Latest / In The Loop / The New Super AI Skill: Management

Transcript

The New Super AI Skill: Management · In The Loop

• 0:00
  So I think the market is shifting.
• 0:02
  We're shifting from knowledge work being radically transformed
• 0:05
  with tools like Claude Code being reimagined for knowledge
• 0:09
  workers with Claude Co work. And we're now moving into an era
• 0:13
  where skill sets such as managing AI tools that are doing
• 0:17
  work for you is more important than just using the tool itself.
• 0:20
  And this big change brings with it brand new skill sets.
• 0:24
  And I think these skill sets are going to be incredibly important
• 0:27
  for you at work, your family, your colleagues.
• 0:30
  Because right now, January hiring and firing data in the US
• 0:35
  looks quite painful. It was the worst since 2009,
• 0:38
  with companies openly saying that they're removing and firing
• 0:41
  people as a result of AI. In today's episode, I'm going to
• 0:44
  talk about the exact skill sets that I think are critical for
• 0:47
  you to learn. Judgement, taste, delegation.
• 0:51
  I'm going to talk about frameworks for deciding when to
• 0:54
  use AI or when to hand the work off to AI, and ultimately what
• 0:59
  you can practically do about this massive change.
• 1:01
  This is in the loop with Jack Horton.
• 1:03
  I hope you enjoy the show. So let's start with what skills
• 1:17
  I think are now really important.
• 1:19
  Now, Ethan Molik did a fantastic piece on all of this.
• 1:23
  And for those who don't know him, Ethan Molik is AAI, a
• 1:26
  researcher and professor that puts out fantastic pieces
• 1:29
  online. I'd highly recommend to go check
• 1:31
  him out. But he did a big deep dive into
• 1:34
  important skill sets. His answer was fantastic.
• 1:37
  It's very, very simple. It's management.
• 1:40
  Not management in the sense of sitting in meetings and, you
• 1:42
  know, reviewing slide decks. Management in the way of
• 1:45
  describing exactly what you need to give fantastic and effective
• 1:52
  and clear feedback and design new ways of evaluating work you
• 1:56
  get back from in this context, agents, if you're good at those
• 1:59
  things, you're going to be able to work with agents very well.
• 2:01
  And what I absolutely loved about this piece is that he adds
• 2:04
  a very important nuance Management has always assumed
• 2:08
  scarcity. You delegate because you can't
• 2:12
  do everything yourself and because talent is very limited
• 2:15
  and very expensive. AI completely flips this talent
• 2:18
  is abundant and it's cheap. What's scarce and difficult all
• 2:21
  of a sudden is knowing what to ask for and how to ask it.
• 2:25
  And for example, senior product managers at Google mapped out
• 2:29
  different skill sets that they believe are now critical to a
• 2:33
  modern worker. The ability to shape a problem,
• 2:36
  so turning vague objectives into precise, testable subtasks.
• 2:41
  The ability to curate context, so deciding what information is
• 2:45
  actually important for this subtask and only giving it that
• 2:49
  context so you can get the perfect output.
• 2:51
  Creating fantastic documentation and systems so you can get 90%
• 2:56
  of the way there on a single attempt.
• 2:59
  Another really important skill they labeled was another
• 3:01
  important skill was actually ability to recognise quality.
• 3:04
  The ability to orchestrate many, many agents, managing sequential
• 3:08
  versus, say, parallel tasks. We'll get on to that.
• 3:11
  The ability to understand guard rails and how to ensure agents
• 3:15
  are performing well. Knowing when to intervene with
• 3:18
  an agent or when to start again, and judgement about the tool
• 3:20
  sets you're using. Understanding when to use an
• 3:23
  agent, when to use a simple model and when just to do it
• 3:27
  yourself. These skills aren't new really.
• 3:30
  Yes, they apply it to AI and there's a learning curve there,
• 3:33
  but this is one O 1 management skills.
• 3:36
  I think the CEO of Shopify, it's a fantastic case study here.
• 3:39
  He's GitHub commits, so the actual code that he's committed
• 3:42
  to, Shopify's code base has gone up from 94 commits in 2024 to
• 3:48
  over 900 in 2026 alone. You can tell he's clearly not
• 3:51
  writing all that code himself. He's the CEO.
• 3:53
  He's probably a busy, busy person, but he's orchestrating
• 3:57
  and managing agents to write lots of code and get it into the
• 4:00
  code base. It's turning a CEO of a massive
• 4:03
  company into a builder again, because again, talent is now
• 4:06
  cheaper and more readily available.
• 4:08
  What matters is your ability to instruct it.
• 4:11
  What matters is the skill sets that I'm going to go into that
• 4:14
  ensures what you're producing is actually good quality work.
• 4:18
  We saw this also with Microsoft CEO Satya Nadella.
• 4:21
  They used to have, say, a designer, a front end engineer,
• 4:24
  a back end, a senior architect, a product manager.
• 4:27
  And what they're doing is rolling all of that skill set
• 4:29
  into one person and calling them a full stack builder.
• 4:33
  Someone that can have an idea, shape it, understand the problem
• 4:37
  they're solving, build a lot of it, and get it into a place
• 4:40
  where others can review without relying on 18 meetings and
• 4:44
  documents. So there's a lot of skill sets
• 5:00
  that are emerging, are becoming increasingly important.
• 5:02
  Another one is taste. Taste is about selection,
• 5:05
  recognizing quality amongst many different options or directions.
• 5:10
  If AI generates US10 versions of something, which one do you go
• 5:14
  for? Which one is actually good?
• 5:16
  Judgement on the other hand, is about weighting trade-offs.
• 5:19
  It's about evaluation, especially when you don't have
• 5:21
  complete information, you don't know necessarily the end
• 5:25
