Latest / In The Loop / Here's why AI is making us work more, not less.

Transcript

Here's why AI is making us work more, not less. · In The Loop

• 0:00
  AI was supposed to make us more productive, which was supposed
• 0:04
  to. Mean working less.
• 0:06
  Yeah, a new study from UC Berkeley put researchers inside
• 0:11
  a tech company for eight months and found, well, the complete
• 0:15
  opposite. People using AI tools worked
• 0:17
  more, not less. And to be honest, this mirrors
• 0:21
  exactly. My experience and our.
• 0:24
  Company's experience, too The study found that people took on
• 0:27
  broader responsibilities, the line between work and rest.
• 0:31
  Blurred. Very.
• 0:32
  Fast and filled up every single. Minute with.
• 0:35
  More and more tasks nobody told them to.
• 0:39
  The tools just made it easy enough that stopping felt like a
• 0:42
  waste of time. Today I'm covering exactly what
• 0:44
  researchers found, why this pattern has repeated with every
• 0:48
  major labour saving technology from the past 100 years, and
• 0:52
  what sociologist theory of social acceleration.
• 0:55
  Tells us about. Why Productivity Tools just
• 0:58
  never seem to produce that spare.
• 1:00
  Leisure time that they seem to promise.
• 1:03
  This February, 20262 researchers at
• 1:24
  Berkeley's High School of Business.
• 1:27
  Published a piece in Harvard Business Review with the title
• 1:29
  that. Well, it tells.
• 1:30
  You a lot of things it says AI doesn't reduce work, it
• 1:33
  intensifies it. They'd spent eight months inside
• 1:36
  a. 200. Person tech company.
• 1:37
  Watching how people actually use.
• 1:40
  AI tools in their day-to-day work.
• 1:42
  So not in a lab, not a self reported study.
• 1:46
  They were on site twice a week from April to December 2025.
• 1:50
  And they did more than. 40 like in depth interviews across
• 1:53
  engineering and product. Design research operations.
• 1:58
  And they found. 3 distinct patterns.
• 2:00
  Of work intensification. And all three of them are
• 2:03
  actually voluntary and nobody was being forced to do more.
• 2:06
  So let me run through those. The first was task expansion,
• 2:10
  the second. Was.
• 2:12
  Blurred boundaries and the third was compulsive multitasking.
• 2:15
  So for the first task expansion they called it.
• 2:18
  Basically, they found that product.
• 2:20
  Managers started writing code. Researchers took on engineering.
• 2:23
  Work People suddenly started absorbing tasks that would have
• 2:27
  definitely been handed off. To another person in their.
• 2:30
  Team or just never attempted because.
• 2:34
  AI just made it feel possible. And the downstream effect of
• 2:38
  this? Wasn't efficiency.
• 2:39
  It was actually. Just load.
• 2:40
  Redistribution, removing some bottlenecks.
• 2:43
  And. Redistributing it across teams,
• 2:46
  engineers found. Themselves becoming as well the
• 2:48
  researcher. Described and as most people
• 2:50
  working with AI coding quality inspectors.
• 2:54
  For what could be described as a.
• 2:55
  Sometimes fantastic, but often junior colleague as.
• 2:59
  I said you had product. Managers starting to write code
• 3:01
  themselves. In the product function, we're
• 3:04
  already starting to see product manage pick up smaller bugs or
• 3:07
  tasks. The same with designers.
• 3:09
  Why would a designer? Design things just in Figma.
• 3:12
  When you could be in core code with access to the code based.
• 3:14
  On your local machine and start. To.
• 3:16
  Create things as well that could actually just be merged there
• 3:19
  and then. We've seen this happen,
• 3:22
  everyone. Across our product.
• 3:23
  Organization is now in an Ida developer environment basically,
• 3:28
  so suddenly everyone in the team has access.
• 3:31
  To the entire code base, both to talk to to.
• 3:33
  Investigate ideas and problems and theories to plan work.
• 3:38
  But also to just. Start fixing things and getting
• 3:40
  stuck in. And if people have the right
• 3:42
  system set. Up what our company has suddenly
• 3:45
  anyone can start to produce something so yeah, that task
• 3:48
  expansion certainly means that people are just.
• 3:51
  Doing more work. That they previously didn't do.
• 3:54
  And this brings us on to the second area of work
• 3:57
  intensification that AI brought which was blurred boundaries so
• 4:01
  because. Prompting an AI.
• 4:03
  Is just so simple and easy it literally takes seconds it feels
• 4:06
  like nothing work. Started to easily seep into a
• 4:09
  lunch break. You could go on your phone.
• 4:11
  I do this all the time and just. Quickly voice note Claude.
• 4:14
  Because you know it's important for the next task and you want
• 4:16
  to make sure that Claude is working whilst you rest and eat.
• 4:20
  I'm terrible for this, whether I'm with friends with whom
• 4:23
  having lunch, whether I'm in a meeting.
• 4:26
  I'm thinking about what the next.
