Latest / Nonsense with cj little and Jeff Parker / The Silicon Takeover: Will Your Next Boss Be a Toaster?

Transcript

The Silicon Takeover: Will Your Next Boss Be a Toaster? · Nonsense with cj little and Jeff Parker

• 0:00
  Welcome to Nonsense Season 4, Episode 17, our big story in our
• 0:03
  second show of the week. I'm Jeff.
• 0:04
  Parker, I'm CJ Little. I cannot wait for an AI to
• 0:07
  replace me doing that opening. This time we're looking at the
• 0:10
  death of today's worker. How to exist in an AI workplace
• 0:13
  up next? Things can be amusing when the
• 0:14
  AI that replaces me also can't say its name.
• 0:18
  That's exactly that's gonna happen.
• 0:19
  It's gonna train in on hundreds of episodes of nonsense and
• 0:21
  like, well, this is how he says his name.
• 0:23
  Not very well. Listen, you and I both belong to
• 0:25
  a profession that is just been eviscerated by AI.
• 0:29
  Yeah, first hand. We know lots of people who have
• 0:32
  been displaced from AI. Basically 75% of all programmers
• 0:36
  could be out of work. So that is potentially true.
• 0:38
  However, I don't think that's going to be the case, and I have
• 0:40
  good reasons to believe that's not the case.
• 0:42
  OK, save those reasons. We'll talk about it in one
• 0:43
  second. But the main thing is you want
• 0:45
  to be the one who brings in the AI.
• 0:46
  Yeah. You want to be the programmer
• 0:48
  who knows how to program. Using the AI fluent person.
• 0:50
  That's going to be the whole point of this episode, I'm
• 0:52
  betting. Without a doubt.
• 0:52
  Be the person who's in front of the AI charge.
• 0:55
  Don't be the person who's resisting and gets obliterate.
• 0:57
  I'm not saying you have to to survive because maybe there's
• 1:00
  some very nuanced niche industries that AI is not going
• 1:03
  to come for, but I would say for pretty much everybody, you're
• 1:05
  going to benefit by leaning in pretty hard to the current tool
• 1:08
  set that's out. There.
• 1:09
  OK, how you can win the battle with AI for your job.
• 1:11
  Let's go. Do not challenge it to math
• 1:13
  problems. The AI revolution doesn't have
• 1:27
  to be a pink slip. In fact, it can be a power suit.
• 1:29
  While the world frets over displacement, the winners are
• 1:32
  busy mastering new skills and practicing integrating to
• 1:34
  survive. Don't try to compete with
• 1:36
  silicon on its home turf. You'll never out calculate an
• 1:39
  algorithm or out process a server.
• 1:40
  Instead, double down on being the first of becoming the expert
• 1:43
  who knows best how to use the tool.
• 1:45
  Thriving requires a shift from worker to architect.
• 1:48
  Use AI to automate the mundane, then redirect your liberated
• 1:52
  energy toward creative strategy and complex problem solving with
• 1:55
  these tools. As interns, not as overlords.
• 1:58
  Curiosity as your new currency. If you remain the person who
• 2:00
  knows which questions to ask and which ethical lines to hold, you
• 2:04
  won't just stay relevant, you'll become the essential navigator
• 2:07
  in a machine driven world. The impacts of AI, however,
• 2:10
  might be less like COVID, which is sort of something that we
• 2:12
  thought we were going to see, and more like the Internet or
• 2:14
  trade with China. So it might not be as immediate,
• 2:17
  but it will be steep. What do you?
• 2:18
  Mean by steep, Well the effects may not be immediately clear
• 2:20
  from aggregate unemployment data and you might see these factors
• 2:24
  like like trade policy or the business cycle could cloud the
• 2:27
  interpretations of those sort of trend lines.
• 2:29
  So what that means is we might be going down on the ship just a
• 2:32
  little slower than, than, than than we thought previously.
• 2:34
  These things always take longer than before.
• 2:36
  Think they're going to take? Yeah, I mean, this is still
• 2:38
  gonna be faster, I suspect, than the previous big shifts that
• 2:40
  we've seen. It's gonna be different because
• 2:42
  instead of being a blue collar recession, it's going to be a
• 2:44
  white collar. Recession.
• 2:45
  That's a big part of the problem.
• 2:46
  The color, the colors, the color.
• 2:48
  So we should break this up. Let's bring something a couple
• 2:49
  trunks think we should talk about the jobs that we thought
• 2:51
  technology would take and the jobs that it actually appears to
• 2:54
  be taking 'cause I do think those are not exactly the same
• 2:56
  thing. We should talk about what roles
• 2:58
  are going to be in demand going forward because I feel like
• 3:00
  that's important. And then we should talk about
• 3:02
  building a, a sort of a, what I call a hybrid skill set that an
• 3:06
  AI can't penetrate and how you can leverage the tech to be
• 3:09
  ahead of the the wave of AI that's crushing folks below.
• 3:12
  It I don't think we're going to have fewer jobs necessarily.
• 3:14
  I do think they're going to be very different jobs.
• 3:16
  Oh yes, if you're thinking. You're going to do your father's
• 3:19
  job. Oh, your father's job, That's.
• 3:21
  Not going to happen. If you're thinking you're going
• 3:23
  to be doing your job, that's not going to happen if you're.
• 3:25
  Well, you're that's true. That's true.
• 3:27
  Yeah, the intern that you hired 10 years ago, you're going to be
• 3:29
  doing his or her job. Yeah, you're not going to be
• 3:32
  doing that. Like I do think that entire
• 3:33
  landscape is going to change drastically of the careers you
• 3:36
  and I know best. I would say what it would be
• 3:38
  more software and specifically, Sure, computers, right,
• 3:41
  Computing the. The computer programmers have
• 3:43
  the largest exposed occupation as of our current measuring.
