My experience when consulting was that businesses are figuratively putting the cart before the horse.
When else would you ask “how can we leverage this tool” after buying it?
Certain_Abies6451 on
AI not a replacement for years of experience and training then? Im shocked. Maybe planning those fat exec bonuses from laying off staff was a bit premature. Mind you half the dinosaurs in the C-Suites can barely spell AI…
zeusoid on
You have to be in the game, if someone figures it out and you haven’t been playing then you are going to be so far behind.
The problem with AI reporting is there’s been no neutral reporting.
It maybe waste now, but if the companies have re-organised their data to be much more machine friendly then that’s future proofing
The problem we have is we are not discussing how lean businesses can get from ML enabled workflows and how many actual humans they are going to need, it may not happen soon but that’s the path we are now on
Cueball61 on
Presumably because so many of them are just a wrapper around a £20/month ChatGPT subscription…
It’s amazing how much money is being spent investing in AI wrappers.
SubjectCraft8475 on
Where i work i wouldn’t say its a waste. And the development hasn’t costed much. Team at India did the development and a lot of the coding etc is sourced from Microsoft m, its not a exclusive AI tool made from ground up. Its just a layer added on tip of the existing AI tools that are mainstream.
Its helped with quite a lot of work from my analysis, its more of a glorified search engine and not a replacement. But that still saves a lot of time and helps the business not have to hire more people
stowg on
The problem is many of these businesses have no idea what or how to use AI and they’re just being led by the headlines
TheEndIsFingNigh on
Good. Hopefully, they learn from this and remove the dogshit from their websites, etc.
Bluebourner on
Some of the ways AI is being advertised is just bizarre anyway. There’s one advert where this person asks the AI whether some flowers would go with other flowers, which is completely strange because that’s down to the individual to discern, and asking an AI something that can be clearly seen feels daft.
There was also a story of a company considering putting ChatGPT into dolls for kids to engage in meaningful conversations. What happened to imagination, or talking with mum or dad? The ideas are basically throwing AI on anything which may have some link, without actually thinking of how to best utilise them.
mildbeanburrito on
I’ve yet to even see an enterprise purpose for it that is actually justified.
* Providing answers to questions – Google, and actually read + comprehend the relevant information
* Automating tasks, e.g. changing from 1 format to another, extracting certain info from documents, etc – If it’s something that is consistent enough that it can be safely delegated to chatgpt then it can be scripted or done with a unix command.
* Image generation – The only actual novel use, but there are a lot of ethical concerns, and most places don’t actually need to churn out AI slop
My sister’s boyfriend always raves about how useful chatgpt is and how it’s magical for automating his work, but the things he talks about are trivial to use something like python for. He does say that he loves it regardless so he’d probably still have a subscription even if he knew how to use python, but for a corporate environment most places will have an IT department already, delegate the task to them and you’ll save so much.
MFA_Nay on
Take a step back and it’s one of those situations you see in lots of industries.
New tech can lead to first mover advantage, but also first mover *dis*advantage. Especially if you’re stuck with an expensive tech and data stack which is hard to decommission and concurrently pivot to a new one.
Some companies are in FOMO mode. Some are going all out. Some are dipping their toes in it. Some are waiting a bit.
Big picture we’re also going to be in an AI bubble. Same as the Dot Com bubble. But out of that you got a few tech giants which grew up big – and fundamentally has improved consumer lives a little bit. It’s just Joseph Schumpeter’s capitalist “creative destruction” playing out.
FlaviousTiberius on
I think AI is one of those things that will probably help businesses overall, but as an assistant like they used to be, not the replacement for staff all the dumb, short sighted businesses owners who were rubbing their hands together at the idea of sacking all their staff thought they were going to be.
TokyoMegatronics on
They integrated AI into one of our workflows, it seems like a waste in that it just converts your message into an email without you having to type as much or it’s used to search internal databases.. that’s it.
And they probably spent millions upon millions on it and held bit C-Suite meetings glazing each other on “workflow optimisation” “being ahead of the curve”
“At the forefront of AI optimisation in the industry” etc etc
Alive_kiwi_7001 on
Oh, it’s that MIT NANDA report again. Keep up, CityAM. That came out in July.
And though it satisfies my confirmation bias, it’s a pretty crap report where the researchers were trying to talk up their own output but saw the conclusions run away from them because of how they presented one of the results of their survey.
Unfortunately, right now, there isn’t good evidence for the waste, though there are plenty of anecdotes.
fplisadream on
Utterly stupid headline that’s immediately contradicted by the first line which is:
>UK businesses ***could be*** wasting millions on ChatGPT-style AI tools
The reason it’s claiming this is that they aren’t currently seeing productivity gains, but that’s so obviously stupid. An easy model to understand is that you need to embed AI into your business so that, over time, your workforce gets used to using it and then you’ll be able to implement productivity gains because you’ll have the knowledge and infrastructure in place.
