Over half of UK businesses who replaced workers with AI regret their decision

It ain’t there yet. It’s got pretty good at fighting bigots on Reddit though

Ai is good for getting rid of your boss. It’s perfect for scheduling budgets etc, wait until people start to realize this.

Lol, no they don’t.

Minimum wage is £12.21 an hour.

If you employ ten people to work in your customer service chats and need them to work Monday 9-5 but because it’s a bank Holiday you have to pay them time and a half.

That’s £18.31ph.

For an 8 hour shift that’s £146.52 per day per employee.

Ten employees is £1465.20 each bank holiday in wages.

The normal rate is £12.21ph 40 hours a week, 52 weeks a year is £25,396 per employee for 10 staff is over quarter of a million pounds per year just in wages, not including holidays, sick pay, pension contributions, NI and all the stress that comes from being an employer.

Or a chat ai that costs you £50,000 a year to run, it’s not as good but talks to 1000x more people, works 24/7 never calls in sick. Never comes to work in a mood or hung over. Never asks for a payrise or employment rights .

Big fucking surprise. Bet they didn’t even try testing the two alongside each other objectively.

This article has very little information. It surprises me how easy it is to trick people.

These companies ‘studies’ are often made with a commercial purpose in mind. Wait for a reputable source, one that doesn’t profit from such news.

My company is pushing AI so hard and are monitoring usage as part of performance reviews (read: using it as a means to reduce the workforce).

If someone leaves, you can’t backfill unless you can prove AI can’t do it. If you want to hire, you need to prove it can’t be done by AI.

I use it when I feel it’s necessary, but the forced nature of this really doesn’t sit well.

It can give shitty answers, make stuff up, and has meant that junior staff members don’t feel the need to learn anything so they can just “AsK ChAtGpT” so I’m constantly having to fix errors when the AI gets it wrong.

I have two concerns about this, one is employers being allowed to force its usage and effectively disciplining those who don’t based on moral and ethical concerns, and the amount of jobs losses it’s going to create while concentrating wealth in the hands of the companies who own the AI.

We’re heading full speed towards tech feudalism and a massive crisis for humanity unless our entire system changes to guaranteed incomes for people.

These models are tools, not replacements (at least for now).
They will replace people because suddenly the effective worker will need less colleagues.

The key issue is that we still haven’t invented AI. LLMs aren’t AI, they can’t think, decide, or interpret anything. Their responses are statistically likely strings of words to a given input.

Computer scientists used one too many human analogies and Marketing just went wild.

Well if they want their employees back they can reimburse them for lost wages with inflation and the new wage agreement thats higher than their prevuous wages all taken into account othereise what guarantee do they have they wont be laid off again?

Talk is cheap, put your money where your mouth is.

As a data scientist LLMs (AI is a broad term, chatgpt is a large language model) is a great tool, it can improve your efficiency if you have the domain knowledge. However, it cannot replace people, the architecture of LLMs will not replace us, even in 100 years time.

I believe, there is currently an “AI bubble”

Men will tell you AI is the most innovative thing of the past twenty years and then just use it to draw themselves as an action figure boxset.

“we are happy to announce we are early adopters of ai…”

Another wave of redundancies just so the CEO can keep the shareholders happy and hit their own bonuses.

I used to work for a company which was bought out by a larger company and the moment they started saying “don’t worry your job is safe” I knew shit would hit the fan sooner or later. Survived three rounds of redundancies, the last one was due to them implementing ai on a large scale. Customers are livid as they can no longer speak to a human being and if they do, it’s someone from India. Life was better before the takeover.

AI marketing teams are snake oil salesmen.

Over promise and under deliver leaving chaos in their wake.

A director where I work had some crazy ideas about using AI but I shot them down pretty quickly by explaining the severe limitations of LLM’s and the considerable risks it posed to the business when it inevitably fucked up.

Upper management get convinced that a technology is wonderful by a sales person over a boozy lunch and buy it without understanding it. The people who actually do the job hate the thing. It causes loads of problems because it’s crap and doesn’t fit well with what the company does. I’m shocked. I’ve never seen that happen before </s>

The downsides of new tech never get measured unless the failure is so massive that it can’t be ignored. It’s like our company buying SAP to do ordering. Someone got a big pat on the back for getting rid of the entry level accounting people who could process a purchase order in 10 minutes. We now have engineers on an hourly rate of five times what the accounts juniors got taking 4 times as long to do the same job but that 20x cost is never measured.

Most people don’t realise that LLM’s aren’t actually intelligent. They’re just very clever pattern matching engines that are very good at predicting what words are coming next. In our training, we learned that the hardest and longest part of setting up an LLM is curating and cleaning the training data. The internet is full of crap. If an LLM is set loose to self train online, it will interest junk and spout junk. The less care that’s taken training it, the worse its performance will be. There are some things it’s good at but many people haven’t heard of Searle’s Chinese Room Argument and don’t realise that’s what an LLM is doing

I honestly do not know how CEOs appear. Is it just from being the biggest ball-licking fellatio expert until a position open?

Tbf, there are a ton of things AI /ML can be used on. Do you write a lengthy email? Quickly brush with ai over it to make it more appealing. Do you have too many tickets/ emails? Try filtering by using ML.

Here we also have a problem. AI should be used to prevent or lessen the workload of colleagues by cutting out steps and giving them their tasks only they can do. You don’t let the colleagues use it all the time but give the dev team the task to implement it onto certain steps.

For example: we still sometimes get faxes from doctors and older folk. We already made it so it would be shown as a pdf in a ticket. Now, we also use AI to tell us what it is without opening it, and with some combination of ocr and AI, it writes automatically a comment directly into the ticket, what it is about and the subject. Depending on what it is, we can also either ignore it or give more information. There is no need to open the pdf anymore.

Maaaaany such small things are perfect for AI and make your day 10x times more productive. Here 1 minute, there 20 seconds, etc. In a day, it can save some good 30 minutes.

Oh yeah, the AI first thing, etc. Is stupid.

I’m sure for some stuff it’s useful.. but you could see this coming a mile away. AI customer service bots for the most part are awful..

We have tried using them for a few small things at work but you still need a human to verify and check the output which also takes time.

Yes it can learn and adapt but it’s not good enough (at least currently) to not need the human element for some tasks – and I think in terms of writing and imagery it’s still very clear when something has been written say by AI

Because it’s also learning from humans or pulling human into it can be wrong. And wrong a lot

Interesting to see how labour things that it will get the disabled and normal unemployed into jobs that don’t exist.