• 3:46
  This is like literally as of the past week or two by the.
• 3:48
  Way you know why that's the case?
• 3:50
  Why? I think that's the case.
• 3:51
  I have some theories. I'll give mine, then you give.
• 3:53
  Yours. You go first.
• 3:54
  Because we want to to make our jobs easier, and we were the
• 3:57
  computer programmers who could make our jobs easier.
• 4:00
  I think that's part of. It in the process we managed to
• 4:02
  be to make the tools so good yes that they just eliminated all of
• 4:06
  the junior roles well look I. Don't think they eliminated the
• 4:08
  junior roles per SE. They limited the current junior
• 4:10
  roles, Yes, for sure. The other reason I think
• 4:13
  computer programming was one of the the the earliest and ones to
• 4:17
  take out with AI is it's all just pure logic.
• 4:19
  It's basically math sure with letters.
• 4:21
  So that's the easiest thing to get an AI to understand because
• 4:25
  it's very black and white. There isn't a lot of liberty.
• 4:28
  Now, I would say let's talk about like a senior software
• 4:31
  developer, you know, the the linguistics, you know how to how
• 4:35
  to write down statements and you know how to how to drive
• 4:38
  solutions. Good senior engineers know how
• 4:40
  to do the art right. They know how to make it look
• 4:42
  well and be maintained. I don't think we figured out any
• 4:45
  of that with AI solutions at all.
• 4:47
  Yeah. And right now we're avoiding
• 4:48
  that problem because they're just saying, well, whatever my
• 4:50
  cost to build it was so cheap, who cares?
• 4:52
  And we're just throwing caution to the wind.
• 4:54
  I do believe that will catch up with us at some point.
• 4:56
  At some point, you're not going to want to just throw compute at
• 4:59
  it. You're going to want to say,
• 5:00
  well, let's make this a little more efficient.
• 5:01
  Or every once in a while does this weird thing and nobody
• 5:04
  knows why. Well, the code is.
• 5:06
  Still going to do that? Yeah, but I, I'm just saying I
• 5:08
  think there's going to be some some balance where it's going to
• 5:11
  be better agents that are better at making the the solution the
• 5:15
  way you want it to be. And the you in this case should
• 5:17
  be someone who is a an architect and an orchestrator and someone
• 5:21
  who knows how to craft what they want.
• 5:22
  They can lever the tools to help craft it, not just a black box
• 5:25
  that appears to do the thing that you wanted to do.
• 5:27
  I do think that's where we're going on, on the software side.
• 5:29
  If I may be so bold as to use language in kind of non standard
• 5:34
  ways, yeah. Engineering is going away.
• 5:36
  Architecture is what will prevail I.
• 5:38
  Understand what you're saying. I do think that like, just like
• 5:41
  for you and I, you, we've used reference before here on the
• 5:43
  show. The days of writing code that a
• 5:46
  is directly interpreted by a chip have been long gone for
• 5:49
  decades. Right.
• 5:50
  You're not writing assembly. Yeah, Yeah.
• 5:52
  You're using an abstraction layer.
• 5:53
  I see this as the next big abstraction layer.
• 5:56
  And I think if we think of it that way, sure.
• 5:57
  Then you say, well, OK, does that mean 75% of the computer
• 6:00
  programming jobs that are covered by this, are those going
• 6:02
  to go away? Like you just don't need them
• 6:04
  anymore, like the roles? Yeah, Yeah, maybe.
• 6:06
  Yeah, they will. But you still have the people,
• 6:08
  and some portion of those people have some expertise, and they
• 6:10
  can allow us to do even more than we could do before.
• 6:13
  And I haven't seen anything that tells me we're going to need
• 6:16
  less software going forward. If anything is set, we're going
• 6:18
  to need a lot more. Yeah.
• 6:19
  So I don't know what percentage. We preface this by saying many a
• 6:23
  times we have said XY and Z jobs are going to be destroyed by
• 6:27
  some elements of tech. Yeah.
• 6:28
  And typically we have not been what I will call quote, UN
• 6:32
  quote, right about that, right. Years ago, there was a prominent
• 6:35
  attempt to measure job offshore ability and they identified
• 6:37
  roughly 1/4 of U.S. jobs as vulnerable.
• 6:40
  But a decade plus later, most of those jobs maintained healthy
• 6:43
  employment growth within the US. Like what?
• 6:45
  Give me a specific what kind? What's a job?
• 6:47
  Like that customer service reps yeah, we did offshore a lot of
• 6:50
  that, but we ended up having a lot more customer service and
• 6:52
  you ended up with gradients of customer service.
• 6:55
  The government can't really predict this, right?
• 6:57
  Their own occupational growth forecasts, while typically they
• 7:00
  have the right direction right up or down, just try to
• 7:02
  extrapolate the previous trends and and move that.
• 7:04
  Forward. But that's true not just of the
• 7:06
  government. That's of course, if you could
• 7:08
  predict those things, if you could predict what the next big
• 7:10
  thing would be, you could be the next billionaire because you
• 7:13
  could be in front of where the world is going.
• 7:15
  Well, that's that's why all of these AI companies have
• 7:18
  valuations in the the near trillions.
• 7:19
  People do believe they are predicting where that's going.
• 7:22
  I'm not saying they're right. I'm just saying that that that's
• 7:24
  there do seem to be some large bets placed on that, sure.
• 7:26
  I'm I'm quite seeing the the profitability.
• 7:28
  But let's go back a little bit to middle of last year.
