves? If anything, this article tells me I should be balls deep in SPY/QQQ because Goldman wants me to sell.
Stunning_Run_7354 | 6 hours ago
I’m curious if anyone has seen AI used effectively in their company.
We have been encouraged to try to incorporate AI into our processes, but almost all the improvements I’ve seen are using AI to summarize meetings, create clip art, and act as an improved search engine. None of these are BAD, but none of these uses has made a noticeable impact either. … well we did have to use more cloud storage for all the extra notes from meetings.
Anyone seen anything near the GAME-CHANGING promises made to the VC and investors?
JoggingGod | 6 hours ago
LLMs have some limited use cases but the AI revolution that's been marketed seems like complete bs. I think most of the AI bubble pops sooner than later. I haven't seen anything indicating otherwise.
thx1138inator | an hour ago
Does software have value? If so, then software that makes software (AI) has value.
blazelet | 5 hours ago
I work in visual effects. I’m using it on the side to speed myself up … if I have a complicated spreadsheet to put together or a GUI I need made around a script I wrote, it’s fantastic for that kind of stuff.
Haven’t seen it used yet in my more complex day to day
mapppa | 2 hours ago
As a software developer, this is exactly my experience as well. It's great for small things that is tedious and has been done thousands of times by different people before. But it falls apart immediately if given any more complex task.
It simply can't keep track of things, and as far as I know that is by design of it being a stateless machine. You only grow the input tokes, and after a while there is so much garbage in the input, that it breaks and starts completely hallucinating, brings up obsolete code, etc.
Even with the small things, you have to babysit every output. They tend to make grave mistakes that can affect a project critically, even though the result is functionally correct. i.e. it switched matrix row/column order midway in some simple functions.
My rule of thumb is: Never ask AI for code that you couldn't write yourself.
Fallout541 | 4 hours ago
I can absolutely say it’s made me a lot more efficient. I’ve seen it make a lot of people worse though because they don’t verify the information it gives.
Chaz_Cheeto | 6 hours ago
Game changing? No. Higher efficiency? Yes.
My wife is a leasing consultant for a property management group. She works with four different properties. She gives tours for apartments, provides customer service, things like that. Her company shifted to using AI last year for tasks. AI is used to schedule tours, field maintenance requests, and create data for decision making.
Since using the programs, they’ve seen a huge increase in tours and maintenance requests. My wife spends less time scheduling, and providing customer service, and more time actually selling the apartments themselves. The skills needed for the job shifted from entry level to being more experienced. Since activity has increased, they’ve hired more maintenance techs and additional leasing consultants to keep up with volume and provide service.
So it’s been a really positive change, but for a while there it was a nightmare because they were not properly staffed to handle the volume at first.
amurgiceblade44 | 5 hours ago
This is a case if AI is being used properly, supporting continuing workers. However in most cases, its being used to replace. Did anyone in your wife's company get laid off due to the switch? If not then it wasn't being used to replace any jobs which is what a lot of worry about AI comes from
kylehatesyou | 4 hours ago
This all seems like something a CRM or something similar would have handled just fine, and is what is used for most situations like this. Why weren't they using something like that, and how were they taking requests before? If it's a small business (four properties doesn't sound like it's too small), I imagine an expensive CRM maybe wasn't cost efficient, but I don't see how AI actually replaced that type of tool at a more cost effective rate, since you'd still need to build a system to manage that type of stuff which would require at least a little application development knowledge, which would mean hiring a dev at least under contract for a minute to make sure it works even if it's just vibe coded, which would be more expensive than just getting a couple CRM licenses and connecting it to a web form and outlook. Maybe the scheduling part would be hard for the CRM, I've never looked into that, but outlook calendars have been around forever, so I imagine there's some sort of tool that allows users to pick times they can schedule for a tour based on availability. I've seen that at my doctor's.
For maintenance, how was AI more beneficial than something that allowed users to fill out forms online and get an automated response that someone would be in contact and then sending the request to a maintenance person's email? I work in an industry that provides leasing and mobile repairs for equipment, and we've had tools that did all of this stuff automatically for 20 years. AI chat bots in a CRM might provide a more personalized response, but the efficiency being touted is about reducing human labor, and it sounds like your wife's company's problem was simply solved by hiring more labor rather than using AI tool. If the AI convinced the execs to make that decision, then that's great for your wife, but not sure it's a great sign for AI.
To add, how does having more maintenance requests as a leasing company make business better? It's the opposite of what you would want. Hiring more maintenance staff because now people can complain about broken things easier sounds like they need to do better inspections rather than get an AI tool. You'd also want the emergency maintenance requests that can be solved with a phone call, or a troubleshooting guide by the user or tenant without a maintenance person going out to be solved that way, like check the breaker, check the pilot light, is it plugged in type stuff so you can avoid sending a tech out. Again, this is stuff a lot of CRM software has had available for more than a decade, so not sure why AI is the tool they chose other than it being cheap right now.
