WEBVTT
00:00:13.199 --> 00:00:15.279
Picture a familiar moment.
00:00:15.519 --> 00:00:17.120
It's Thursday afternoon.
00:00:17.440 --> 00:00:22.000
A leader looks up from the screen and notices something strange.
00:00:22.320 --> 00:00:24.160
Nobody has knocked today.
00:00:24.399 --> 00:00:28.640
Nobody has sent a message that starts with quick question.
00:00:28.960 --> 00:00:35.359
Nobody has leaned into the doorway with that half sentence, hey, do you have a second?
00:00:35.679 --> 00:00:39.600
For about 10 minutes, it feels wonderful.
00:00:40.079 --> 00:00:42.479
Finally, room to think.
00:00:42.719 --> 00:00:44.560
Then a second thought arrives.
00:00:44.799 --> 00:00:47.520
Quite hard to shake.
00:00:48.000 --> 00:00:50.880
Where did all the questions go?
00:00:51.359 --> 00:00:53.119
Because the work did not stop.
00:00:53.280 --> 00:00:57.920
The projects are moving, the deadlines are real, the questions still exist.
00:00:58.320 --> 00:01:01.359
They are just landing somewhere else now.
00:01:01.600 --> 00:01:09.280
In a chat window, in a tool that never sighs, never looks at the clock, never says, Can we talk about this tomorrow?
00:01:09.519 --> 00:01:13.680
And the leader is left holding a question nobody asked out loud.
00:01:14.079 --> 00:01:20.799
What I am no longer hearing because AI is answering first.
00:01:21.200 --> 00:01:24.000
Welcome to the AI Cafe Conversations Podcast.
00:01:24.079 --> 00:01:25.519
I'm Sahar Andradi.
00:01:25.680 --> 00:01:32.159
I am a neuroleadership coach, and this is where the neuroscience of pressure meets the future of leadership.
00:01:32.400 --> 00:01:36.239
Everything I share comes from my experience and my opinion as a coach.
00:01:36.400 --> 00:01:39.920
It's not a diagnosis and it's not a fact about you.
00:01:40.159 --> 00:01:41.680
You decide what fits.
00:01:41.920 --> 00:01:46.400
Today's question is simple to say and hard to sit with.
00:01:46.640 --> 00:01:52.879
When a team asks AI first, what stops reaching the leader?
00:01:53.200 --> 00:01:57.760
Last week I wrote about the day that turns into answering everyone's questions.
00:01:58.000 --> 00:02:01.439
The day that disappears into gut a minute or just one thing.
00:02:01.599 --> 00:02:04.640
That's a real problem, and many leaders are living it.
00:02:04.799 --> 00:02:07.040
Today is the opposite problem.
00:02:07.280 --> 00:02:10.319
Today is about the day the questions stop arriving.
00:02:10.479 --> 00:02:13.759
Let me share some numbers and let me be careful with them.
00:02:14.000 --> 00:02:26.400
In July of 2026, tech.co reported on an Adobe business survey of more than a thousand full-time US employees who use AI every week.
00:02:26.639 --> 00:02:33.439
When those employees had an obvious question, 68% said they turned to an AI agent for the answer.
00:02:33.759 --> 00:02:37.439
4% said they ask their manager.
00:02:37.919 --> 00:02:39.520
Sit with that for a second.
00:02:39.759 --> 00:02:42.000
68% to 4.
00:02:42.159 --> 00:02:45.120
Now, this is a survey of reported preferences.
00:02:45.280 --> 00:02:47.759
It tells us what people say they do.
00:02:48.159 --> 00:02:51.680
It doesn't follow anyone around the office with a clipboard.
00:02:52.240 --> 00:02:55.919
And it's about obvious questions, the kind with a clear answer.
00:02:56.080 --> 00:03:07.840
Still, if even part of that picture is true on a given team, a lot of small questions that once walked through a leader's door are now doing or going to a machine.
00:03:08.080 --> 00:03:09.520
Here is a second survey.
00:03:09.840 --> 00:03:17.840
MyIQ ran a global survey from June to July 2026, reported by FM magazine in August.
00:03:18.080 --> 00:03:24.879
More than 22,000 employees in 10 regions, almost regular AI users.
00:03:25.120 --> 00:03:32.400
74% said that they now direct questions to a chatbot that they would previously have asked a colleague.
00:03:38.800 --> 00:03:44.000
And 48% reported fewer spontaneous conversations during the workday.