• 7:31
  So this is May of 2025. Dario Amodi.
• 7:33
  So he's the CEO anthropic of Anthropic.
• 7:36
  And he said that AI could drive unemployment up to 10 to 20% in
• 7:40
  the next one to five years and quote, wipe out half of all
• 7:43
  entry level white collar jobs. That was his sentiment about a
• 7:46
  year ago, Jim Farley, the CEO of Ford estimated that AI would
• 7:50
  eliminate, quote, literally half of all white collar workers in a
• 7:53
  decade. Wow.
• 7:54
  Those are folks saying, hey, look, there's something that's
• 7:56
  coming. A couple things that that we
• 7:57
  alluded to earlier that I didn't want to point out.
• 7:59
  Previous tech waves have pretty much always hit the blue collar
• 8:03
  workers the hardest. Looks like blue collar is the
• 8:05
  safest. Yeah, if you spin a wrench, your
• 8:07
  job's secure. For now at least.
• 8:08
  At least for the moment. But right now where we see the
• 8:10
  real risk is in the, you know, if you've got higher education
• 8:14
  and higher salary, you likely have higher exposure to AI.
• 8:17
  Yeah, that's our current sentiment.
• 8:18
  I, I have some, some arguments that sort of work against that.
• 8:21
  But generally speaking, those are the ones that we're seeing
• 8:24
  the biggest impact by AI. We also sort of as a as a a
• 8:27
  subtext to that the workers in the most exposed professions are
• 8:30
  more likely to be older, female and more educated and higher
• 8:33
  paid, which I already mentioned. Yeah, sure.
• 8:34
  Those are the ones which you would think would be the least
• 8:37
  to go. But those are the ones that I
• 8:39
  think we're going to see the most impact on the soonest.
• 8:41
  And that's what we want to like. How can we be in front of that
• 8:44
  wave, especially for those folks?
• 8:45
  Yeah, I mentioned customer service reps That's over 70%
• 8:48
  exposure, data entry at 67%. How data entry is still a thing
• 8:52
  is beyond me, right. Even something as specific and
• 8:54
  again, very logical like medical records specialist that's sort
• 8:57
  of got 2/3 exposure, you know, 67 ish percent.
• 9:00
  There's a whole lot of of these industries.
• 9:03
  We go, wow, there's a lot of expertise there, but now you're
• 9:05
  at risk if you're in market research or if you're doing
• 9:08
  financial investment analysis, you are at severe exposure,
• 9:10
  which likely means severe risk. Analyzing financials is my day
• 9:14
  job, no. No, no, hold on.
• 9:15
  First of all, I'm not saying everyone, right?
• 9:17
  I didn't say 100%, but you and I both know a whole bunch of of
• 9:21
  analysts that like, yeah, they should be.
• 9:22
  They should be. They should be eradicated.
• 9:24
  The most important thing about being a financial analyst, if
• 9:26
  you, especially if you're doing your own, yeah, capital
• 9:27
  allocation, is more temperament than it necessarily is skill
• 9:31
  set. Yeah, a 5 year old could do the
• 9:32
  math. Yeah, but it's a very different
• 9:34
  thing to have the temperament to be able to do something and know
• 9:36
  you're doing the right thing when everyone else is doing
• 9:38
  something. Else, but more just making the
• 9:40
  point that these are all very white Collarish roles.
• 9:43
  Yeah. And those are ones that we're
• 9:44
  seeing sort of at risk. So that's a bucket.
• 9:45
  We should look at this data this episode through.
• 9:48
  You can add in accountants, you can add in lawyers.
• 9:51
  Yep, a non exhaustive list of jobs that have zero coverage by
• 9:55
  AI. I currently OK cooks, motorcycle
• 9:57
  mechanics, lifeguards, bartenders, dishwashers.
• 10:01
  They've 0 exposure to AI currently.
• 10:03
  Yeah, that may change, but right now it's.
• 10:05
  Not worth it to make the robot that goes and collects The Dirty
• 10:08
  dishes off the restaurant tables and takes them into the kitchen
• 10:11
  yet. Yet I mean a bunch of people at
• 10:13
  CES think it is and and apparently Tesla, the non car,
• 10:16
  the early car company thinks it is.
• 10:18
  And what I think is interesting about this is I alluded to the
• 10:20
  the sort of blue collar roles we've hit previously.
• 10:23
  I would say we thought tech was more coming for or specific with
• 10:26
  white collar more than what I call the middle manager chunk of
• 10:29
  white collar jobs. And that was kind of the
• 10:31
  expectation you're going to have tools like better workflow
• 10:34
  management, right? Whether it's Jira, which makes
• 10:36
  me grumble even thinking about or the Mondays of the world or,
• 10:39
  or things that would make like what would classically be a
• 10:41
  program manager's job and a middle manager's job make it
• 10:44
  easier for them. So you need less of them to
• 10:46
  manage more people. Sure.
• 10:47
  That's at least sort of where I thought this was going.
• 10:49
  But what's interesting is now you go, well, those folks might
• 10:51
  actually have the skill set that is more in demand.
• 10:53
  If they can manage people, they might be able to manage agents.
• 10:57
  Is that what you're saying? Yeah.
• 10:58
  So let's. Talk about that.
• 10:59
  We talk about the the roles that are likely going away.
• 11:01
  I'm pretty bearish over some of the things people talk about
• 11:04
  right now. Like there's a lot of talk about
• 11:05
  being a like what? Like being a prompt.
• 11:06
  Engineer. Oh yeah, yeah, yeah.
• 11:07
  And I'm like, OK, first of all, you know how to talk to the
• 11:10
  current interface. Good on you, right?