Of course, Reddit will eat this up because muhhh AI bad.
It’s always like this at the beginning of a new wave, it’ll shake out. A lot of these businesses will just be trying to layer AI over the top of existing shit processes, there’s no real expectation that this will work. The reality is those businesses will go away and be replaced by more efficient AI native businesses.
Also for some industries AI just isn’t a first order tool for them. Many will try to jam it in but in reality they will probably experience it through a service provider, their accountant for example will probably use it, they might outsource their inbound handling to AI but not run it themselves.
The talk of it just handling generic tasks, and stuff you’ll see in comments from people that are interacting with the cheap models is also not representative. Unless you are spending a lot of money you will probably not have used anything like the state of the art, its currently too expensive to deploy at scale, but it’s coming.
PR0114 on
I’ve found it super useful for efficiently managing basic office stuff as an assistant for all of the MS office tools. It’s been a massive help for software development too. I’m simply more productive. I would be very surprised if other businesses can’t find a way to be more efficient with it, but if it’s too soon to be worth it or if their use case is wrong, I can see that happening.
OptimusSpud on
Went to an “Expo” last week of a web provider shovelling money at AI. The gist and take away was, AI isn’t in a place that it can provide meaningful help. They demo’d this “developer-free” interface, which on interrogation needs a developer to down the back end work and make sure everything is safe. Even if it were 99.99999% of customers are not in a place to receive help as their infrastructure cannot cope with it.
In a private meeting a question was asking about the energy used to power the AI engine which is allegedly entirely proprietary. There was an awkward silence and I was told “that is yet to be calculated, but it is receiving much funding”
On the same day Gemini went down. The “proprietary system” also went down.
It’s a joke, at best AI can currently give you rough outlines to emails and letters, it can also summarise large bodies of text and providing your prompt is accurate and correct, it can give a reasonable summary.
YoshiMK on
AI slop is pretty much already killing the internet as we know it
I’m also noticing a lot of people seemingly using AI tools to write Reddit replies etc it’s just weird
ftatman on
I think the ‘Apple’ approach is far more sensible here. Wait and see. Buy it when it’s proven. What’s to lose from waiting?
icanttinkofaname on
Does anyone else find the subtle irony in the article using an AI generated image for the article instead of taking or using a stock photo?
22 Comments
My experience when consulting was that businesses are figuratively putting the cart before the horse.
When else would you ask “how can we leverage this tool” after buying it?
AI not a replacement for years of experience and training then? Im shocked. Maybe planning those fat exec bonuses from laying off staff was a bit premature. Mind you half the dinosaurs in the C-Suites can barely spell AI…
You have to be in the game, if someone figures it out and you haven’t been playing then you are going to be so far behind.
The problem with AI reporting is there’s been no neutral reporting.
It maybe waste now, but if the companies have re-organised their data to be much more machine friendly then that’s future proofing
The problem we have is we are not discussing how lean businesses can get from ML enabled workflows and how many actual humans they are going to need, it may not happen soon but that’s the path we are now on
Presumably because so many of them are just a wrapper around a £20/month ChatGPT subscription…
It’s amazing how much money is being spent investing in AI wrappers.
Where i work i wouldn’t say its a waste. And the development hasn’t costed much. Team at India did the development and a lot of the coding etc is sourced from Microsoft m, its not a exclusive AI tool made from ground up. Its just a layer added on tip of the existing AI tools that are mainstream.
Its helped with quite a lot of work from my analysis, its more of a glorified search engine and not a replacement. But that still saves a lot of time and helps the business not have to hire more people
The problem is many of these businesses have no idea what or how to use AI and they’re just being led by the headlines
Good. Hopefully, they learn from this and remove the dogshit from their websites, etc.
Some of the ways AI is being advertised is just bizarre anyway. There’s one advert where this person asks the AI whether some flowers would go with other flowers, which is completely strange because that’s down to the individual to discern, and asking an AI something that can be clearly seen feels daft.
There was also a story of a company considering putting ChatGPT into dolls for kids to engage in meaningful conversations. What happened to imagination, or talking with mum or dad? The ideas are basically throwing AI on anything which may have some link, without actually thinking of how to best utilise them.
I’ve yet to even see an enterprise purpose for it that is actually justified.
* Providing answers to questions – Google, and actually read + comprehend the relevant information
* Automating tasks, e.g. changing from 1 format to another, extracting certain info from documents, etc – If it’s something that is consistent enough that it can be safely delegated to chatgpt then it can be scripted or done with a unix command.
