Posts by PinchOfSalt

Jack Dorsey’s fintech outfit Block announces 40% layoffs, blames AI, gets 23% stock bump

Re: CEOs/CFOs are now in their corners drooling

For some this may represent a short term gain. For most, it's just another shift of cost from one opex line to another without any benefit in revenue or margins with a huge risk attached to it that the price for AI will skyrocket once it's been adopted and you're addicted.

As a proof point of this, if you separate the S&P500 into technology providers and technology consumers, the margins for the providers has moved from 17% in 2011 to 29% today. For their customers, they've gone from around 9.5% to 11%. So all those business cases about tech enabling margin increases is BS. What it actually does is shifts the value from the customers to the tech providers.

For others such as PS companies, marketing companies, or anyone heavily reliant on expertise or IP, you have a far worse problem. The chances are that if this AI thing is in any way good, your revenue will shrink faster than you can adapt your business model. Sadly this doesn't take a lot to do.

There are three questions we ask our clients:

Is AI good enough to allow 10% of your customers to not use you any more

Is AI good enough to allow your customers to stop giving you 10% of your work

When will it be good enough to have that effect, based on current improvements in capabilities.

10% shifts is enough to take huge percentages off your valuation, and make it far more difficult to cover loans and will eat into contribution margin that's paying the overall bills of the business.

Of course this might never happen, or might happen in two years. But I've already got clients saying to me that they're losing clients to using AI and worrying about what to do about it.

Always ask about the motives

I wonder whether this is more about other investments that he has than this specific one.

As we've seen recently, it's not that difficult to make the market shift with a change in sentiment due to this sort of announcement, so this could have more to do with him trying to shift focus to the 'AI will crush everything' narrative which will boost his other investments whilst sinking this one in the medium term.

Altman: You think AI is wasted energy? Try raising 100 billion humans

Re: In other words

Perhaps this is the disconnect.

If the goal is just easier, then that's one thing.

But perhaps it should be to make humans happier, which evidence is showing technology tends to not do.

OpenClaw is the most fun I've had with a computer in 50 years

Re: Surely not

Try defining unbiased.

6,000 execs struggle to find the AI productivity boom

Re: The reason seems to be in history repeating itself

I'm not assuming anything really.

Just looking back at history of computerisation in the 70s, 80s & 90s, electrification of factories in the 1800s, The rising use of the Internet post 2000. They all show the same approach.

Stage 1, just give everyone the new thing.

Stage 2, update their existing tools to include the new thing

Stage 3, Re-think how your business works internally so you can take advantage of the thing across a whole business process or processes

Stage 4, Re-think how you interact with clients using this new thing

It's not necessarily sequential, and almost certainly shouldn't be. But the stages are pretty easily identifiable and the reasons that 1 & 2 don't have the desired effect are consistent.

To your point about whether an LLM is the right answer to a given problem is completely fair, and yet more evidence of using AI inappropriately. A little like asking an LLM how many Rs there are in a given word. The useful approach is to ask an LLM to write a script that can determine how many Rs there are in any number of words which it will do very, very well.

In exactly the same way that we don't ask programmers to count how many products are passing through a factory., We ask them to write software that measures this repeatedly.

In two prompts I was presented with a LinkedIn post formatting tool. Literally 3 minutes work. I will never, ever pay for a tool to do that now. But I would not ask an LLM to format my LinkedIn posts as I know it would do so in a non-deterministic manner.

So, LLMs are good at some stuff, but not at all.

The thing I think most people are not really paying attention to is that AI does not need to be perfect to replace humans in a given scenario. It just has to be good enough at a given price point for it to be a viable alternative, given it will work 24x7.

I think we're going to hear the term 'good enough' quite a lot in future.

Re: The reason seems to be in history repeating itself

Thanks for spotting that. I had it in my memory from a different document laid out as A-D but rewrote it here with 1-4 for a less commercial audience.

We ridicule AI today for all it's faults. But this is no more sensible than ridiculing the intern for not knowing how you work, whilst ignoring that they have just arrived with a completely different approach.