OwnAmbition- | 4 hours ago
Kind of around the same benefits as you. We’ve used it to create product lifestyle image, but majority of imagery needs to be fixed. A lot of the images are also low resolution so we can’t even use them for print.
It’s a nightmare trying to get higher up to understand its limitations when they see random AI video/photos online.
DrNebels | 4 hours ago
I do a lot of mechanical and chemical engineering stuff. I have find chat gpt and copilot very helpful IF i know what I’m looking for and ask the proper questions. I think before internet andAI the edge was to be able to find information no one else had access to. Now the edge is being able to filter through information overload and ask the right questions.
Sad_Resolve_4888 | 4 hours ago
My wife uses it to cross reference her diagnosis, summarize her reports and patient notes after the 30+ minute pre screening. In it's current form, it's the best "colleague" she's had to bounce ideas off of but needs to scrutinize it's answers due to hallucinating.
Edit: Also forgot it helped her in research with Stats
tehifimk2 | 3 hours ago
We use it a little bit to start off new scripts occasionally. It basically does formatting and we write everything else and do it properly. Saves like 20 minutes a fortnight or so.
Apparently that's enough for my company to justify laying off almost 50% of its workforce last year, about 1800 people.
bk7f2 | 3 hours ago
Competent professionals use (or will use) LLMs to increase productivity without reducing of quality but this is not the case for incompetent ones. LLLs also can replace workers in many error-tolerant tasks like first level of customer support, for example. LLMs can be useful where there are repeating patterns in decisions but they can be less competitive than other machine reasoning technologies. Thus, they are useful and will be even more useful in the future but they have exactly zero chances to become general AI. In my opinion, the current AI boom is a regular investor hype, as many other before.
FuklzTheDrnkClwn | 4 hours ago
Call center I used to work at tried to implement an AI service desk. Customers raged until they killed it. I wonder how much money they wasted.
EmoJarsh | 3 hours ago
I'm in IT and we've been encouraged to integrate it into our work flow (very large company):
Creation of Knowledge Base Articles - For this it will basically repurpose your notes on a case into a KB. If your notes are crap, the KB is crap, which is usually the case.
Searching of Knowledge Base/Documents - This seems no better than any previous methods of searching. It will hone in on very specific keywords even when inappropriate, so kind of no better/no worse.
Meeting Notes/Recording - This seems useful although I don't use it personally. We already had a way to record but the highlights can be useful for people who's entire role is to be in meetings.
Image Recognition - This is on the customer side so I don't deal with it directly. There are many issues with it, same for other LLMs where they're doing facial recognition and so on.
That's all we get out of it on my side of the house. I don't know how much we've spent on AI, probably a lot from the Executive stuff I see.
Ok_Addition_356 | 37 minutes ago
IT is an inherently complex domain for AI IMO. A lot of what you do is knowledge based, not text and task based it seems. At least not the most critical stuff like... flipping the giant UPDATE THE NETWORK SECURITY SOFTWARE switch lol
EmoJarsh | 20 minutes ago
Very true, I think it will be one of the least impacted fields. That runs counter to what most people would assume but then again there's a lot of assumptions about IT and almost all of them are wrong. :P
The real threat is LLM gobbling up the lower support roles, which is how someone like me came up in the ranks. That is a lot of following processes and doing simple tasks, then as you go up things become more complex.
I've certainly seen some of that when working with 3rd Party companies. IT has always been a "net loss" on the balance sheet, not in reality, but balance sheets seem to win.
Usakami | 3 hours ago
Doubt it. It's an algorithm that is great at summarizing and doing the boring repeated work that people hate. It is definitely not for what is marketed. Ironically it can effectively replace the low to middle management, who just create spreadsheets, write emails and implement orders from above.
I am sometimes still impressed with LLMs. Recently I ordered a package into a box inside a huge shopping center. Google maps didn't really help much. Before I left home, I wanted to look it up and typed the thing into google, but as it sometimes does, instead of the browser, it fed my query into the LLM. To my surprise, it was able to tell me exactly where to find the box and how to get there, where google maps showed the other side of the center and would make me completely lost. That's a summary again tho, that relies on having data to summarize, data from people. When it doesn't have them, it hallucinates an answer.
