"The saving of 46 minutes per each Copilot-using employee at LBG is strangely specific"

Because it's made up. They couldn't put out a statement that says "we can't exactly say how much time is saved because we can't accurately measure it". That wouldn't look good when it comes to somebody's appraisal/bonus - which, let's face it, is probably the main driver here (alongside trying to justify the eye-watering amount of money wasted on implementing AI where it is neither warranted or needed just so they can reduce headcount).

I wonder if that '46 minutes' takes into account the time taken to proof read what Copilot generated. They still do that... don't they? DON'T THEY??

"they also noted the AI tools occasionally make mistakes"

s/occasionally/always/

"every occasion" still counts as "occasionally"

Well, to be fair it's only made up at the front end, in the sense that the making up is when employees are asked a loaded question "how much time does Copilot save you each day?" Most don't record their time, and probably don't record their outputs, but what they do know is that "which idiot introduced this time wasting shite?" is not on the list of acceptable answers. So they take a guess at how much time they've saved, and record that, all their guesses (and lies) get totted up and divided by the number of responses, and yeah and verily, it becometh The One Truth that Copilot doeth save the plebs much time.

and half plus one wrote "one hour" while the other half (minus one) "thirty minutes", and so the mean value became 46 minutes.

I'm not convinced they proof read what they produced before Copilot after my experience with their mortgage people.

The worry is with them using CoPout that "computer says no/yes" is seen as immutable and irrevocable.

I remember when MS's ROI calculator for Office 2000 told me my company could save £3M per year in IT cost. Only issue with that was our annual budget was £20k.

46 minutes sounds a sensible number, but does it include toilet/smoke/coffee breaks? That's where the real time is lost.

> toilet/smoke/coffee breaks? That's where the real time is lost.

Time is lost? Aren't those the times, like the proverbial "water cooler chats", when cross-fertilisation occurs* between team members, getting the chance to rub shoulders* with those outside your immediate cubicle mates, boasting camaraderie and increasing company value? Or just a few minutes of peace when the old noodle can do its magic.

From an engineering/development p.o.v. it's often seemed sensible to give your thinkers a number of paid timeslots for taking a good, long, shower, given how many times you hear "the answer just came to me as I was lathering up". Never convinced management of this, though.

* Okay, maybe a bad choice of words with regards to the first of those time-wasters.

your immediate cubicle mates

If you're referring again to toilet/smoke/coffee breaks, I'd wonder where the hell you've been working

Or going home early?

More faffing about and looking like "things are being done" does not equate to productivity. These wankers might be in control of your money and they're playing bot on bot orgy simulator. Cunts.

If people stopped sending emails that needed summarizing they could save even more money.

If I send an email with technical info To technicians and CC manglement, the manglers may want an executive summary: "CYA" would cover most emails.

If I have to resend it To the manglers, I'll tell them why I'm escalating and what I need them to do.

Indeed, but writing concisely for any modestly specific or complex topic is a skill not many people have, and it takes more far more time and thought than rambling on. I know, I'm a world expert at rambling on. As the saying attributed to many different people goes "I would have written a shorter letter, but I did not have the time". I think the French mathematician Pascal gets the ultimate credit for a version of that.

« Je n'ai fait celle-ci plus longue que parce que je n'ai pas eu le loisir de la faire plus courte. »

[1]Les Provinciales XVI (1657)

[1] https://fr.wikisource.org/wiki/Les_Provinciales_(Vall%C3%A9e)/16

Hmmm, maybe I should think about moving my bank account...

Remind me never to back with Lloyds. Any bank that uses a search engine (ie, “AI”) to generate… well, ANYTHING(!!) has no business being in the business

The "golden rule," is to "never use the output without checking it." - so the 46 minutes is lost to checking CoPilot's homework?

Could of course just get rid of a whole tier of middle management whose entire function appears to be preparing reports for other managers to read. Could just get one AI to write the reports and another to read them. No need for endless pointless meeting either.

I once worked for a business that had the leanest middle management of any organisation I've ever worked for. It was a marvellous cover for the incompetence and fraud that led to the bankruptcy of the £200m turnover business. It is certainly possible to have too many middle managers, it is rarely realised that its also possible to have too few.

If a document needs sumarising, then stop spending time putting unnecessary detail in it.