I get very frustrated with using AI, but I know that I am not using it correctly from advice I've received from friends. Don't ask it to carry out large tasks. Ask it to write scripts or software to carry out large or sequential tasks. Ask it to verify that it has produced against the original request. And so on.

I caught up with two friends in the last week, one who runs the AI division of a system integrator and another that runs IT programmes for a global enterprise. Both had live examples of enterprise applications based on SaaS being ripped out and replaced with bespoke written software using AI coding tools. Not PoCs, but live, mission critical SaaS applications being decommissioned and replaced.

In one case this took 6 weeks, in the other 16 weeks. In both cases, it was a team of 5/6 that wrote the software.

I'm not fanboy for AI. The impact that this will have is going to be disturbing at the very least and already is. But now the genie is out of the bottle, it's a case of whether you want to be squashed by it or try to eek out your career a bit longer by going with the flow and trying to do our best with it.

The reason seems to be in history repeating itself

I've been looking into this in some detail.

The chances are that most people are doing AI wrong, in the same way that we've always adopted new things inappropriately.

There are 4 approaches that organisations could take:

1, Throw AI licences at employees

2, Buy the latest versions of their existing products with AI embedded in them

3, Redesign their internal processes so that they are oriented towards AI usage to improve it end-to-end

4, Rethink their customer facing experience and either add to it (chatbots, conversational interfaces etc) or empower the existing experience with AI (improved personalisation, better product selection, etc)

I suspect that almost everyone is doing 1 & 2. It's simple, quick, and relatively painless. it's also distracting as people faff about with this new widget without proper instruction on how to use it.

History shows that neither of these produce any measurable results at a business level as it improves a task but not the overall work throughput. For analogies look to the early stages of electrification of factories, early stages of the wide spread use of computers, internet adoption, etc.

C&D is where ROI can be shown, however this is more disruptive and people are less willing to throw cash at big change without big proof of success, which is not yet visible.

BBC bumps telly tax to £180 as Netflix lurks with cheaper tiers

I'm with you.

I'd happily pay the licence fee just to have I'm Sorry I haven't a Clue and The News Quiz on Radio 4.

AI video company arouses fury by boasting about replacing creative jobs

Again, I feel this is the wrong benchmark.

If you want a deterministic outcome, we've known how to do this for coming up a century.

This is not for that and so the comparison is not helpful. Like comparing apples and honey badgers.

You're right in that is is not entirely reliable, but which human is? We hear a lot about the failures of AI simply because it's new, not because it's worse.

Even today I was doing research on the performance of the S&P500 over the last 25 years and it made several errors. Just like an intern would.

I'm in no way saying that I'm liking this outcome. I can already see that despite 35 years of experience in product, operations and commercial management that people are increasingly turning to AI rather than giving me a call. Not because it's better, but it is easier and better than they are on their own.

I remember a fairly heated row I had with a colleague when talking about whether the Internet would take over as the predominant means of corporate networking.

He was absolutely convinced that it would not. Packet loss, deterministic behaviour, guaranteed SLAs for him meant that private networking was insurmountable.

My counter position was that the Internet did not have to be as good as private networking. It just had to be good enough.

And here we are. Private networking is used for edge cases, the Internet has taken over the vast majority.

This is the most likely outcome for AI v humans,

I can in two prompts have myself a LinkedIn formatting tool. I don't have to pay for it, write specifications, employ a developer. I just use it. It's ugly, a little clunky, but it functionally works and having done this, I wouldn't pay for one again.

The same thing has happened to Google. ChatGPT is 'good enough' to stop using the 'standard' way of doing things for many people, despite the shortcomings.

This will happen across all sorts of industries from legal to accounting to marketing to software. Not because it's good, but because it's good enough.

Keep it simple, stupid: Agentic AI tools choke on complexity

Re: "software agents that mimic human decision-making to solve problems in real time"

The key difference is that if you hire the right humans they'll admit that they don't know what they're doing and ask for help.

Palantir CEO claims AI will mean western economies won't need immigration

Economics

So, what he's suggesting is that we don't pay people who then pay tax in our country, spend money in our country and so on and instead pay large corporations who don't.