So, it does have uses. It can help researchers with pattern seeking, it can shorten repeated, boring tasks, like writing polite emails, but it can't really replace workers. Oh, and what it is also amazing at is translating. Paste a text in any language and it will do a much better job than google translate. But there is no revolution really. It is not and will not be AI as we imagine. It is an algorithm, an impressive one for sure, but still. It's able to pattern seek which words are associated with which other words, so it appears like it "understands," what the word represents, but it doesn't.
yvrev | 2 hours ago
In IT, super useful for things I could solve by googling anyway, but it can be quicker.
Notable example is "I want to do X in framework Y that I'm unfamiliar with" and it's all right at scouring docs to give some alternatives. Next step is always reading the real docs though.
Also for small ad-hoc scripts in languages I will never bother memorizing (this means bash). I feel very content in my decision to not bother memorizing Linux commands for super niche use cases, AI just covers that.
And dumb repetitive things I used to use Vim macros for, like one time formatting of some text file to some other format.
How much does this increase my productivity? Eh, not a lot honestly. Definitely not 20%, but it's nice and saves me some mental load for simple things.
Ok_Addition_356 | 35 minutes ago
Same boat here but I get the sense this is common in system/software admin jobs like ours. Even though my title is "software engineer", I spend most of my time managing/updating/monitoring/fixing several computer networks onsite and in the cloud so I feel you.
osborndesignworks | 24 minutes ago
Yes. AI is like having an army of junior level thinkers and digitally native servants.
The impact available is really only limited by the employee’s discernment and curiosity. Or is more likely limited by the organizations throttling of those qualities in employees.
IceInMyVain | 3 hours ago
AI can be applied to many fields… I think when your talking about it, you’re talking about LLMs.
I work for a huge company and yes, we have developed an app that helps our engineers with a task (it’s basically a sophisticated RAG system)
But the most useful use of AI is using it to detect and classify objects on images/videos. It works great and doesn’t need much ressources.
Optimal-Archer3973 | 3 hours ago
No, so far AI is bullshit. CEOs are using AI implementation for PR buzz and to justify downsizing. There are even calculators for implementing it in small ways knowing that the actual work will fall on the remaining employees and how many employees can be fired before you have mass quitting or demands for raises.
This is one reason companies try to make sure employees do not communicate with each other outside of the job. They are terrified that since it has been shown a no office system works, the employees will get together to compete with them as a human run company versus an AI "implementation" company.
30 years ago this happened with phone systems. Having a live human answering a phone every time versus a phone system to answer and redirect calls or delay callers. People preferred human interaction back then and would call the companies with live humans over listening to a machine.
PotentialAnt9670 | 3 hours ago
My team was heavily encouraged to begin using AI in our tasks. I tried it out for a bit and then stopped. There was basically no change in efficiency for me. At least not in any noticeable way.
jaqueh | 26 minutes ago
learn how to use it before you get left behind. if your job mostly involves things that can be input into a computer, then an llm can do it 100x better given the right instructions.
PotentialAnt9670 | 16 minutes ago
I mean I know how to use it. It's just not very useful or efficient for what I do.
jaqueh | 13 minutes ago
My point is if you don’t find it useful then you’re not using it correctly
PotentialAnt9670 | 8 minutes ago
How do you normally use it in your workflow?
[OP] FXgram_ | 7 hours ago
An article on the same topic in The Washington Post sparked numerous questions among readers: https://www.washingtonpost.com/technology/2026/02/23/ai-economic-growth-gdp-mirage/
"What would the economic impact be if the same investments were made into healthcare? Or affordable housing? Or bringing factories back to America? The debate over AI is complex, and it will take years of using it to known if it's a net positive. In the process we have people being laid off, the abuses of engines like Grok, even military uses. There are so many other investments that we so desperately need."
"When are states like Virginia going to stop subsidizing these data centers by passing energy costs to consumers?"
"This article completely misses the main question. Debt driven spending on AI is propping up the stock market, no big surprise there. The real question is will AI actually justify this debt. Will it actually transform companies, the way the true believers hope. A few articles in other magazines have said, so far… no. That question isn’t even asked here."
There is also a strong sentiment that AI investments are enriching a few at the expense of broader economic stability.
bigGoatCoin | 7 hours ago
> Or affordable housing?
Funny about that AI will make housing worse. AI can be used to oppose or heavily scrutinize new development by analyzing complex, voluminous land-use data, zoning laws, and environmental reports to identify regulatory non-compliance, potential environmental risks, and inconsistencies in proposals. It can analyze planning histories to uncover potential issues.
before now if you want to block the new apartment complex coming up down the street you need to pay a lawyer a fuckload of money....but now you can just get a pro account and let it run.