00:03:44.240 --> 00:03:52.000
The report itself says its findings do not establish that AI use Wiccans workplace relationships.
00:03:52.319 --> 00:03:54.719
Fewer spontaneous conversations.
00:03:54.960 --> 00:03:57.360
That one stays with me.
00:03:57.680 --> 00:04:02.960
Because spontaneous conversation is where a lot of leadership information used to live.
00:04:03.199 --> 00:04:08.560
The hallway, the end of the call, the oh, one more thing on the way out.
00:04:08.800 --> 00:04:10.560
And here is a third piece.
00:04:10.800 --> 00:04:20.720
In October 2026, RGP released a study of 500 professionals who actively used AI, surveyed in August and September.
00:04:21.040 --> 00:04:28.959
For complex questions, not just obvious ones, 51% of Gen Z said they turned to AI first.
00:04:29.199 --> 00:04:34.560
Compare that with 19% of millennials and 11% of Gen X.
00:04:34.800 --> 00:04:36.000
And the C-suite?
00:04:36.720 --> 00:04:41.199
65% of C-suite's executives said they turn first to a colleague.
00:04:41.439 --> 00:04:43.279
7% turn to AI.
00:04:43.439 --> 00:04:44.720
So picture the gap.
00:04:44.959 --> 00:04:48.959
The people at the top still walk down the hall and ask a human.
00:04:49.199 --> 00:04:51.839
Parts of the team open a chat window instead.
00:04:52.000 --> 00:04:56.160
Two different habits, at least in what people report.
00:04:56.639 --> 00:04:58.319
Again, honesty first.
00:04:58.720 --> 00:05:00.639
These are reported preferences.
00:05:00.800 --> 00:05:03.600
They do not prove that leaders are losing information.
00:05:03.759 --> 00:05:05.199
Nobody measured that.
00:05:05.439 --> 00:05:09.759
But they point at a question every leader can check for themselves.
00:05:10.000 --> 00:05:12.319
Now let me bring in the human voice.
00:05:12.560 --> 00:05:20.639
A manager in a public online forum wrote this, and I want to read it exactly, verbatim, as written.
00:05:24.160 --> 00:05:27.519
They are managing around me now.
00:05:28.319 --> 00:05:30.160
Read that one more time in your head.
00:05:30.319 --> 00:05:32.800
They are managing around me now.
00:05:33.360 --> 00:05:36.319
One way to hear that sentence, they don't trust you.
00:05:36.560 --> 00:05:39.439
They don't trust me anymore or trust you anymore.
00:05:39.680 --> 00:05:42.800
They are scared or they are avoiding me.
00:05:42.959 --> 00:05:46.399
And that may be true for some teams, it's a real possibility.
00:05:46.639 --> 00:05:52.399
When asking a person feels risky, asking a machine feels safe.
00:05:52.800 --> 00:05:56.399
A chatbot does not remember that you asked a basic question.
00:05:56.560 --> 00:05:58.399
It does not raise an eyebrow.
00:05:58.639 --> 00:06:01.360
It does not bring it up in the performance review.
00:06:01.600 --> 00:06:04.480
But here is the misread I want to name today.
00:06:04.720 --> 00:06:07.120
Quiet does not always mean fear.
00:06:07.279 --> 00:06:09.759
Sometimes quite means convenience.
00:06:10.079 --> 00:06:15.439
AI is fast, AI is patient, AI is there at 11 at night when the leader is asleep.
00:06:15.759 --> 00:06:18.160
AI does not need a calendar invite.
00:06:18.399 --> 00:06:23.519
For an obvious question, of course, people go to the thing that answers in four seconds.
00:06:23.839 --> 00:06:26.480
That is not a judgment of the leader.
00:06:26.720 --> 00:06:29.199
That is just the shortest path.
00:06:29.439 --> 00:06:33.360
So there are two readings of the same silence.
00:06:33.759 --> 00:06:41.839
One reading says the team is afraid to ask, the other says the team found a faster way to get the easy answers.
00:06:42.079 --> 00:06:45.519
I'm not going to tell you which one is true for any particular team.
00:06:45.600 --> 00:06:47.519
I cannot know that from here.
00:06:47.600 --> 00:06:54.879
And honestly, a leader often cannot know it from the insider either, at least not right away.
00:06:55.120 --> 00:07:02.319
What I can say is this: in my experience, either way, the leader's early signal can dry up.