This does not sound like a good idea at all.

Anthropic CEO: Selling H200s to China is like giving nukes to North Korea

Re: Canada plans on blah blah blah as the years pass

There are many types of war.

Economic is Trumps favourite, but he forgets that he is only able to do that whilst everyone else lets him. That it's everyone else that owns America and can bankrupt the place at a point of their choosing.

As Putin is finding out, it's very difficult to carry out a military war if your economy is in the toilet.

ERP isn't dead yet – but most execs are planning the wake

Surely if AI is so good that you're going to rely on it running chunks of your business, it's also good enough to write an ERP system that will deliver exactly what you want without being tied to some sort cash extraction software provider?

Workday project at Washington University hits $266M

Re: Kerching

I get what you're trying to say here, but it's a bit more. than just that.

Anything this complex will require a programme manager on top of the day-to-day project management.

You'll also need at leat 2 business analysts to document the requirements.

You'll also need a product owner to prioritise the requirements into something coherent.

You'll also need some folk who will operate the system nice live and keep it running 24x7.

Then hosting costs too.

I suspect that your model still comes out cheaper, just not quite so cheap.

Alibaba Cloud can’t deploy servers fast enough to satisfy demand for AI

Demand does not mean there's no bubble

Just because infrastructure is in high demand does not equate to a sustainable business model.

If the folks above you are all scrambling to add users and are eating the loses between what customers will pay and the cost of their underpinning services, you're still going to hit crunch point fairly soon.

LLM based offerings are useful. They're great at a few use cases.

But they're a long way from being worth paying vast sums for, which is what would need to happen for the entire stack to be sustainable.

O2 cranks prices mid-contract, essentially telling customers to like it or lump it

Cats with laser pointers?

Is that like Sharks with freakin' lasers on their heads?

Azure's bad night fuels fresh calls for cloud diversification in Europe

Re: Curious how this gets pastr compliance in big corporations ?

Having been supply side to a bank, the banking regulators have been saying to them for some time that they have to diversify their usage across clouds.

The concentration risk of all banks being on so few providers is a systemic risk, not just a risk to an individual banking provider.

Not sure about that

I doubt this is the case as this is the point of a P&L.

You can spread the capital out over time in the P&L so you don't see those highs and lows.

In my experience, the reason for this shift was quite often that earlier it was cheaper to do with the service than the on-prem option due to adoption oriented pricing. Now people are addicted, the pricing is not so preferential, so it might be a good time to dig out the cost comparison spreadsheets.

I'd also wager that a good number of these migrations to cloud were driven by CIO egos wanting to get this onto their CVs...

UK.gov vows to hack through regulation to get benefit from AI

Choices of task

Why do we insist on inventing tools to replace ourselves rather than augment ourselves to do the things that we find exceptionally difficult?

Using AI to identify ways to manage epilepsy makes sense. It's a problem we simply have not solved. Even using it to look at the brain scans that are done to understand the form of epilepsy since this apparently takes months to do.

But no, we need to use it to write emails (I can only assume that more emails will then be sent), use Sora to create scene from 'It ain't half hot mum' with the actors replaced by babies.

And yes, I'm sure someone somewhere is going the right thing with it, but honestly, as a percentage of workload, it will be miniscule.

Amazon brain drain finally sent AWS down the spout

There is a reason, just not one we like.

They are built like a machine. Each cog has its place and should not move from that place without specific instruction. Thinking is dangerous in such environment for it might break the machine and lead to chaos.

It's effective but not enjoyable. Nor does it allow you to explore and find what you're really good at since you can't try things.

It's what we used to call running the business by the 'idiot principle', where the supreme leader maintains themselves as the most clever person in the business, and treats all others as idiots.

Re: last dns outage I had

What's with the thing against MBAs?

I see it all the time, yet there's never any explanation why this specific qualification is problematic.

Tech vs marketing just seems a low-brow argument. As if tech had any purpose without marketing and marketing (now) without tech.

I do not have an MBA, but have listened to those that do and do not find what they say offensive or worthy or ridicule.

