AI’s Ergonomic Inflation

We’ve been down this road before

In my latest episode of the CoolTimeLife podcast, I look at the often-overlooked productivity paradox of generative AI. Can AI really deliver on its promise of increased efficiency, or are we just creating more work for ourselves? Hint: have you ever heard of the term “induced demand?”

Listen to the podcast at our podcast host, Blubrry.com, or find it on your platform of choice, including iTunes, Spotify, Amazon, Audible, iHeart, and YouTube.

Transcript

I’m going to talk about Generative AI here, but not in the way you might think. There are already enough people spouting on about its numerous abilities and others talking about its threats and dangers, and hopefully people are paying attention to both sides. But one of the great promises of the Gen-AI revolution, especially in the marketing of all these new apps that you never knew you needed, is that it would increase productivity. That’s a wonderful idea. But it’s not true. And this is something that people need to get used to. Technology may have changed – arguably for the better – but we humans have not. That’s going to take more than a planet’s worth of computing power. And if we can’t change, AI might be a zero-sum game. Or worse.

Hello and welcome to CoolTimeLife. I’m Steve Prentice. Each of our CoolTimeLife podcasts focuses on a topic dealing with people, productivity, technology, and work-life, and each offers ideas and facts you need to know about to thrive in today’s busy world. An index of our podcasts is available at cooltimelife.com.

To illustrate my point, let’s first look at the concept called induced demand. When a city discovers it has a problem with traffic congestion – too many cars and trucks on its highways, long traffic jams, and the chaos that comes when even a single lane of a multi-lane highway is blocked – someone eventually gets the bright idea to build another highway. This seems like a wonderful solution, one that is often eagerly lapped up by local politicians as a great way to spend money and reward cronies that will also be gratefully received by the taxpaying public, who will of course, foot the bill.

The problem is, adding new highways doesn’t work. Because as soon as a new highway is built, a whole bunch more cars and trucks appear, and immediately fill it up too, so now you have another jammed highway. Where did all these other cars and trucks come from? Well, they – or more precisely their owners – learned about the new highway and decided to use it. It’s a concept called induced demand, and it happens all the time. The presence of the new highway attracts more people – some may have been using mass transit, some may have been travelling by alternate routes, including the already jammed highways, but now they are spilling onto the new system, and others dutifully follow.

The only way to reduce highway congestion is to have fewer cars, not to build more highways. But that is not a politically popular idea, at least in North America. I live close to an expensive toll highway which runs east to west over the top of Toronto. It is often delightfully empty. Very few cars, almost no trucks, no slowdowns, not even any billboards or buildings. Just a big, wide highway cutting mostly through farmers’ fields. The positive side of this is you can pretty much expect a fast, uneventful ride, allowing you to arrive at your destination at the time you had hoped, or maybe earlier. The downside is you’ll be out the best part of $100 for a return trip.

I once had the opportunity of asking an executive from the company that managed this private highway why the tolls were so expensive, and why shouldn’t it be made free to use, like all the other highways in the area. His reply was quite straightforward. If you remove the tolls, this nice quiet highway would become immediately jammed with cars and trucks. They will just come. And his key point was, even if that were to happen, it would not – could not – alleviate the congestion on the other highways. There will just be more of everything, more cars, more trucks, more highways, more congestion, more delays.

Those who do the simple math will say, a thousand cars divided by two highways means 500 cars on each. That’s good. But they are basing that calculation on an assumption that a thousand is the top number, and that those humans who were put off by the congestion on the first highway won’t be compelled to join the fray. But they will. Induced demand shows that where there were a thousand cars there will now be two thousand cars or more. Humans are messy that way.

Which brings me back to the promise of increased productivity through the use of AI. Without very careful and diligently applied rules of behavior, all this technology – for most end users anyway – has the potential of doing the exact same inflationary thing.

Let me give you another example: email. Email has been around for pretty much 40 years now. But there was a time when email did not exist, and people relied on the postal service and inter-office mail to move information around. When electronic mail came onto the scene, it was not an innovation. It was nothing new in its execution. It was simply an improvement in speed. It still was based on the act of writing a letter, and then sending it. Just like a new highway being built across a city, it was perceived as a faster and better way to do the same thing: send letters back and forth. It was part of the information superhighway. Thanks, Al Gore.