Funnily enough it's red states that are moving far faster at ripping zoning/permit control out of the hands of municipalities. So you'll probably see an even larger divergence of housing affordability between red v blue states. Sure california has that slew of bills, but those bills only provide work arounds legally costly work arounds and they are loaded with poison pills. While some red states just straight up bash their muni zoning powers straight into oblivion.
ktaktb | 6 hours ago
This article is being misinterpreted all over the internet.
This article is saying that the idea that all of the spending on AI is boosting US GDP is wrong.
(This article is not about AI productivity).
It it simply saying that the spending US tech companies are doing on AI is actually boosting GDP in Taiwan and Korea.
The analyst from Goldman here did not set out to comment on AIs impact so far on productivity in workplaces. That is a different question. This is about investment in AI and where that boost is being seen in GDP.
The U.S. is essentially acting as a pass-through for capital that is actually boosting the GDP of East Asian manufacturing hubs.
themiracy | 7 hours ago
Welcome to the dystopian future where humans have to provide a business case for access to water that, as a scarce resource, could be "better" allocated to data centers that... do... umm... *checks notes* ... ahem... a lot of computations.
jmstallard | 6 hours ago
The following two statements seem contradictory to me, but I lack the expertise to understand the ramifications of, or even verify, the contradiction:
4immati | 6 hours ago
Productivity growth was not driven by AI but investment in AI (data center construction, etc.) proped up the GDP.
Ok_Addition_356 | 33 minutes ago
Not too complicated...
Investors dropping cash on the stock market does not inherently mean industries are growing in sustainable forward-thinking ways.
You can already see the major AI companies starting to feel the heat and fighting to see who's going to survive the pop.
tacs97 | 5 hours ago
As if these billionaires want economic growth. The only growth they want to grow is the appetite for their tech and their own personal bank accounts.
SanDiegoDude | 7 hours ago
Interesting article, though not quite as cut and dry as the title makes it sound. AI being built with foreign chips and foreign servers means that the infrastructure spends aren't really benefiting the US too much. Sounds about right though from this perspective, the chips themselves are agnostic to what they run, it's US models and US companies using the compute from those datacenters. What none of these articles seem to bring up is demand though is there's a reason why datacenters are popping up everywhere, and that's because of demand for compute that is AI driven. So where's the demand coming from if it's a useless technology (per the doomer headlines)? They wouldn't be building them if the demand wasn't there.
Business enterprise AI tooling is taking off gangbusters, all those 'useless AI technologies' are now getting purpose built tools utilizing agents and agentic swarms and are no longer just legacy software with a chatbox bolted on.
128-NotePolyVA | 4 hours ago
Because it’s not in general use… because like all new technology there is a learning curve. Most people are using it like a search engine, to answer questions, or write a paper, resume, email. That’s surface level stuff.
But when connected to additional tools and plugins it can do tasks that replace the need for additional staff, hiring a freelancer or contractor. Computer programming, web design, financial advisor, graphic artist, commercial music, training videos, tutor in any subject. The list goes on and on.
How well it does these things depends on learning to query and develop a plan or path to the desired outcome. And it can even explain to you how to do that.
I actually hate AI because our way of life has allowed people to earn a living in all these fields. But here it is, it’s up to us to learn how to use it or be like the generation before us that struggled to use a PC and mouse.
Active-Car864 | 4 hours ago
I had a comment from a client, AI can resolve 90% of the issues customers face but not the 10% remaining, which a human cannot solve who has not acquired practice addressing the 90%, so it is actually less efficient more costly.
jankyt | 2 hours ago
Not even close. We have people using it to make slide decks but then manually updating trackers in excel or repeating activities for analysis. They don't even use it to proof documents so typos persist
zxc123zxc123 | 17 minutes ago
>Goldman Sachs Says
Out the window it goes then.
Goldman Sachs didn't get their infamous sharkiest sharks on wall street reputation by being nice guys who give you good advice.
These are the same mofos who bought into EM early in the 2000 after the tech bubble they helped create popped. And then got an academic/economist to invent the BRICs term and write a research that they used to sell EM investments to their own INVESTMENT CLIENTS who were PAYING them for research/info.
But here's the kicker: Goldman had already bought in PRIOR to that and basically offloaded their bags onto their clients the next decade plus.
While the economic paper's talk about the massive population, resource, and GDP growth potential weren't lies.... the investment returns on the BRIC counties have been dismal since 2000 with China/Brazil peaking in 2007, Russia doing poorly before getting wiped out in 2022, and India lagging for decades.
Now if Goldman were willing to do this to the folks who have their assets with Goldman AND were paying them for research. Then how the fuck do you believe they aren't putting FREE information to max benefit themselves? If anything, this article tells me I should be balls deep in SPY/QQQ because Goldman wants me to sell.