Plus Ça Change …

The constant here is us messy humans

Picture This

When looking at the future, even when it seems like everything is changing and all the old certainties have dissolved into a frothing tumult, it’s worth remembering there is one constant. Us. Human behaviour does’t really change. It’s why Shakespeare and Greek plays are still relevant today.

Which brings me to the Powerpoint problem.

I’ve written before about how difficult it used to be to communicate in writing, back before we had computers. Days of back and forth with the typing pool to get what you had written and what got typed on the paper to align. And then time for the physical transport of the letter or memo. And then the same time-consuming process at the other end before you got a reply.

Email truncated all of that into a day, then hours, and now minutes. The result? We are all buried in ‘written’ communication, much of it unnecessary and performative.

Back then, presenting information was even harder. You had to use an overhead projector, and produce ‘overheads’ or ‘slides’, which you placed onto the projector so that they would be displayed on a large screen.

The ‘overheads’ were sheets of clear acetate that you could write and draw upon, or even type on. If you wanted professional looking graphics, they had to be produced by a printer (I mean an actual person, not just a machine), which was expensive and time-consuming. Each ‘overhead’ was the size of an LP (that’s a vinyl disc for you youngsters) and had to be lugged around and sorted into the correct order for your presentation.

It was a pain in the arse and expensive, so you were very careful about how many you produced and when you used them. You front-loaded the process with careful thought and consideration about how to use this scarce resource.

Then along came Powerpoint and anyone could produce slides in minutes and immediately show them on their laptop.

We should have seen where this was going to go. Not just from the email experience, but also from IBM. You see, IBM were the big cheeses of computing back then, and they always had lots of glossily produced ‘overheads’ when they did presentation. But then their computers cost the the annual GDP of a small country back then, so they wanted everything to look reassuringly expensive.

They also had a special word for ‘overheads’, they called them ‘foils’. The quality and number of your ‘foils’ was a mark of your status in IBM, and your expertise at using them gave you great kudos. Your ‘deck’ of ‘foils’ was your measure of virility.

So what happened when Powerpoint democratised all this?

Yes, the volume of slides went up exponentially - and the quality and number of your slides became a status symbol. The more senior you were, the more slick and eye-catching your slides had to be. More and more resources were poured into producing slides, or ‘Powerpoint decks’ as they quickly became known. New roles were created for graphic designers to produce them. They started to include pictures, sounds, video, moving graphics. The decks grew in scale and complexity, more and more resembling film production.

Even at the lower level, where the resources were tighter, people still spent hours jazzing up their powerpoint, vie-ing with each other to have the biggest, whizziest decks. Powerpoint production became a massive time and energy suck.

And so a new phrase was born - Death by Powerpoint. The deck became the place where good ideas went to die. So much effort was put into creating the decks, which were too voluminous for people to go through properly anyway, that no-one got around to developing the idea and producing an actual product or service.

Organisations today are littered with Powerpoint decks, the hard drives and cloud storage systems stuffed with this digital cruft, gumming up the system, blocking out the light and costing money to keep them in their mummified state.

Has communication improved? Are people more thoughtful about what they present? Are organisations more productive, creative and innovative because of Powerpoint?

Are they buggery.

Creating a powerpoint deck has become a replacement for thinking, for sifting idea and reducing them down to their core. Much easier to put your semi-formed thoughts and vaguely related facts into a bunch of bullet points shown landscape on a screen because you think that makes you look smart. What’s worse, other people think it makes you look smart, or at least like your working. And the more slides you have, the more weighty your idea must be, and the harder you must have worked on it, right? (Ed: No!)

Why do the hard work of building something when you can create a powerpoint, and you get to wang on and look important whilst you read out what everyone else can read for themselves? The means has become the work. The artefact is the end product.

So, how do you think AI revolutionising productivity is going to go?

Don’t You Want Me

Well, Navarun Bhattacharya has identified a problem with the deployment of AI that he calls ‘marginal utility collapse’. He points out that whilst AI is being presented as a supply side issue (capacity is coming , it will get cheaper etc.), there’s actually a demand side issue.

Put simply, AI outputs are adding to a flow of content that is already overwhelming us, they are choking a system that is already saturated. As he explains in this LinkedIN post:

‘Within … 5 yrs, LLMs have collapsed the cost of producing “intelligence at work”. You can generate 10 strategies instead of one;50 scenarios instead of five. The supply curve has shifted dramatically outward. But the demand curve hasn’t moved much. Because demand, in this context, is not desire. It is the ability to read, evaluate, understand, prioritize, trust the output, and importantly act. And that capacity is limited, by the human involved.’

That sounds a bit like the Powerpoint problem, doesn’t it?

Navarun posits that those that ‘… companies that will win in this environment will be those that manage marginal utility. They will: generate less, but more deliberately filter aggressively before outputs accumulate; attach ownership to decisions, not just insights; design systems that convert fewer inputs into clearer actions.’

I agree with him. Only I don’t see many organisations that have a good track record of implementing this sort of ‘less is more’ philosophy. In fact, most do the exact opposite.