Email was supposed to make things more productive by being super-fast. But what happened? We just started sending more emails. They seemed to cost nothing after all, and in theory we would be able to have a whole email conversation in far less time than it would have taken to have paper letters transported back and forth by the postal service. So, the emails multiplied, and with it came all the time required to compose them, edit them, respond to them, and focus on them when they arrived in the middle of an already busy day.

To this very day, I ask people in the seminars that I run to take stock of the average number of emails they deal with each day. The term “deal with” means: composing new emails, reading emails in the inbox and responding to emails. It also includes emails that have attachments or other tasks, like scheduling a meeting, that have nothing to do with the work that is in front of you, which itself is now getting delayed by the processing of the emails. Some seminar attendees tell me they deal with 30 or 40 emails a day. Some say it’s as high as 150. When I next ask them to tell me the average time required for each email to be processed, even if they say two minutes per message, that adds up to between one hour and six hours a day, spent just on email. This discovery shocks a lot of people, but they don’t deny the numbers. It’s the same thing with calories and liquid assets – in other words, cash. We simply don’t want to, or can’t, keep track of the realities present.

The problem comes from induced demand. Emails beget more emails the same way rabbits beget more rabbits – very quickly. The time-saving benefits of using email were very quickly absorbed by the ergonomic inflation of having more emails to deal with, with a resultant net gain in productivity of zero or less. The irony is, that paper letter that your grandparents had to write out was likely more efficient because it took time to think of the right things to say, and less time overall to resolve the issue that the email was about.

You might say using email is a rather dated example to use, but that’s precisely the point: we’re still using it, and still losing millions of hours of productivity per day to it.

In my courses I teach people how to influence other people to send fewer emails, or to respect periods of time when you are focusing on your work and unable to respond. This is a challenging skill to master. The pressure of emails made it difficult to say no to people’s expectations.

When collaborative technologies like Slack came along, there was a promise of less time spent on emails due to the smoother and less formal style that Slack chats offered. Unfortunately, people simply found themselves overwhelmed by too many chat messages and continued to struggle with the social pressure of turning their Slack channel to the “Do Not Disturb” setting.

It all comes down to a simple reality: if a new technology allows you to do something faster, maybe winning you back two, three, maybe four more hours of productivity per day, that would be great, unless you fill those hours with the exact same activities – not activities that advance a project’s progress, but instead ten more back-and-forth emails, whose existence is simply allowed by the fact there appears to be more time in which to do them, but whose value does not justify that existence.

Clearly AI and Generative AI are in their earliest moments, and their main reasons-for-being have yet to be invented. Almost all of our technological innovations throughout history have been applied to some task we already knew about. The very first telephone was devised to help people hear – the idea of real telephony was too abstract. The first motion pictures were simply there to record Vaudeville stage plays. And of course, web browsers were designed as a big old index of printed material.

So right now, we have a bunch of AI tools with cute names doing things that we used to do manually, like do online research or edit videos, and in many cases, they do help us do these things slightly better. Lots of other people have already written about how helpful ChatGPT and its competitors can be, and this will only get better and more reliable as the technologies develop. But what is less easy to develop is human nature. It is our nature to slide back down the productivity curve. Video chat technologies like Zoom allow us to have meetings without incurring the time and expense to travel to them. This has led to more meetings, but not necessarily more efficiency.

Using ChatGPT to create a first draft of an article is cool – and it’s fun to watch it do it, but then time is required to re-read and rewrite that copy and possibly fact check, grammar check, hallucination check, and further double check that it’s OK to send. Sure, some time is saved in the initial creative process, especially where writer’s block is concerned, but the net-time saved is not as huge as we all expect.

I am not knocking the technology here. I think it has great potential, but I am questioning the human capacity to roll with it and to truly become more efficient. Just like the crowded highway scenario, just because a car can go fast doesn’t mean it will be able to, given speed limits, traffic lights, and congestion on highways and in parking lots. That’s why car commercials always show their models driving on empty city streets or along deserted coastal highways. That’s the promise of an ideal life that most people would love to experience, but in reality, will not – at least for 99.9% of the time they own that vehicle.