EU biometric border system launch hits inevitable teething problems

Re: suffers teeting problems

Biting the hand that feeds ITeet

Datacenters face rising thirst as Europe dries up

Social problem

What i see here is a market challenge.

We have a scarce resource that has three use cases:

1, Where we use it to drink and keep ourselves clean

2, Where we use it for irrigation so we can feed ourselves

3, Where we can use it for industrial uses, hosting cat videos and other things

I suspect that we have an inverse incentive.

That the users in use case 1 should only be charged the lowest possible rate - its low on the hierarchy of needs after all.

That the users in use case 2 are also in need of lower rates to avoid inflating the price of 2, which is also low on the hierarchy of needs.

The users in use case 3 are by default the highest in the hierarchy of needs and consequently should pay the most. However, they also use vast quantities and therefore demand low prices or subsidies to 'bring jobs to the area'.

These lower prices or subsidies are paid for either by 1 or 2, either in their water prices or in terms of higher taxation to pay for 3.

So we end up in a perverse situation where the least valuable use case gets a preferential rate vs those of the higher value.

One option is to have a 'left over' model. We calculate the amount of water that's required according to the hierarchy of needs at each level and then what's left is passed up to the next level. If there's none left for cat videos, that's decision made. Of course the cat video folks can go talk to the water management companies and ask for more water to be created somehow, but this should have no impact on the current application of other use cases.

Yes, we'd need to have some sort of hierarchy of importance of cat videos vs farming, but that might be a useful exercise in public education. Most people are concerned with current, small, but noisy problems, rather than realising how lucky they are and how good things are.

Social media users rubbish at spotting sneaky ads, say boffins

The unhappiness industry

An old CEO I worked for in the marketing technology space used to describe advertising as the unhappiness industry.

It purports to show you how to be happy, but what it really shows you is all the ways you are inadequate.

Millions of age checks performed as UK Online Safety Act gets rolling

Re: Millions of

An ex chief constable told me that 80% of men watch porn. He was CEO of the charity that did all verification and classification of illegal content, so had a fair amount of knowledge on this. It was a few years ago now though, so might be out of date

Re: Alternatives?

I do agree that technology isn't really the answer, but it's been pretty heavily involved in making money from creating the problem.

Perhaps you could outline your approach to supervising your children whilst they're at school, on their way home from school, the times they're at a friends house, out playing (assuming this is still done)?

The funny thing is that this solution probably has worked.

The younger kids probably aren't able to directly spend money on their phones and are therefore not using VPNs to circumvent the filter.

The adults are now giving some extra cash to the VPN providers to get access to the adult content.

It won't stop everything, but it will almost certainly dampen it down.

Let's see what the traffic numbers show over the next few months.

Re: Alternatives?

Yes, this is a possibility.

Certainly this was the assumption for why a thirteen year old boy hanged himself in his bedroom just down the road from me.

It was assumed that he'd seen the numerous videos being circulated on various platforms, including via messaging systems between children, had encouraged him to try this.

But as you point out, there is no proof as his phone was locked when they found him, and the police couldn't get into it.

It would seem surprising if he hadn't found out via the Internet as where else would he have obtained such an idea at that age and without any previous signs of distress?

Re: There are no good guys

There was a radio 4 comedy show that took a look at the 'behaviour' of TikTok on a brand new phone with a brand new user account with a never used before email account used for verification.

They set the account up as an imaginary 12 year old girl with the most common name for a girl born 12 years before.

The default content that was shown upon opening the app was truly shocking.

Bullying, violence, offensive language were pretty much the first things to be shown.

Searches for content would always bring up something unpleasant and definitely not content you'd sit down with a daughter to watch in any other setting.

If you search for The Naked Week on Radio 4, it's episode 1, starts around 17.30 minutes in.

Re: Alternatives?

Yes, this seems quite nice as a solution, but requires the government to build a system to do it. I guess it could be an extension of an existing system to save a bit on cost. The fundamental difference (I think) with your solution is that it's the govt. doing the checking rather than a private company.