I'd say it depends on the audience.

Techies may well need the full document. It may contain detail pertinent to people at their level.

Middle management need a higher level. A generic idea about the contents of the document.

C-Suite can't cope with anything more than a one-liner. Usually colour coded (because green is good, red is bad) and in a PPT.

So you end up with 3 documents to cover the one subject.

I've been here so many times. I have to write in fine detail if there is a problem (the 5 whys). Middle management just want to know what service(s) were impacted. C-Suite get a PPT with a colour coded box on a 1-pager.

OK, so structure report as

1 line summary

half-page overview

300 pages of detail

Then people can read as far as they want into it.

300 pages of detail.

Then get Co-Pilot to add the half page overview.

"This document contains three hundred pages. Many of those pages include white space and punctuation."

"This page is intentionally blank"

But what happens when 'AI' summarises the 300 page document for middle management, and then that is summarised for top management.

Is any meaning lost? How does anyone know?

"It promotes growth and is very powerful."

I write the key points at the top, then expand on them down below.

Is this the same LBG that is on the hook for £2bn in car finance compensation? That took over HBOS?

I suppose if your business model is based on imagining you can make money out of any old rubbish, then AI must look like a good thing!

" it's about reimagining the entire banking experience. "

Ah. Sorry. Silly me. I now see it is actually banking — easy mistake to make.

Anyone recalls the TV series After Henry (with Prunella Scales) especially the episode where the daughter took Granny to see the French film Betty Blue ? Granny thought from the grandaughter's description that she was seeing a French B a nking movie.

Reimaging the entire banking esperience:

Close most branches

Of those few that remain, make them as user-hostile as possible to deter people from coming back. Avoid providing any useful advice to those who ask for advice

Whatever the issue, the answer is "do it on the app"

Offer dozens and dozens of nearly-identical savings account and credit card options (presumably coming up with all these keeps their staff "busy")

And as we have seen today, make sure that the functioning of "the app" is reliant on some big cloudy provider, so that if they fuck up the whole thing falls over!

The time required to verify every document and email by the people that have to deal with the AI slop is not counted?

Anyway, they could use this to offset the time people lose every day going to their locker to pick up their stuff, finding an empty desk, and setting up their office environment because "flex work spots are more efficient".

46 minutes saved per employee per week means 98 employees can now do the work of 100. No excuse for mass redundancies as yet.

Don't bring your filthy logic here! To the board of Lloyds, that 46 minutes is very early adoption savings, and ample justification to have a huge reorganisation and a round of mass redundancies.

You could have probably done with having the article summarised for you, as they claim 46 minutes daily.

These 46 minutes - how much is that worth to LBG, how much is it currently costing them in fees and, if(!) the fees change*, what is the cutoff when it is no longer worth pursuing?

You know, just some basic facts, the sort of bottom line entries that banks calculate all the time (?!) and they usually like to boast about, to demonstrate to the board/investors/mere customers that they've done the due diligence.

Or do they not have any real numbers? Maybe this is just something Copilot told them was happening?

* say, AI floggers feel like they'd like to makke a profit on the processing costs

Not much based on the salaries for branch staff

"never use the output without checking it."

Given that the amount of effort required to check the outpourings of the stochastic parrots will usually outweigh the supposed time "saved" by using ShatGPT etc. it just isn't going to happen.

In the last week I've received 3 x internal emails that are clearly "AI" (LLM) generated as they bear little or no relationship to the questions being asked and are written in the standard Clippy style of LLMs so I'm just going to ignore them and if anybody complains I'll ask them what checking they did on the results.

Another article citing the "golden rule" of AI assistants, namely to check the output. But if you're not already well versed in the subject matter of the thing you've asked it to help you with, how are you going to be in a position to reliably check its outputs properly? And if you *are* already an expert, why do you need to use an AI assistant in the first place?

Remember when businesses were lucky to have a single phone line that would be answered between 8am and 3pm if nobody else happened to call at the same time and got you a busy signal? Maybe we should go back to that. Parasitic profit-growth-at-all-costs seeping into every crevice of our limited mortal time on this planet isn't a good thing, and workers shouldn't be spending it trying to figure out how to make GenAI more lucrative for investors by juicing the usage statistics.