It seems so easy to write and then send a message via email, SMS, Slack, or WhatsApp, but do all those messages, along with correcting the typos that voice recognition just couldn’t catch – is all that faster than a quick five minute phone call?

Again, I’m not saying these technologies are bad. They are mostly very good. But the shininess of our toys tends to blind us to what true productivity is. Writing ten messages where one would suffice is not productivity. That’s ergonomic inflation. Allowing back-to-back meetings to be piled up in your calendar by people or by an AI tool is not an improvement. It’s just a freshly laid path to meeting Hell in which participants will remember even less of the meetings than ever. Even if an AI assistant were to say, “slow down there, big fella, give yourself time between these meetings to wrap things up and breathe a little,” most people would overrule those in order to feel they were getting more done, or out of the fear of offending people by saying “no” to their request.

Ultimately, if the technology were able to help everyone get five day’s worth of work done in four days, thus allowing us the much vaunted four-day workweek, what would happen? We would simply redefine the work of five-days-crammed-into-four as the new normal for four days. You would be pressured to return to the office ostensibly to get even more done on Friday. But looking at it from a high-level view, is all of that work true productivity, or just being busy? Busy-work being done because more time has been freed up. In other words, induced demand.

I would wager that even if the four-day workweek were mandated as a law, many individuals would simply look for a second job to work on Fridays as a side hustle – in order to make a little more money. Woe betide them when they discover that income tax laws will render that side-hustle a zero-sum game.

If all of this sounds like a real downer, it doesn’t have to be. It’s not a matter of saying these technologies are useless, it’s just that when it comes to positive terms like “improvement” or “productivity,” we focus a little too much on the technologies and the work, and not enough on human nature.

Maybe it would be a good idea to redefine what productivity is, after all. Is it doing more of the same thing? Like creating more widgets per hour? Or is it about doing a good thing in less time? Or does time have anything to do with it? If it takes ten hours to make a widget, but ten days to make a really superb widget, is that time worth it? There are many memorable expressions that bear this out. “Haste makes waste,” for example. Many a product has been broken or damaged because someone moved too fast and broke it. When days are planned too hastily, and people agree to too many meetings without thinking – that’s not a pathway to efficiency or productivity.

In fact it’s quite an irony that one of the key selling points of current AI-based technologies is the capacity to summarize meetings quickly and efficiently. Sure, that sounds like a great idea, but just like the new highway being built around the city, will the ability to take minutes more effectively simply entice people to spend more time in meetings? Generative AI is seen as the savior of an overworked marketplace, but can it guarantee quality while bringing people to market more quickly?

Redefining what quality is might mean figuring out how to have fewer meetings and how to send fewer emails. It would entail that managers learn how to manage less – to allow working people to get on with the job they were hired to do and to trust them to get it done. It might mean discovering just how much more important listening is, as compared to telling, and how important trust and acknowledgement are in place of oversight and officiousness.

Far too often, when a company finds itself in trouble, the first reaction is to cut costs. And where do they cut costs first? By cutting positions. There is an all-too-common sentiment among some in the higher ranks of an organization, as well as those who are paid to maximize profits, that the people in an organization are merely costly appliances consuming the company’s cash in the form of take-home pay, and that the company’s value and revenue generation comes from somewhere else. These people who are let go as the expression has it, could be or could have been some of the company’s finest talents, if productivity time had been given over to better upskilling and delegation. So, too, the survivors, those who were not fired, at least not yet, can find themselves in a state of perpetual unease, waiting for the next swing of the axe.

And this is where my last analogy comes in. In times of excessive financial inflation, politicians and similarly short-sighted individuals will simply say, “print more money,” which economists know is the worst thing to do. When it comes to technology, it can be very easy to say “use more technology,” and sadly the inflationary pressures render their same results. Perhaps the quality, efficiency, and productivity that people are looking for inside their shiny new AI tools has been with them all along, hidden within the minds of its employees, needing only opportunity and support to grow into something powerfully efficient. I say all of this as someone who loves technology and who has made a career out of helping people understand it. But ultimately, productivity, once properly defined and understood, is a human thing.

Thank you for visiting. Do you have comments or thoughts about this episode? Feel free to get in touch through our Contact page.

Keywords:

AI, artificial intelligence, induced demand, traffic jams, email, Slack, ChatGPT, productivity