  direction or an outcome that you're going for.
• 5:27
  So yeah, suddenly decision making integrates all these
• 5:30
  different skill sets of taste of judgement, of evaluation.
• 5:34
  And then management skills is the coordination of all these
• 5:37
  different decisions across many different agents and tools to
• 5:41
  get more work done than ever before.
• 5:44
  Another really important skill set here is curiosity, which I
• 5:47
  think is arguably at the moment my key most desirable skill set.
• 5:52
  You could have extraordinary taste, but without any agency
• 5:57
  and curiosity, you won't create anything.
• 6:00
  You need to be constantly asking, is there another
• 6:02
  direction to take? Can I prompt it to explore this
• 6:04
  direction? And if it doesn't get me
• 6:06
  anywhere, that's fine. But then I've been curious
• 6:08
  enough to explore it. And again, that judgement and
• 6:11
  taste to know when it's good enough and we're ready to ship.
• 6:15
  So someone could have extraordinary taste, but no,
• 6:18
  let's say agency. So the ability to make a
• 6:20
  decision, therefore you can't create anything.
• 6:24
  You could have lots of agency, but awful judgement.
• 6:27
  So you could move really fast, but in the wrong direction.
• 6:30
  You could have all the curiosity in the world, but without the
• 6:34
  ability to follow through and take action, nothing happens.
• 6:38
  So all these skills combined suddenly become extremely
• 6:41
  important to develop in yourself.
• 6:42
  And we've seen these types of comments from big leaders all
• 6:45
  over the world right now. So the Co founder of Y
• 6:47
  Combinator, probably the most famous starter by accelerator in
• 6:50
  the entire world, believes judgement is now the skill set
• 6:53
  that matters. In his words, when anyone can
• 6:57
  make anything, it's the ability to decide what to make the
• 7:00
  counts. Greg Brockman, the president of
• 7:03
  Open AI, said that taste is now his most important skill.
• 7:07
  And the CTO of Cloudflare, so the thing that basically makes
• 7:10
  the Internet work, said that taste in 2026 is now the
• 7:14
  engineering differentiator. So yeah, I've discussed what
• 7:17
  skills are really important now and why they're really
• 7:19
  important. Because ultimately, as I said,
• 7:22
  you can be a full stop builder in whatever career trajectory,
• 7:26
  whatever career path you're currently doing, you can be a
• 7:28
  full stop builder, someone who can produce more than anyone
• 7:31
  else in the world. But it requires you to have
• 7:34
  those important skill sets, to be trusted in the organization,
• 7:38
  to actually go and do those things.
• 7:52
  So then I think now let's move on to, I guess the equation for
• 7:54
  when to delegate to AI. Think this is a really important
• 7:57
  one to get started with and a really good framework of
• 8:00
  thinking about these big questions.
• 8:01
  Now Ethan Mollick believes it comes down to just three
• 8:04
  variables. 1st, the human baseline of time.
• 8:08
  So how long would a task take for you to complete?
• 8:11
  Second, the probability of success.
• 8:14
  How how likely is AI to give you a good answer and output?
• 8:19
  And 3rd is AI process time. How long would it take you to
• 8:22
  write the prompt, get the answer and evaluate the output?
• 8:26
  Now, on the face of it, it seems quite simple, but that requires
• 8:28
  you to have in depth understanding of the tools, how
• 8:31
  to manage agents, what it's good at and bad at, to have stress
• 8:34
  tested all those different avenues, and therefore to really
• 8:36
  touch yourself into it. And the reason I'm emphasizing
• 8:40
  this point so much about these new skill sets, the way of
• 8:42
  deciding when to delegate or not delegate something to AI is
• 8:47
  because of the huge amount of layoffs that are starting to
• 8:50
  happen across the US. I'm sure the data is going to
• 8:52
  back it up in the UK too. You know, US employers announced
• 8:55
  over 108,000 layoffs in January, which is 120% increase year on
• 8:59
  year. Hiring intentions fell to their
• 9:01
  lowest level since records began and AI was cited in over 7 1/2
• 9:05
  thousand job cuts. And although that only
• 9:08
  represents 7% of those total jobs, I truly believe that it's
• 9:11
  understating the impact AI is starting to have on the job
• 9:14
  market. So what should you take away
• 9:28
  from this episode? Well, ultimately, we're seeing a
• 9:32
  shift from execution to direction.
• 9:34
  And we're seeing this especially in the world of programming and
• 9:37
  developers. But as I've said a number of
• 9:40
  times on this podcast, it's coming for everybody.
• 9:44
  The skills that matter now, our taste, our judgement, agency,
• 9:48
  curiosity, and the ability to coordinate work across many
• 9:51
  different agents. These are the sorts of tasks
• 9:53
  that often people didn't develop in themselves because they were
• 9:57
  so busy on executing things themselves.
• 9:59
  Whereas now everybody is becoming a manager.
• 10:02
  And the economic reality of this change is uneven and in places
• 10:07
  it's going to be really unfair. I think layoffs in this next
• 10:10
  period are probably going to be structural, not temporary, but
• 10:13
  there are many skills that everybody must learn and
• 10:16
  demonstrate if they're going to navigate these quite difficult
• 10:20
  times. Nobody can fully predict what
• 10:23
  work's going to look like, what everybody's a manager, but I
• 10:25
  think those that will thrive are those that can describe exactly
• 10:29
  what they need and make the perfect judgement of when it is
• 10:32
  good enough to ship. Anyway, that's a short episode
• 10:34
  this week on important skill sets that we're seeing emerge.
• 10:38
  I hope you enjoyed the episode, I hope you found it useful and I
• 10:41
  will see you next week.