• 11:12
  To think that interface is going to be consistent is insane
• 11:14
  'cause it's not going to be. In the last week, we've added
• 11:17
  four new ways you can talk. To exactly, exactly AI and it's
• 11:20
  just getting better. So like, yeah, you are at best
• 11:23
  a, a, you know, you're one of the little pads you step on when
• 11:26
  playing Frogger, but that's it. Like it's just, I'm just walking
• 11:29
  across you to get to the other side.
• 11:31
  This whole prompt engineer concept, to me, this is the
• 11:33
  equivalent of like in the late 90s when people would write that
• 11:35
  they're quote proficient in Microsoft Word on their resume,
• 11:38
  right? Like, right.
• 11:39
  I guess maybe that was nuanced then and different because you
• 11:41
  knew where the bold button was. But now, like people are just
• 11:43
  like, wait, what? You use Microsoft Word?
• 11:45
  Why? What are you doing?
• 11:48
  It's no longer relevant. So I think the whole prompt
• 11:50
  engineer is sort of going that way and there have been N
• 11:53
  variations of prompt engineer and I think they're all
• 11:55
  generally in the same bucket. It's like you get into the
• 11:57
  prompt engineering and you're like, oh wait, no, this is going
• 11:59
  away. I should be this and you're
• 12:00
  like, I'm a context engineer. It's like, yeah, that's going
• 12:02
  away too. Yeah.
• 12:03
  You know, short term, sure, maybe you do something
• 12:04
  interesting. So this is where we start to
• 12:07
  think about, OK, well, from a jobs perspective, what are the
• 12:10
  things that are going to be valuable?
• 12:11
  What I do see being long tail in the future is having multiple
• 12:14
  things, let's just call them agents right now, doing tasks
• 12:18
  for you, right? That does seem to be at least
• 12:20
  for the, let's call it the medium term, the way we're going
• 12:22
  to execute on these things. So you.
• 12:24
  Need to do the orchestrator of the agents.
• 12:26
  You want to be the orchestrator of the agents.
• 12:27
  You don't have to know how to execute the thing right.
• 12:29
  You want to be able to orchestrate them and probably be
• 12:31
  able to do some modicum of what I'll call quality assurance when
• 12:34
  it spits the thing out, just to know if it's right or wrong.
• 12:36
  Basically, you're a fact checker.
• 12:38
  Kind of a fact checker, you know, maybe it's like a fact
• 12:40
  sniff, like check. You're like, OK, this doesn't
• 12:42
  look terrible. This is probably right.
• 12:44
  By the way, that's useful. All these white collar
• 12:46
  apprenticeship jobs like law, accounting and marketing, those
• 12:48
  kind of things. Yeah, they don't need that entry
• 12:50
  level person who looks up Jeff, VCJ.
• 12:52
  Yeah, Case remembers that from law school.
• 12:54
  But they do need someone one to Fact Check all those things.
• 12:57
  Yeah, the accounting people do need someone to Fact Check all
• 12:59
  the accounting and make sure that the AI got it all right.
• 13:01
  Those jobs haven't gone gone away.
• 13:03
  This. Is what's kind of interesting
• 13:04
  about this is so if you think about the like those
• 13:07
  internships, right, that that stepping stone job, a lot of
• 13:10
  those are being eradicated as, as Jeff just mentioned, right?
• 13:13
  Those are ones where it's like you don't really need someone in
• 13:15
  that role, someone who's willing to work for a relatively low
• 13:18
  wage to just get the experience and do the stuff.
• 13:20
  Nobody sort of does the grunt work.
• 13:21
  You're a paralegal for a while and then you're just, you know,
• 13:24
  you're writing briefs or whatever you're doing sort of
• 13:25
  the next step. A lot of that is being
• 13:27
  displaced. And there is one train of
• 13:29
  thought that says, well, now you're not training the next
• 13:32
  generation of whatever those interns become.
• 13:34
  And I can, I can sort of see that.
• 13:36
  But I also argue like what part of that experience is relevant
• 13:39
  and the parts that are going to be relevant in an AI first or AI
• 13:42
  only world, How do you replicate generating that?
• 13:45
  That doesn't take five years of being an intern and getting
• 13:48
  yelled at. I do think there's a path there.
• 13:50
  We don't know what it is yet, but I do believe we will figure
• 13:53
  it out. It will be identified pretty
• 13:54
  quickly. In some careers, we'll figure it
• 13:56
  out fairly fast. The careers like law, which.
• 13:58
  Ones do you think we want? I think.
• 14:00
  It's much more difficult in law. I think it's much more difficult
• 14:02
  in a lot in marketing. There's there's a lot of
• 14:04
  subtlety there that does require working with those people who
• 14:08
  were used to be over you sure now, well you don't have those
• 14:11
  those entry level jobs, so you don't have those people over
• 14:13
  you. So how do you get to that?
• 14:15
  Through that knowledge gap maybe?
• 14:16
  You still do work with them, you just don't work with them in the
• 14:18
  same way. Yeah, that's right in the.
• 14:20
  Past typically knowledge and experience is of great value,
• 14:24
  right? You've, you're a senior engineer
• 14:26
  because you've made a ton of mistakes and you figured stuff
• 14:28
  out. And a lot of times your
• 14:29
  compensation will be proportional to your knowledge
• 14:32
  set because you know a lot and there are less of you that have
• 14:34
  stuck around and done that. And I do think that's valuable.
• 14:38
  I think it'll be valuable going forward.
• 14:39
  However, there is another metric that is going to explode even
• 14:42
  more. Now.
• 14:43
  The landscape's changing quite a bit.
