Latest / MBA Training AI / Durable AI skills: decomposition, verification, taste

Transcript

Durable AI skills: decomposition, verification, taste · MBA Training AI

• 0:00
  Host: Welcome to the Leaders Insights Podcast.
• 0:07
  Host: Today's episode, AI skills that stay relevant as tools change.
• 0:11
  Host: Last week, OpenAI shipped yet another model, and half my LinkedIn feed is people panicking
• 0:17
  Host: that the prompt writing tricks they spent 2025 learning are already dead.
• 0:22
  Host: Are they right?
• 0:23
  Expert: Mostly yes.
• 0:25
  Expert: If you built your whole edge on knowing that GPT-4 liked being told you are an expert, congratulations.
• 0:32
  Expert: You optimized for a quirk that vendors patch out on purpose.
• 0:35
  Expert: The models are being trained to need less handholding.
• 0:38
  Expert: Your clever incantations have a shelf life measured in months.
• 0:41
  Host: That's brutal for the people selling $200 prompt courses.
• 0:45
  Expert: It is, and I don't feel bad about it.
• 0:47
  Expert: There's a real statistic here.
• 0:49
  Expert: The article points to how tool-specific certifications age faster than the tools
• 0:53
  Expert: themselves.
• 0:55
  Expert: Anthropics' Claude interface in September 2026 barely resembles what it was 18 months ago.
• 1:01
  Expert: If you memorized menus, you learned furniture, not carpentry.
• 1:05
  Host: Okay, so what's the carpentry?
• 1:07
  Host: Give me a skill that survives the next model drop.
• 1:10
  Expert: Problem decomposition.
• 1:12
  Expert: The unglamorous ability to take a fuzzy business goal and break it into steps a machine
• 1:16
  Expert: can actually execute.
• 1:19
  Expert: Increase retention is not a prompt.
• 1:21
  Expert: Pull the customers who canceled within 30 days.
• 1:25
  Expert: Cluster them by why they left.
• 1:27
  Expert: Draft three win-back emails per cluster.
• 1:30
  Expert: That's a plan.
• 1:31
  Expert: The model changes.
• 1:32
  Expert: That thinking doesn't.
• 1:34
  Host: That sounds like project management with extra steps.
• 1:37
  Host: Why does it matter more now?
• 1:39
  Expert: Because the models got good enough to punish vagueness in a new way.
• 1:42
  Expert: They'll confidently do the wrong thing beautifully.
• 1:45
  Expert: A 2026 report from GitHub, and yes, they sell the coding assistant so
• 1:50
  Expert: they can integrate accordingly.
• 1:52
  Expert: Found developers accept AI suggestions faster than they verify them.
• 1:57
  Expert: Speed without decomposition just means you ship the mistake sooner.
• 2:01
  Host: You mentioned verification.
• 2:03
  Host: Isn't checking the AI's work exactly the thing everyone's too lazy to do?
• 2:06
  Expert: That's the second durable skill, and it's the one people skip.
• 2:10
  Expert: Call it verification literacy, knowing how to test an answer instead of vibing it.
• 2:15
  Expert: If the model gives you a market size, you ask which sources over what period, and
• 2:20
  Expert: you spot check one number by hand.
• 2:22
  Expert: The professionals keeping their edge treat every output as a claim, not a fact.
• 2:27
  Expert: The tool improves.
• 2:29
  Expert: Your skepticism should not.
• 2:31
  Host: Give me a case where skipping that blew up in someone's face.
• 2:34
  Expert: Lawyers, multiple firms in 2025 filed briefs citing cases the model invented, hallucinated,
• 2:41
  Expert: and got sanctioned by actual judges.
• 2:44
  Expert: The tool wasn't the problem.
• 2:46
  Expert: Nobody ran the one check that mattered.
• 2:48
  Expert: Does this case exist?
• 2:50
  Expert: A paralegal with verification literacy would have caught it in 90 seconds.
• 2:54
  Host: So decomposition, verification, what's the third thing or are we not doing three?
• 2:59
  Expert: We're not doing a tidy list.
• 3:01
  Expert: But there is a third that's underrated, taste.
• 3:04
  Expert: Knowing what good looks like in your field.
• 3:07
  Expert: The model can generate 40 versions of a landing page.
• 3:10
  Expert: It cannot tell you which one your specific customer trusts.
• 3:14
  Expert: That judgment comes from domain experience, and it's the least automatable thing you
• 3:18
  Expert: own.
• 3:19
  Expert: People with taste direct the tool.
• 3:22
  Expert: People without it get directed by it.
• 3:24
  Host: Isn't taste just a fancy word for gatekeeping?
• 3:27
  Host: How does a junior person build it?
• 3:29
  Expert: By doing the boring work anyway.
• 3:31
  Expert: You still write the first draft yourself sometimes, so you know what a bad one feels like.
• 3:36
  Expert: Juniors who let the model do everything from day one never develop the reference point.
• 3:41
  Expert: It's like learning to cook only by ordering takeout.
• 3:44
  Expert: You can't tell when the sauce is off.
• 3:46
  Host: Let's get concrete.
• 3:47
  Host: Someone's listening mid-career worried their AI skills evaporate every quarter.
• 3:52
  Host: What do they actually do Monday morning?
• 3:54
  Expert: Take one recurring task you already use AI for.
• 3:58
  Expert: Before you touch the tool, write out the steps a competent human would take in order.
• 4:03
  Expert: That's your decomposition.
• 4:05
  Expert: Then define in one sentence how you'll know the output is wrong.
• 4:09
  Expert: That's your verification test.
• 4:11
  Expert: Do that for a week and you've built something no model update can delete.
• 4:15
  Host: So stop collecting prompts.
• 4:17
  Host: Start writing your own checklist.
• 4:18
  Expert: Exactly.
• 4:19
  Expert: The prompt is disposable.
• 4:21
  Expert: The thinking behind it is the asset.
• 4:23
  Expert: Guard that and every new tool just makes you faster instead of obsolete.
• 4:28
  Host: That's the note to end on.
• 4:30
  Host: Skills over shortcuts.
• 4:31
  Host: That's a wrap.
• 4:32
  Host: Fresh AI briefings drop daily at MBA-training.com.