I'm unsure of the way that tax ids are generated, so I'm not certain how easy that would be to abuse. Plenty of places that a tax ID is stored over the years across all those employers that we've all worked for. Pension providers, etc.

I'd be interested to see what the call from the website to the validating platform is doing and how much data is being collected.

It's much more difficult to solve this in this circumstance vs mobile operators due to the multi-tenancy of household networks and no unique identifier to tell one household person from another.

The other option is that the ISP implements the blocking solution, rather than the sites. This has a different set of consequences as the controls would have to be very broad. If the ISP were doing it, it would be easy to use some simple credentials to bypass any blocking system.

Re: Alternatives?

Your definition of effective is being confused with ineffective in this context I sense.

There's sufficient pressure from parents who have seen their children impacted by the negative side of content that I think doing nothing is really not an option.

Alternatives?

Okay, we've all now had a go at the approach that's been put in.

What are the alternative approaches that people can suggest that would be more effective and with fewer downsides?

Re: Does it really matter?

True, unless you find you are dealing with a sieve

Re: Does it really matter?

Thanks for the comments.

I'd suggest having a proper look at what the Internet and it's existing data pools can tell about you.

It's either terrifying or amazing, depending on your point of view.

As someone pointed out, the Nazis were quick to use public records to identify those they wanted to persecute. But, that data pool already exists on Facebook and LinkedIn. Correlate between the two and you'll get a bit more. Then overlay X and Threads, and you'll be able to find most.

If we do end up with an authoritarian government from either end of the spectrum, then they have direct access to the data from censuses, health records, driving records etc anyway.

I'm not suggesting it isn't important. More that we need to look at it cohesively as a data leakage problem, rather than focusing on this one element.

Does it really matter?

I'm not sure this is really all that important.

The average user on the Internet is giving away so much personal information every time they use their computer that this age check thing is just a distraction.

I'd wager that over 80% of people click the 'all cookies' option on every site they visit, never use private browsers or any other form of data leakage prevention, so this tiny bit of extra info is somewhat irrelevant.

Zuck tries to justify AI splurge with talk of 'superintelligence' for all

What's the point?

Someone once said 'Imagine the average person and how dumb they are. Now remember that 50% of people are dumber than that'.

Which leads to some questions...

What will those with lower intelligence gain from this

What will those with average intelligence gain from this

What will those of above average intelligence gain from this

What will those who want to do good achieve with this

What will those with nefarious intent achieve with this

What will those who provide such a thing obtain from this

I accept this is a bit of a venn diagram, but you get the point.

We've already carried out one experiment which hasn't gone so well. Giving people access to the Internet, allowing them to read anything they like, publish what they like, be fed things in an unfettered way, has not led to a more harmonious, happier or more intelligent set of societies.

I'm not entirely sure that repeating this experiment without changing the context is going to be any better.

Meta reveals plan for several multi-gigawatt datacenter clusters

We went from measuring datacenters by area to power a long time ago as it's the primary constraining factor.

It used to be that space was the constraining factor as to how much compute you could squeeze in.

Now it's all about power, so it's sensible to use this as a way of gauging the size of a datacenter.

Compute doesn't really make much sense as each DC will have it's own mix of storage, compute and networking. Also, compute alters with time, so also doesn't create a great benchmark.

China’s trying to slim down, which will fatten the smartwatch market

China has a much stronger coherence of vision than we do, so I suspect that this is less likely to be the case.

Of course the other description for this is that the population is taught / indoctrinated from a young age to believe what the government tells them.

Datacenters have a public image problem, industry confesses to The Reg

Re: Stupid clever people

You are completely right.

When I wrote that I questioned the use of the term nice since it's subjective.

However, I very, very much doubt that people are really complaining about the water usage, etc.

The primary concern that I've seen reported from the reports in the US and UK is that these buildings are huge and disruptive to the locals who live there. They're often pushed through against the wishes of the local people because the economic benefit to the region is greater than the desire to maintain the current local look and feel.

The secondary things that then come up are water usage etc, once they lose the argument about maintaining the local feel of the proposed location.