• 4:27
  Prompt is or. The next thing is that I need
• 4:29
  this different set of AIS to go and do for me and so very
• 4:32
  easily. I have 456 tasks going on and
• 4:35
  I'm doing them and managing them at all times of the day.
• 4:38
  So yeah, the research found that workers.
• 4:39
  Realized, often in hindsight, that.
• 4:42
  Downtime no longer. Provided that.
• 4:44
  Same sense of. Recovery.
• 4:46
  And as a result, work fell even less bounded and a bit.
• 4:49
  More ambient, something that could.
• 4:51
  Always be. Advanced a little.
• 4:53
  Bit further with just a little bit of effort.
• 4:55
  So again, that was another. Way that AI.
• 4:58
  Intensified work. And this comes on to the third.
• 5:01
  Way that AI intensified work. Which was?
• 5:05
  Compulsive multitasking. Suddenly they found that workers
• 5:08
  are running multiple. AI processes just at the same
• 5:11
  time, you know, keeping several threads.
• 5:13
  Alive at once. Reviving you know certain tasks
• 5:17
  because the AI could just handle them in the background.
• 5:20
  And in this research? Study, they actually said, and I
• 5:23
  quote this. Created a rhythm where the both
• 5:25
  the human and the. Machine were constantly in
• 5:27
  motion and I think one engineer captured this fantastically in
• 5:31
  this paper. You had a.
• 5:32
  Thought that maybe. Oh, because you could be more.
• 5:35
  Predictive with AI. Then you could save some time
• 5:37
  and you could. Therefore work less.
• 5:39
  But then really. You don't work less, you just
• 5:41
  work the same amount. Or even more.
• 5:43
  And our team often jokes about this Barry, the CEO who who, you
• 5:46
  know, programs every single day, often jokes that it's made him
• 5:49
  tireder than ever before in his entire career.
• 5:52
  And I just said I'm terrible for this.
• 5:53
  Just suddenly multitasking all the time with different tasks.
• 5:56
  And it could even be personal tasks as well and personal
• 5:59
  projects I'm making sure I've got on the go.
• 6:01
  And you know, just naturally you suddenly have increased
• 6:04
  expectations to. Produce more work.
• 6:07
  And as a result of those expectations rising.
• 6:10
  There's no real. Productivity game because you
• 6:12
  fill that spare time doing more to meet that expectation.
• 6:16
  So yeah, these three patterns created a self reinforcing cycle
• 6:19
  and the the research report talks about this.
• 6:22
  Essentially AI accelerated. Certain tasks which raised
• 6:24
  expectations for speed, not through certainly explicit
• 6:27
  demands, but just what became visible.
• 6:30
  And normalized in everyday. Work, you know, higher speed
• 6:34
  made work is more reliant on AI increased reliance.
• 6:37
  On AI, widen the scope of what? Therefore, people attempted to
• 6:41
  do, you know, taking on task from another person's job role
• 6:43
  and as a result of that. Wider scope.
• 6:46
  The total volume of work Therefore, suddenly.
• 6:48
  Increased and then the cycle continues with nobody really at
• 6:51
  the controls. So yeah, increased capability
• 6:54
  leads to increased output, which leads to increased expectations,
• 6:57
  which then increase the amount of pressure to expand and
• 7:02
  improve further. So essentially self regulation
• 7:05
  failed. You know, self regulation of all
• 7:07
  of us individuals has completely failed and.
• 7:11
  The researchers were quite explicit.
• 7:12
  About why they think this happened.
• 7:14
  And they put it. Down to three reasons, which is
• 7:17
  #1 the intensification felt really rewarding, you know?
• 7:20
  Workers described. Momentum and the ability to
• 7:23
  expand capability. Or the thrill of suddenly.
• 7:26
  Being able to do things they've never done.
• 7:27
  Before I I've. Certainly felt that as a huge
• 7:29
  driver. Second, because the work was
• 7:31
  voluntary. Neither the workers nor their
• 7:34
  managers. Recognized it as.
• 7:36
  Overwork and 3rd as colleagues expanded their output informal.
• 7:41
  Expectations also reset. So what was once, you know,
• 7:43
  extra effort became just, you know, the minimum standard of
• 7:47
  expectation for performance. And I think for those that have
• 7:50
  gone really far. Into using AI within teams
• 7:53
  structuring everything around AI and how to get the best out of
• 7:56
  AI like our entire way of working at our company is
• 7:59
  stricture around AI you're probably.
• 8:01
  Going to resonate with a lot of what I'm saying.
• 8:03
  It's described exactly what's happened at Mindset.
• 8:05
  Our output is insane and. What you.
• 8:08
  Might expect from a single person over weeks a year ago.
• 8:13
  I have an expectation to produce in days.
• 8:41
  Because I know it's. Possible if you're.
• 8:43
  Using AI properly again, back to that schools discussion last
• 8:46
  week and it's really interesting I mean for those that listen
• 8:49
  every single week, any opportunity to look back in time
• 8:52
  and trying to understand, you know, why does this feel like it
• 8:55
  keeps happening because it isn't new.