• 14:44
  So your your knowledge and experience that from 20 years
• 14:47
  ago is going to be less valuable.
• 14:49
  Oh yeah, for sure. You know, today and even less
• 14:51
  valuable tomorrow, but productivity is another lever
• 14:54
  that we have to account for. And this is the thing that I
• 14:57
  think we're going to see the biggest explosion in.
• 14:59
  In the past, it was hard to compare productivity person to
• 15:01
  person like especially with software.
• 15:03
  What are you looking at? Lines of code?
• 15:04
  It's a terrible metric you can't use.
• 15:06
  Lines of code IBM used to do that and it was dumb.
• 15:08
  Terrible. So you try to look at like the
• 15:10
  size of the feature, the thing that they're building, you know,
• 15:13
  and like you try to go, well, OK, this is a big one and this
• 15:15
  person's doing at this speed, that one's doing at that.
• 15:17
  But even that's not great. The thing is now that you've got
• 15:19
  this almost unlimited ability to scale agents to help you do it
• 15:23
  like the limiting factors have move.
• 15:26
  So productivity and output, I think are going to become more
• 15:28
  and more important, especially when you're thinking about
• 15:31
  compensation and what role you're in.
• 15:33
  It's not going to just be doing lots of work in the past I'm.
• 15:36
  Cautious on your saying compensation, but I think 'cause
• 15:40
  I think compensation is still going to be well competitively
• 15:43
  based. They work in the I can
• 15:44
  absolutely 50 times the work tomorrow that I put out today
• 15:48
  does not mean I'm worth 50 times more money.
• 15:51
  I'm not saying it'll be linear at all, but that might be the
• 15:53
  new baseline. I I'm saying more of that, like
• 15:55
  if that's, it's the baseline where before you might say,
• 15:57
  well, OK, if you've got 20 years of experience in web
• 16:00
  development, you're, you're at a certain mark, right?
• 16:02
  And you're going to get into some.
• 16:04
  I'm really using compensation as a, as like a, like a title
• 16:06
  placeholder. Like the, the compensation might
• 16:09
  follow the title, but like you're in a band there because
• 16:11
  you've got this experience set and, and the way you get there
• 16:13
  is by doing a bunch of stuff and likely spending a decent amount
• 16:15
  of time doing it. I'm saying productivity is going
• 16:18
  to be another measure that gets you onto one of these bands.
• 16:21
  That's really what I'm saying. Now the compensation is still
• 16:23
  going to be competitive at market for sure, but it will be
• 16:25
  a supply demand and if you can do 10 times more than the person
• 16:28
  next to you, you're likely going to be in a different band.
• 16:30
  Does not mean you get 10 times as much.
• 16:32
  That is not what I'm suggesting at all.
• 16:33
  Correct, but you might be at a 2X or roll correct?
• 16:36
  Or just some different set of perks because you're doing so
• 16:38
  much more money. And compensation is not the only
• 16:40
  lever there either. If I could have someone who was,
• 16:42
  you know, able to do things agentically choice as fast as
• 16:45
  someone next to them, do they need to work the same number of
• 16:47
  hours? Like what's what's my what am I
• 16:49
  measuring their success by? Is it time?
• 16:51
  Is it output? No, but in about 5 minutes, the
• 16:53
  kids who just graduated from college and do know how to do
• 16:56
  the same thing with agents, Yeah, they're going to displace
• 16:59
  all of those people who don't So.
• 17:00
  And you're going to if you're. Getting like for like in the
• 17:02
  same quality output and you know, and it's not so much a
• 17:04
  gobbledygook, which which you're not, but let's just say that you
• 17:06
  were yeah then yes, your supply now is able to do a lot more.
• 17:10
  And if I start putting them into the bucket, it might just be not
• 17:12
  that you're gonna get compensated less.
• 17:13
  You're not going to get compensated.
• 17:14
  Yeah, you're not going to be in that role.
• 17:16
  And I'm really just trying to point out that that productivity
• 17:18
  in the amount that you spit out is an important metric.
• 17:20
  And it's one that I think is going to explode more and likely
• 17:23
  be measured more than we've seen before, at least in in computer
• 17:26
  software, at least in software engineering, I suspect in others
• 17:29
  as. Well, I like the idea.
• 17:30
  Again, the benefit to you always is to be the expert in whatever
• 17:34
  it is that you're doing. Yeah, be better competitively
• 17:37
  than everyone else around. You know, all the tools in the
• 17:39
  space. Yeah.
• 17:40
  I mean, how often do you see managers get asked what's
• 17:42
  precise teams have you managed? Like somehow that's a metric of
• 17:45
  success? I've never found it to be
• 17:47
  proportional to metrics of success.
• 17:48
  Sure, but I could see a world where it's like, well, how many
• 17:50
  agents do you manage on a weekly basis?
• 17:51
  If your answer is 2, you're in one bucket.
• 17:53
  If your answer is 20, you're in another bucket.
• 17:55
  That's more what I'm trying to say.
• 17:56
  Is it it? I do think.
• 17:57
  Well, it depends on the quality of the output.
• 17:59
  You might of. Course, of course you might.
• 18:00
  Be able to manage hundreds of agents and you put out garbage.
• 18:03
  I'm saying hold the quality standard across that right.
• 18:06
  There's some bar of of that. Someone who can manage more
• 18:08
  because they've got some system or have just built a thing that
• 18:11
  does it is typically going to be more valuable or at least more
• 18:13
  interesting. Provided a quality of output is
• 18:15
  similar, which it never is. Oh, of course.
• 18:17
  But, you know, congruent at least.
• 18:18
  Maybe not similar, but congruent.