Although the third annual survey of CEOs by Oliver Wyman Forum and the New York Stock Exchange purports to show how top CEOs are responding to this challenge. It’s not encouraging.

They are reducing junior roles and retaining older employees, particularly middle managers. That’s because they see the need for experience and judgement to be able to utilise or vet AI deployments and outputs. This is characterised as moving from a pyramid structure to a diamond. The obvious comment is that this is not terribly stable. I mean, where does the future middle come from? Oh well, that’s a problem for the next guy, I guess.

The survey also shows that CEOs are increasingly working on plans for the next 12 months or less, so the danger of the diamond toppling over is not even on their time horizon.

The survey additionally says that those who are getting the best ROI from AI deployment, above their expectations, are also redesigning workflows at the fastest rate. This suggests to me that AI is actually forcing organisations to redesign the way they work but also that some of the ROI attributed to AI may, in fact, be coming from reorganisation. Conflating the two is convenient because it justifies the huge expense (much like widespread practice of mis-attributing layoffs to ‘AI efficiencies’ does) but muddies the picture considerably.

We Built This City

But before we move on, let’s take a quick look at the supply side.

We know that suppliers are not charging anywhere near the real cost of their LLM services. Flat rate charging, whether it $20, $200 or £2000 a month, is going to disappear soon. Github CoPilot is moving to usage-based billing in June, so it will be interesting to see how that impacts usage. And those enterprises who have already embedded AI usage into their workflows, sacked the people who used to do the work, will now see a major hike in their costs which they haven’t budgeted for. Ooops.

But that’s not the only supply problem. Whilst there have been lots of announcements of new AI datacentre capacity being built, rather less is actually coming on stream. This is where the rubber hits the road, where the realities of building physical infrastructure drags the gravity-defying bubble of the AI boosters down to earth.

You need people to build stuff. Engineers, electricians, HVAC specialists, construction people, all sorts of trades and specialists. And the USA, where most of these data centres are planned, don’t have enough of them. They don’t train enough of them, either. And now they can’t bring them in from elsewhere because, well, immigration is not exactly favoured by the administration.

This is a major bottleneck and it doesn’t get solved any time soon. And no, I don’t think Elon’s robots are going to save the day. (It’s also a problem in the UK, for pretty much the same reasons. We’ve neglected training tradespeople too.)

That’s not taking into account the need to completely reconfigure the power network and massively expand capacity. And do the same for the water infrastructure.

It seems even AI agents need somewhere to sit. And there aren’t going to be enough virtual desks for some time. If ever.

It’s a buggers muddle, isn’t it?

Power To The People

So the answer to the question, “How will AI impact work in the future?”, is still unclear. Maybe we should ask the question “How much will AI impact work in the future?” as well.

It will be uneven, I think. There are reports from investment and fund management of AIs that can do the work in minutes that would take a team of PhDs several months. That’s highly-skilled, highly paid jobs being displaced. But it would be foolish to extrapolate across all industries and say all high-value jobs are at risk. It’s also clear that translators are being replaced by AI but there’s no shortage of hair dressers and nail technicians in my high street and they seem to have plenty of demand.

And it will certainly create jobs as well. When Excel was released, it was predicted that accountants would be hit hard. The opposite happened and the numbers of accountants grew three-fold (and don’t we know it!). And let’s not forget all those roles for crafting Powerpoint slides that came about.

Michel Zannini, who co-authored ‘Humanocracy’ with Gary Hamel, recently pointed out that many of the systems organisations built to create efficiency have unintentionally created bureaucracy.

That bureaucracy often starts with good intentions, such as creating consistency, reducing risk and improving governance. However, over time, it becomes more complex, it bloats. More layers, more decision gates, more standards to follow. Decision making slows and innovation becomes more and more difficult.

Not because people lack talent or motivation — but because the environment suppresses initiative before it has a chance to grow. This is how bureaucracy marginalises innovation. The people closest to problems often have the best ideas, yet traditional structures frequently distance decision-making from the front line.

This is not a new phenomenon. It’s an age-old problem of large organisations. If you’ve ever worked in one, you’ll know it well. If you’ve ever interacted with one, as we all have, then you’ll know it as the source of the maddening, frustrating and illogical interactions you rant on about when chatting to friends.

AI might address this but if it’s by delegating decisions to its algorithms, I can see that creating plenty of new problems. And I can see many way that AI can make the system worse, not just by gumming it up, as I’ve mentioned above.

But there are ways of addressing it, which are referred to in the book. Since its publication, other examples have emerged. For example, Bayer have reinvented themselves with their ‘Dynamic Shared Ownership’ model, introducing more self-management, fluidity and dynamism into their work organisation. It also makes it a much better place to work, a more human-centric one.

If AI does force organisations to consider how they might reorganise their work, then maybe some will see there are better ways to do it. Not centred around AI, but around the people.

Because organisations are communities of people and always will be.

---

I love to hear from my readers, so please leave a comment or contact me at colin@colinnewlyn.com. Or if you fancy a chat, go to my Calendly page.

You can read more of my sparkling prose at my other substack, ‘Surviving Corporate’