• 8:57
  So there's a really interesting history to this and a really
• 9:00
  interesting theory about why us as humans cannot just be
• 9:04
  satisfied with that increased output and suddenly have leisure
• 9:08
  time to go enjoy with our family and and our hobbies and and do
• 9:12
  things that are nice. Because yeah, this isn't new.
• 9:14
  And this pattern has repeated with almost.
• 9:16
  Every technology sold as a time saver.
• 9:18
  For over 100 years in the 1920's, the average American
• 9:21
  woman spent roughly 60 hours a week on housework.
• 9:24
  So cooking, cleaning, washing, fixing things.
• 9:27
  And so by the 50s there was. Now a washing machine, a vacuum
• 9:32
  cleaner, refrigerator and a gas stove in most people's homes.
• 9:35
  And so the difference. Between the 1920s spending 60
• 9:38
  hours a week and the 50s with all that new technology to make
• 9:41
  things easier. You know, what do we what do we
• 9:43
  think the. Housewife spent on cleaning per
• 9:47
  week. It's about 60 hours a week
• 9:50
  again, and you can look at other technologies for the
• 9:51
  spreadsheet. 20 hours of manual calculation suddenly done in 15
• 9:55
  minutes. Well.
• 9:57
  What was the outcome? Accounting clerks of people
• 10:00
  doing accounting fell by 400,000 people when it was invented, yet
• 10:04
  counters and auditors rose by over 600,000 with e-mail.
• 10:08
  Handful of memos. Per day.
• 10:10
  Became 100. And 21 emails and 11 hours of
• 10:13
  work a week for cars, a researcher found.
• 10:15
  That in 1994, humans spent roughly one hour.
• 10:18
  Per day. Commuting regardless of era
• 10:21
  culture. And even technology.
• 10:23
  So basically. As cows got better, we traveled.
• 10:24
  For further inside them. Because they were.
• 10:27
  Better and it was easier to travel.
• 10:29
  The point being the. Mechanism is always.
• 10:31
  The same so technology gets better and reduces the cost per
• 10:35
  unit of that activity. So it makes it easier to do that
• 10:38
  thing yet. As a result, the volume of that
• 10:41
  activity just. Increases, so we do more of it
• 10:44
  and we just consume the savings in time and effort that we would
• 10:48
  have gained regardless of the technology.
• 10:51
  So that new volume just becomes the baseline.
• 11:16
  And there's a lot of theories behind this.
• 11:17
  There's a well. Known thinker who?
• 11:18
  Explains how this happens perfectly called Hartman.
• 11:21
  Rosa, a German sociologist, and it gives you a really
• 11:24
  interesting perspective on not just this discussion.
• 11:27
  But why the world often in modern?
• 11:30
  Times just feels quite chaotic. You know Rosa.
• 11:32
  'S core argument, which was laid out in the.
• 11:34
  Book. Social acceleration, is that
• 11:36
  modern? Societies are caught in what he.
• 11:38
  Describes a triple acceleration, so you get.
• 11:41
  Technology acceleration. So things just get faster.
• 11:44
  Acceleration of social change. So institutions, norms and
• 11:47
  expectations shift really fast as a consequence.
• 11:50
  And the acceleration of the pace of life.
• 11:53
  Despite all the time saving technology, individuals just
• 11:55
  feel more time pressure and the critical insight I think is that
• 11:59
  these three just feed into each other, into a loop that has no
• 12:02
  natural stopping point. And which is the just the story
• 12:05
  of? Capitalism, Rosa actually called
• 12:07
  this the process of dynamic stabilization, which is
• 12:11
  essentially a fancy word for the modern institutions and
• 12:14
  companies and economies can only maintain their stability by
• 12:18
  continually trying to grow and accelerate and go faster.
• 12:22
  And standing still means you're falling behind.
• 12:25
  Just like the bicycle. You have to just constantly keep
• 12:28
  paddling, otherwise you're just going to fall over.
• 12:30
  So what? What's?
• 12:30
  Really interesting about this is and what does.
• 12:32
  What does it explain why everything continuity still
• 12:35
  feels the same no matter how much cool technology comes
• 12:38
  about? Well.
• 12:40
  Rosa called that. Standstill, which is when
• 12:42
  everything is in constant motion but nothing changes.
• 12:46
  So the act of constantly being in motion means that nothing can
• 12:49
  change. Technology shift, fashion
• 12:51
  changes, work gets busier. And yet the basic.
• 12:53
  Experience of. Falling or feeling like you're
• 12:55
  behind, Feeling stretched, feeling like things are never
• 12:59
  enough. That just stays the same because
• 13:01
  the treadmill is only getting faster.
• 13:02
  And I think that Berkeley study is a case.
• 13:04
  Study of this playing out. On the level.
• 13:06
  Of individual workers in AI. But the more they did, the more.
• 13:09
  That was expected and you know, this expectation becomes the new
• 13:13
  flaw and that's why. AI so far.
• 13:15
  Doesn't save us time, it makes us work more.
• 13:19
  Anyway, I hope you enjoyed the episode and I'll see you next
• 13:22
  week.