• 18:20
  Yeah. I also do think you're gonna see
• 18:21
  this world where, like, remember, like when saddles,
• 18:23
  like you had horse saddles. Sure.
• 18:25
  And that was like a, like, it was like a, like a yeah,
• 18:27
  workman's tool. Right.
• 18:28
  Like if you were a cowboy. Sure.
• 18:29
  Or if you were, and if you were in farming, you had a horse and
• 18:32
  you had a saddle and it was a very utilitarian sort of thing.
• 18:34
  Yeah. And then at some point, saddles
• 18:35
  became, not that they became like, luxurious.
• 18:38
  Right. And like they sold a lot fewer.
• 18:39
  Of them what human things are gonna become luxurious in in an
• 18:43
  AI world. I don't know the answer to that,
• 18:44
  but it might be like, oh, here's a human being playing the flute
• 18:47
  and you're like, whoa, that's super unique.
• 18:49
  I've read about this sure. Because now all you've ever
• 18:51
  heard are the synthetic AI MIDI flute, right?
• 18:53
  Like that's all you've ever heard if you know you're born in
• 18:55
  2035. The version of MIDI, right?
• 18:58
  Whatever it is going. Forward like I just I think it's
• 19:00
  kind of fascinating to think about what are what are the
• 19:02
  things that are gonna be human, but that's like, you know what,
• 19:04
  like small market like artistry, right like oh.
• 19:07
  That's the artisan stuff. Yes, there will.
• 19:09
  There will still be a market for artisan stuff.
• 19:11
  There will be still a market for things that are made by humans.
• 19:13
  Yeah. Forget that for a second.
• 19:14
  Yeah. Think about doctors.
• 19:15
  OK. Will you need more doctors when
• 19:17
  AI becomes a very, very good doctor?
• 19:20
  Yeah. Now a doctor does other things
• 19:21
  besides diagnose your problems. Yeah.
• 19:23
  He also is doing bedside manner. And he's looking at you, you
• 19:26
  know, and how you're reacting to things and telling you in ways
• 19:29
  that you can understand things. Sure.
• 19:30
  AI eventually will be able to do those things as well.
• 19:32
  So will medicine become more? You know, right now we have this
• 19:36
  huge wealth gap in medicine. Will AI democratize medicine?
• 19:40
  I think there's a good chance that it will make a big dent in
• 19:43
  it. I just don't know how.
• 19:45
  Like, I see AI being a lovely assistant to my doctor.
• 19:50
  I don't see AI replacing my doctor yet.
• 19:52
  Yeah. Could I see an end state where,
• 19:54
  like, it becomes my first tier of doctor?
• 19:56
  Like instead of going right to the to the urgent care, right, I
• 19:59
  go to the AI and like, at least go through the OK, It's not
• 20:02
  this, It's not that. It's probably not this.
• 20:03
  And then when I do go finally see a human being, a doctor,
• 20:06
  it's probably them navigating the AI of being like, well, OK,
• 20:09
  it could be this. Let's look at these things.
• 20:11
  Or just overseeing a bunch of AIS as they are helping.
• 20:14
  Patients like telemetry so much what a doctor does is telemetry
• 20:17
  and then a cross set of like your history, yeah, what you
• 20:20
  currently have happening and like what you know, the possible
• 20:22
  sets of what it could be. That is perfect for an AI to do,
• 20:25
  you know, take a picture of your throat.
• 20:26
  It's like probably not strep, right.
• 20:28
  And can you get into a world you've got rapid tests deployed
• 20:30
  at home or some device on your phone that can do not all but
• 20:33
  like 15 or the most common rapid tests?
• 20:35
  Sure. Yeah, I see that being a a real.
• 20:37
  Thing I wasn't even thinking about necessarily at home, Yeah,
• 20:39
  I was thinking about if you go and do a doctor visit, but but
• 20:41
  an AI is going to be very good, yeah, hearing your symptoms, of
• 20:44
  course, and looking at your history and coming to very
• 20:47
  intelligent conclusions. No, I see it being, I see it
• 20:49
  being a a necessary assistance, sort of like EMR.
• 20:52
  Like how could you run a doctor's office without EMR
• 20:54
  these days? Explain to people.
• 20:55
  What EMR is? Electronic medical records, I
• 20:57
  just don't think you could provide a level of service like
• 21:00
  you just it would be so odd of everything in paper, I just
• 21:02
  don't see that really being. I'm sure there are some that
• 21:04
  still do, but man that seems odd.
• 21:05
  I see these being a very good assistance and sort of like
• 21:09
  colleagues for doctors for sure. But I'm carrying that farther
• 21:13
  and saying if your first step is urgent care or telemedicine,
• 21:16
  this could easily replace that. Yeah.
• 21:18
  And I, I'm excited about that because there's so many things.
• 21:20
  Of course, I'm like, man, I'm going in for this rodent, you
• 21:23
  know, diagnostics. I'm wasting my doctor's time.
• 21:26
  They should be working on someone that's more complicated,
• 21:27
  right? Problem set than this.
• 21:29
  And your insurance company's money 'cause they're spending a
• 21:31
  fortune for you to go and do that.
• 21:33
  Yeah, thing that is a very simple.
• 21:34
  Thing I mean, I suspect if you had really good, a really good
• 21:37
  AI first tier, you know, pre telemedicine, pre urgent care
• 21:41
  step, I don't think insurance companies would embrace that
• 21:44
  they can't make money off that transaction.
• 21:45
  I don't see them being a good right.
• 21:47
  When have they ever wanted to make something more efficient?
• 21:48
  Right, they can't. Right.
• 21:49
  Like that's not their job. They thrive in less efficiency.