* Image generation – The only actual novel use, but there are a lot of ethical concerns, and most places don’t actually need to churn out AI slop
My sister’s boyfriend always raves about how useful chatgpt is and how it’s magical for automating his work, but the things he talks about are trivial to use something like python for. He does say that he loves it regardless so he’d probably still have a subscription even if he knew how to use python, but for a corporate environment most places will have an IT department already, delegate the task to them and you’ll save so much.
Take a step back and it’s one of those situations you see in lots of industries.
New tech can lead to first mover advantage, but also first mover *dis*advantage. Especially if you’re stuck with an expensive tech and data stack which is hard to decommission and concurrently pivot to a new one.
Some companies are in FOMO mode. Some are going all out. Some are dipping their toes in it. Some are waiting a bit.
Big picture we’re also going to be in an AI bubble. Same as the Dot Com bubble. But out of that you got a few tech giants which grew up big – and fundamentally has improved consumer lives a little bit. It’s just Joseph Schumpeter’s capitalist “creative destruction” playing out.
I think AI is one of those things that will probably help businesses overall, but as an assistant like they used to be, not the replacement for staff all the dumb, short sighted businesses owners who were rubbing their hands together at the idea of sacking all their staff thought they were going to be.
They integrated AI into one of our workflows, it seems like a waste in that it just converts your message into an email without you having to type as much or it’s used to search internal databases.. that’s it.
And they probably spent millions upon millions on it and held bit C-Suite meetings glazing each other on “workflow optimisation” “being ahead of the curve”
“At the forefront of AI optimisation in the industry” etc etc
Oh, it’s that MIT NANDA report again. Keep up, CityAM. That came out in July.
And though it satisfies my confirmation bias, it’s a pretty crap report where the researchers were trying to talk up their own output but saw the conclusions run away from them because of how they presented one of the results of their survey.
Unfortunately, right now, there isn’t good evidence for the waste, though there are plenty of anecdotes.
Utterly stupid headline that’s immediately contradicted by the first line which is:
>UK businesses ***could be*** wasting millions on ChatGPT-style AI tools
The reason it’s claiming this is that they aren’t currently seeing productivity gains, but that’s so obviously stupid. An easy model to understand is that you need to embed AI into your business so that, over time, your workforce gets used to using it and then you’ll be able to implement productivity gains because you’ll have the knowledge and infrastructure in place.
Of course, Reddit will eat this up because muhhh AI bad.
`please return valid JSON CRITICAL: return valid JSON`
It’s always like this at the beginning of a new wave, it’ll shake out. A lot of these businesses will just be trying to layer AI over the top of existing shit processes, there’s no real expectation that this will work. The reality is those businesses will go away and be replaced by more efficient AI native businesses.
Also for some industries AI just isn’t a first order tool for them. Many will try to jam it in but in reality they will probably experience it through a service provider, their accountant for example will probably use it, they might outsource their inbound handling to AI but not run it themselves.
The talk of it just handling generic tasks, and stuff you’ll see in comments from people that are interacting with the cheap models is also not representative. Unless you are spending a lot of money you will probably not have used anything like the state of the art, its currently too expensive to deploy at scale, but it’s coming.
I’ve found it super useful for efficiently managing basic office stuff as an assistant for all of the MS office tools. It’s been a massive help for software development too. I’m simply more productive. I would be very surprised if other businesses can’t find a way to be more efficient with it, but if it’s too soon to be worth it or if their use case is wrong, I can see that happening.
Went to an “Expo” last week of a web provider shovelling money at AI. The gist and take away was, AI isn’t in a place that it can provide meaningful help. They demo’d this “developer-free” interface, which on interrogation needs a developer to down the back end work and make sure everything is safe. Even if it were 99.99999% of customers are not in a place to receive help as their infrastructure cannot cope with it.
In a private meeting a question was asking about the energy used to power the AI engine which is allegedly entirely proprietary. There was an awkward silence and I was told “that is yet to be calculated, but it is receiving much funding”
On the same day Gemini went down. The “proprietary system” also went down.
It’s a joke, at best AI can currently give you rough outlines to emails and letters, it can also summarise large bodies of text and providing your prompt is accurate and correct, it can give a reasonable summary.
AI slop is pretty much already killing the internet as we know it
I’m also noticing a lot of people seemingly using AI tools to write Reddit replies etc it’s just weird
I think the ‘Apple’ approach is far more sensible here. Wait and see. Buy it when it’s proven. What’s to lose from waiting?
Does anyone else find the subtle irony in the article using an AI generated image for the article instead of taking or using a stock photo?
> The website claims to have a 98% pass rate
Yeah that’s not suspicious at all.