We pretty much do have that, but the "busy signal" is an automated voice telling you that there's an unprecedented exceptionally high call volume, and your call is important

In the context of 'paper-shifting' commerce (and in academia), beyond a point likely long-since reached, increased 'productivity', i.e. output for a given input resource (in these instances employees), may be intangible or (as with academia) counter-productive. Also, 'productivity' may only correlate with 'profitability' (or social utility) within a discrete range of each variable. Even in the manufacture of physical widgets, production beyond a certain point becomes of questionable worth.

Turning this on its head, one may ask whether increased output in consequence of the fabled 'AI' should be taken as an opportunity to decrease human input. Of course, that is the motivation of people for whom 'profit maximisation' is the key mantra bestowed by business schools. Yet, this can be turned further by deciding to maintain the production cost status quo whilst releasing people from mind-numbing activity.

Suppose, being satisfied by the status quo , communities (e.g. nations) within which commerce and manufacture are embedded, decide to exercise supervisory powers over the antics of business, and demand that for each displaced worker the business must pay to the state the same sum of money it would have given a worker in that particular category. This stream of income can then go into supplementing universal basic income (UBI) enabling all to live with dignity and, as a bonus, unleashing a considerable pool of untapped talent into activities transcending mere commerce and manufacture; the last mentioned being necessary activities, but not inspirational towards mankind's material and cultural betterment.

Setting aside matters of detail concerning how society is recompensed long-term for allowing business to adopt differing working practices, it should be apparent that the self-exalted leaders of big-business would 'get their cake and it'. Although incurring the same direct labour force costs, a host of incidental costs (e.g. large physical environments for work forces) would cease, and their notional worth would be absorbed into profit.

Is this the same Lloyds that were impacted by a single point of failure out at AWS US East 1 this morning?

Let hope that they weren’t using Copilot to configure their cloud deployments, eh?

OK, so we can look at the claimed time saved, equate that to 'saving' two employees per 100 (per. the response earlier), and multiply that up to the full staff count of Lloyds to estimate they could downsize their workforce of 65,000 people by a little over one thousand. A total 'finger in the air' guess is that this could "save" £55 million a year (est. average salary at Lloyds of £55,000, which is probably high), at the cost of the department that has to implement and service the Ai tools, and the subscription costs of those tools.

If every single remaining employee at Lloyds gets a pro subscription to OpenAI, then the cost to save £55 million is... £153 million. Oh dear.

So, OpenAPI would perhaps at best get about £40 million from a large client like Lloyds. If you multiply that out to cover the whole UK white collar workforce (est. 25 million), we could 'save' around £10 billion (average white collar wage in the UK is half what I've estimated at Lloyds), at a cost of £7 billion. All very vague numbers, but it gives us an order of magnitude.

So the total revenue OpenAI might get, if we rolled this out to every single white collar job in the UK, is maybe £7 billion. Every single white collar job. That's revenue, not earnings - someone else can do the maths to work out how much it costs them to service that many users.

We can perhaps draw two conclusions from this:

1) OpenAI are screwed.

2) Re-organising your workforce and processes to avoid 46 minutes 'busy time' each day (maybe cut down on a meeting or two? simplify that email? or just don't send it?) would be far more efficient.

Oh... and this also gives us a base estimate of the productivity gains if we were to roll out AI to every single worker in the UK: about £3 billion.

That's it. Completely turn your entire workforce on its head, make thousands redundant and introduce inaccuracy and uncertainty into every document and process... and you can increase productivity by about £3 billion.

Or 0.1% of GDP.

And then you have to increase taxes in order to pay unemployment benefits to the now redundant workers.

Salary + software licences + other overheads. £55k/employee probably isn't too far off the mark.

Lovely cliché that - "due diligence", but almost universally misunderstood. "Due" in this case means "appropriate and sufficient", so if you palm the decision making off on a machine, your "diligence" is by definition not due, as that would require you to assess the situation, think out what is required to manage it and take responsibility for the outcomes , whether good or bad.

However, I'm not surprised at banks cutting corners like this -- it's not for nothing that the collective noun for them is "wunch".

I've had my main account with Lloyds for 42 years. Nothing has stirred me enough to look around. This morning was only annoying. That they are using copilot is a reason to find an alternative. Anyone got any non AI based suggestions?

Just waiting for the nationwide advert