• 21:52
  Less efficiency gives them more opportunity to scrape out money,
• 21:55
  either by putting in some fee or by just taking your money and
• 21:58
  not paying the thing like that. They've never struck me as folks
• 22:01
  that are like, we want to be efficient.
• 22:02
  That's probably accurate. If they were, we'd have much
• 22:05
  better health care. But there's no part of the
• 22:07
  general health care system screams efficiency.
• 22:10
  From PBMS to health insurance providers.
• 22:13
  Yeah, sure. They're never like, oh, let's
• 22:14
  make this efficient. They're always like, what can we
• 22:16
  do to make this more friction? Yeah, Now it's interesting to
• 22:18
  think about jobs like doctors where you think there's so much
• 22:20
  experience and so much wisdom that's been gained over years
• 22:23
  and years of of of practice, of medical practice.
• 22:26
  And then you realize, oh wow, the AI can gain that experience
• 22:28
  in a matter of like minutes. Yeah, I already has that
• 22:30
  experience, right? It already has benefited all
• 22:32
  that experience. Right.
• 22:33
  A doctor looks at a chart as a medical scan of in some way, and
• 22:37
  it's very hard, but they can see where this little problem is
• 22:40
  over here. Yeah, the AI can not only see
• 22:42
  the little problem over here. Yeah, you can see the five other
• 22:44
  problems that the doctor isn't even looking.
• 22:45
  For yeah, yeah, yeah, totally. It's a bright future.
• 22:47
  I love it. It is a bright future.
• 22:48
  I'm excited. Unless you're a doctor, no, I
• 22:50
  think. You were the doctor.
• 22:51
  It's going to be things going to be bright and do you.
• 22:52
  Think the order of events is important to this a couple
• 22:54
  times, and I keep wanting to say I do think it is important for
• 22:57
  the AI to be answering to the human, at least in the near
• 23:00
  term. Yeah, I do not like the way MLB
• 23:02
  is going to be calling balls and strikes where the umpire will
• 23:05
  make the call and then if you want to challenge it, then you
• 23:08
  go to the computer who then tells the umpire you're wrong.
• 23:10
  Go to the other way around. Much rather have it be the other
• 23:12
  way around, where the computer is making all the calls and if
• 23:15
  the umpire wants to override what the computer has said he
• 23:18
  can do. That OK, The only time that
• 23:20
  would actually need need to happen is if like a bird flew in
• 23:23
  front of all the cameras at the same time.
• 23:24
  It would just get it right forever, right?
• 23:26
  You just have an up in the back. He would just be like a sack of
• 23:28
  potatoes of the mask on. That's all it would.
• 23:30
  Be and mostly that should be the case, but occasionally you're
• 23:32
  gonna get some anomaly that the computer isn't doing right, and
• 23:35
  it's really nice that the UMP now can override.
• 23:37
  You're right what the computer has said that would.
• 23:39
  Happen one time someone would debug that line of code would
• 23:41
  never happen. Again, right, That's the.
• 23:43
  That's how it should work. That's exactly how it should.
• 23:45
  Work, but if you're gonna automate the UMPS automate the F
• 23:47
  in players too. While you're at it.
• 23:49
  Just automate the whole thing all the way down, OK?
• 23:52
  Major League robot baseball I am.
• 23:54
  In I like the idea of the Doctor, I'm in supervising the
• 23:58
  AI, yes, and making sure the AI are doing things, especially in
• 24:01
  the short. Term, yes, we're going to see
• 24:02
  that for a while. Yes, I think that's a really
• 24:04
  important step. Until what?
• 24:05
  Well, until the AI starts doing it better than the doctors,
• 24:07
  until their outcomes are better than humans.
• 24:09
  And that'll that'll come. And that might not be very far.
• 24:11
  Away, by the way, the same thing in baseball.
• 24:13
  Eventually we work out all those edge cases, all those all those
• 24:15
  weird bug things, and eventually they're much better I.
• 24:17
  Fully suspect that the current systems that they have deployed
• 24:21
  now can call balls and strikes better than UMPS today.
• 24:24
  In 99.9% of the cases, but once in a while they'll get something
• 24:28
  that's just. Dead wrong I know but but but
• 24:30
  I'm just saying I I'm not saying they won't get it.
• 24:32
  Wrong, I know. If you factor that in percentage
• 24:33
  wise, their percentage is still better than a human of.
• 24:35
  Course, yeah. So why don't we switch over now?
• 24:37
  There's more. This human thing of like, Oh no,
• 24:39
  we just, we, we can't do that. We got these UMPS what are they?
• 24:41
  What are they gonna do? They got families, right?
• 24:43
  And. What what everybody says is
• 24:44
  cause the umpires have a very strong union, which is
• 24:46
  absolutely could not be less the case.
• 24:48
  Yeah, I don't think they have absolutely no strength though.
• 24:50
  They have nothing to. I don't think that's it at all.
• 24:52
  It's part of the game. I don't I I think the the
• 24:54
  observers of the game and have a harder time with.
• 24:56
  It I'm an observer of the game, I have no hard time with it.
• 24:58
  Get the calls right. You're unique.
• 24:59
  There's a contest between two teams.
• 25:01
  It is not about this person who makes calls wrong.
• 25:04
  I'm a little passionate about this, aren't?
• 25:05
  You are a little passionate about this.
• 25:06
  Alright, we have to get out here real quick before we do.
• 25:08
  Have you seen it read anything good this last week?
• 25:10
  Oh, I sure have. It's not about baseball.
• 25:12
  Not Well, it is. It's kind of about baseball.
• 25:14
  I'm gonna bring it back to baseball.