Stupid clever people

Classic IQ vs EQ problem.

How is it news that everyone wants the nice bits of something without having to suffer the consequences of that nice thing?

I want decent roads, but I don't want to pay the tax to make them decent.

I want a stable electricity supply, but I don't want any power stations or distribution network disturbing my view.

We live in a world where the gap in understanding between production and consumption is so vast that I'm bewildered that supposedly clever people are surprised.

Adobe turns subscription screw again, telling users to pay up or downgrade

Really?

A single $56 subscription allows you to generate $10k+?

Most creative businesses have a net margin of around 8-10% once overheads and directors salaries are properly accounted for (ie not hidden in dividends).

What sort of services are you delivering that allows for a single person to be generating more than $10k per month with just this software suite?

I'm not arguing that you don't feel it's good value, for you it might well be. But when you're in most creative agencies, subscriptions for software is a meaningful chunk of your cost base and increasing far faster than the rate card can.

Sci-fi author Neal Stephenson wants AIs fighting AIs so those most fit to live with us survive

AI vs AI battles

I'm not sure about this.

I'm less concerned about some sort of revolution of AIs taking over the world.

I'm more concerned that the use of AI will cause us to make ourselves extinct through boredom or lack of interest in other people and reality.

Anthropic’s law firm throws Claude under the bus over citation errors in court filing

Error correction

At a previous employer we were using forms of AI and other methods to validate that marketing content complied with client brand guidelines.

It was actually reasonably good at this and better than humans.

The reason for this is that it didn't get bored and it was just as adept at spotting patterns as anti-patterns, whereas humans are only really good at pattern matching and very quickly tired of trying to identify anti-patterns as it's very intensive brain work.

So, having AI draft something and then asking a human to proof read it is not exactly ideal.

There is of course no corresponding data to show how many times citations are incorrectly made in court through pure human error as that's not very newsworthy, so before we through this under the bus, probably worth understanding that.

Trump ends Biden-era dream to cap US AI chip exports

Copyright infringement anyone?

I'm confused.

Either there's a desire for all information to be freely available for the improvement of AI, or it isn't.

Why is NVidia getting a shout out here vs authors?

One can only wonder...

The Telegraph jumps the gun on World War III

Process flow

As I recall, there was much fanfare about reducing the number of steps to publish from 50 odd to 17.

Sounds like it now has too few steps.

AI models routinely lie when honesty conflicts with their goals

Conflicts of interest

I'm not sure the term lying is entirely fair.

There are two basic measures at play here:

1, Being factually accurate

2, Be likable to all users

These two things are in obvious conflict.

What we're seeing is the various developers and trainers playing with the balance between the two. I wouldn't therefore call it lying.

As a result, I don't feel this is a design flaw. It's a design feature.

Procter & Gamble study finds AI could help make Pringles tastier, spice up Old Spice, sharpen Gillette

Game changer?

I'm not sure if this is a challenge of the the summarisation of their findings, but in the main body it talks of 'being comparable to working with another human' and finishes with 'this is a game changer'.

Nothing game changing about being able to do what we can already do and use far less energy and water.

On the issue of AI copyright, Blair Institute favors tech bros over Cool Britannia

What's the point?

If we take a step back for a moment, we have a sustainability problem.

Today, a large part of the reason for creativity is money making. Whether you write a book, paint, or produce adverts.

The reason it makes money for you is that you have a reasonable expectation that people consuming your works pay for them.

If we remove that motivation, then who is going to continue to bother writing books, painting or producing adverts?

If people stop doing this because they don't get paid, then what are these AI tools going to get trained on?

The simplest possible solution in the short term could be paywalls for valuable content. Let the bots read the simplest possible versions for free, but anything of value needs a payment. This is how the record industry deals with it and it seems pretty sensible and scalable.

Asda's tech separation from Walmart nears £1B as delays mount

Are they also working for Birmingham City council?

Just wondering...

Oracle Cloud says it's not true someone broke into its login servers and stole data

Re: Encrypted passwords?

We use compute power to make AI cat videos.

There's more than enough compute lying around to do this.