• 25:15
  OK? Corey Doctorow coined the term
• 25:17
  inshunification in in 2022. Yes, it was to describe the
• 25:21
  process in which A2 sided online products and services decline in
• 25:25
  quality over time. Of course, for folks who don't
• 25:27
  know, this is where like initially, A vendor will create
• 25:28
  a high quality offering to attract users.
• 25:30
  Often losing money in that offering?
• 25:32
  Sure, because they're trying to attract.
• 25:34
  Users maybe, maybe not yeah, the point is it's a good service.
• 25:36
  You like it, you come, you join and then degrade those offerings
• 25:39
  to better serve business customers and then finally,
• 25:42
  right, degrade their services to both users and business
• 25:44
  customers to maximize short term profits for shareholders.
• 25:46
  Sure, you can see this sort of classic cycle Doctorow
• 25:49
  frequently cites. Google is the the ultimate
• 25:51
  example. Microsoft and Meta are often
• 25:53
  cited as well. There's lots of insatificators
• 25:55
  out there. The Norwegian Consumer Council
• 25:57
  has noticed that digital products and services are
• 26:00
  steadily becoming worse. I love it.
• 26:01
  They sort of paid attention, Sure.
• 26:03
  And to promote this new report that they released that's
• 26:05
  actually really good. It's called Breaking free
• 26:07
  Pathways to a Fair Technological future.
• 26:09
  They created a video promoting their fictional company.
• 26:12
  And I'm, I'm saying fictional. Maybe, maybe this could be a
• 26:14
  real company, Sure. Whose sole purpose is to make
• 26:17
  things shittier. Oh, it's the absurdity of it, of
• 26:20
  course, is what makes it of course, and I suspect that most
• 26:23
  of our listeners have already seen it, being the nerds that
• 26:25
  you all are. But if you haven't, click on the
• 26:27
  link in the show notes and check it out.
• 26:28
  It's called the Day of the Life of the Insatificator on YouTube.
• 26:31
  It's 4 minutes long. It's really fun.
• 26:33
  It's really, really fun, as only the Norwegians can do.
• 26:36
  I am looking forward to. That oh, you will love it.
• 26:38
  You haven't seen it yet? Have you not seen?
• 26:39
  I have not seen. It oh, it made all the rounds
• 26:40
  last week. Oh, it's great.
• 26:41
  I will definitely put a link in the show notes.
• 26:43
  You've got to check it out. How about you?
• 26:44
  What have you. Seen the documentary?
• 26:45
  Sunday Night is a vibrant, fast-paced tribute to The Ed
• 26:48
  Sullivan Show, the ultimate cultural camp of the 20th
• 26:51
  century. It captures how a stiff,
• 26:53
  unlikely columnist became the gatekeeper of American cool.
• 26:57
  The film masterfully weaves together iconic debuts from The
• 27:00
  Beatles, British Invasion to Elvis's pelvis with the high
• 27:03
  stakes drama Behind the Curtain. It highlights Sullivan's
• 27:05
  uncurated role as a civil rights ally, consistently featuring
• 27:10
  Black artists like The Supremes and James Brown during peak
• 27:13
  segregation. It's a compelling look at a man
• 27:15
  who curated the soundtrack of a generation.
• 27:17
  It's fantastic. It's a really good movie.
• 27:19
  You would not believe what a civil rights champion this
• 27:22
  person was. I love it.
• 27:23
  There's a really good documentary that's about Fred
• 27:25
  Rogers. And Fred Rogers flew in the face
• 27:27
  of segregation in 100 different ways.
• 27:30
  There were times when there were, like, outrage that black
• 27:32
  people and white people were swimming in the same swimming
• 27:34
  pool. And that week he had on his
• 27:35
  mailman, who was black, And the two of them took off their shoes
• 27:38
  and socks and put their feet into a wading pool.
• 27:40
  He was an amazing. Man, I'm shocked how much Ed
• 27:42
  Sullivan was that person for adults.
• 27:44
  Was he really? I had no idea.
• 27:46
  Yeah. Oh, wow.
• 27:46
  Is this a recent doc? Yes, I got to check this.
• 27:48
  Out It's So good. Sunday night on Netflix now.
• 27:50
  Definitely check it out. Good wreck that is the episode.
• 27:53
  Thank you for joining us for all this nonsense.
• 27:55
  Actually terrible podcast from the awful company.
• 27:57
  This is on the web at Nonsense dot Productions.
• 27:58
  I'm CJ. I'm Jeff Parker.
• 28:00
  If you like this program, please follow, download, subscribe and
• 28:02
  like at Apple Podcast, YouTube Music, Amazon Music, Antenna
• 28:05
  Pod, iHeartRadio, Spotify, My Favorite Overcast, wherever you
• 28:08
  may get your podcast from. Nonsense is produced using the
• 28:11
  Ardour multitrack digital audio Workstation.
• 28:13
  Amazing full-featured software, open source and free.
• 28:16
  This trick where you can understand me talking while
• 28:18
  music plays underneath is done with odd door.
• 28:21
  Check it out. We'll be here every Tuesday and
• 28:22
  Thursday mornings. For more nonsense, please join
• 28:24
  us at. Least until the AI replaces.
• 28:26
  Which will be next week. Looking forward to it.
• 28:40
  I'm not worried about the AI of 2026 replacing us, the AI of
• 28:44
  2027, because the AI of 2023 could pretty much replace us.
• 28:49
  There's nothing really I'm worried about in the future,
• 28:51
  like it just hasn't decided to come along yet.
• 28:54
  The AI of 1972 could. Do this job?
• 28:56
  Sure, Totally. Yeah, a Commodore 64 press play
• 28:59
  on tape could basically get 85% of what we do.