00:07:02.560 --> 00:07:08.000
Because the questions were never really just questions, they were information.
00:07:08.560 --> 00:07:22.560
A question tells a leader what someone is working on, where they are stuck, or what confuses them, or even what they are worried about, what they think the leader cares about.
00:07:22.879 --> 00:07:27.920
A steady stream of questions is a live feed of the team's reality.
00:07:28.319 --> 00:07:31.680
When that feed goes quiet, the leader doesn't get an alert.
00:07:31.839 --> 00:07:32.959
Nothing pinks.
00:07:33.120 --> 00:07:37.920
There is no notification that says three questions you would have needed did not arrive this week.
00:07:38.240 --> 00:07:43.519
There is just a slightly quieter inbox and a slightly calmer afternoon.
00:07:43.839 --> 00:07:45.279
That is the crisis.
00:07:45.519 --> 00:07:48.480
Not that AI answers question, that's useful.
00:07:48.720 --> 00:07:55.920
The crisis is that the leader can lose a source of early warning without ever noticing it left.
00:07:56.560 --> 00:07:58.480
What does that actually cost?
00:07:58.720 --> 00:08:00.480
Let me walk you through it.
00:08:00.720 --> 00:08:04.319
Outward, the way it tends to show up in organizations.
00:08:04.800 --> 00:08:06.879
The first cost is timing.
00:08:07.120 --> 00:08:11.519
In my experience, small questions often carry early warnings.
00:08:11.839 --> 00:08:15.040
Is it normal that the vendor change the delivery date?
00:08:15.279 --> 00:08:17.600
Should this number look like this?
00:08:17.920 --> 00:08:21.120
Who owns this now that the project moved?
00:08:21.439 --> 00:08:23.279
None of those sound like a crisis.
00:08:23.360 --> 00:08:25.120
They sound like housekeeping.
00:08:25.360 --> 00:08:31.920
But sometimes they are the first visible edge of a problem that will be big in six weeks.
00:08:32.240 --> 00:08:36.480
When those questions go to AI, the person may get a reasonable answer.
00:08:36.720 --> 00:08:39.919
The tool may say, delivery days change often.
00:08:40.159 --> 00:08:42.080
Here is how to update the plan.
00:08:42.480 --> 00:08:43.120
Fine.
00:08:43.759 --> 00:08:45.039
Accurate even.
00:08:45.360 --> 00:08:51.759
But the leader who knows that this vendor changed it twice, already discorder, never hears it.
00:08:52.000 --> 00:08:53.759
The leader had the contacts.
00:08:54.000 --> 00:08:56.159
The machine had the question.
00:08:56.559 --> 00:08:57.759
They never met.
00:08:58.159 --> 00:09:00.320
So the cost is not a wrong answer.
00:09:00.480 --> 00:09:05.200
The cost is that the right person never saw the right question at the right time.
00:09:05.600 --> 00:09:07.679
The second cost is texture.
00:09:07.919 --> 00:09:12.879
Leaders learn a lot from how people ask, not just what they ask.
00:09:13.200 --> 00:09:17.679
The hesitation, the this might be a dumb question.
00:09:17.919 --> 00:09:21.360
The person who asks the same kind of question three weeks in a row.
00:09:21.679 --> 00:09:30.559
A leader who hears these things starts to see who is stretched, who is new to something, who is quietly carrying more than the raw.
00:09:31.120 --> 00:09:35.200
That texture is hard to get from a dashboard.
00:09:35.440 --> 00:09:37.919
It mostly comes from the questions.
00:09:38.240 --> 00:09:41.120
The third cost is the second opinion.
00:09:41.360 --> 00:09:50.879
Remember that my IQ figure, 59% of regular AI users said they asked colleagues for second opinions less often.
00:09:51.360 --> 00:09:57.279
A second opinion is not only about getting a better answer, it's how a team checks itself.
00:09:57.679 --> 00:10:03.039
It's how one person's blind spot gets caught by another person's experience.
00:10:03.279 --> 00:10:08.720
If more of those checks go to a machine, the team may still get answers.
00:10:08.960 --> 00:10:13.919
What it may lose, in my opinion, is the habit of checking each other.
00:10:14.399 --> 00:10:21.360
The fourth cost is the lowest and the hardest to see is the leader's map.
00:10:21.679 --> 00:10:25.519
Every leader carries a map of the team in their head.
00:10:25.840 --> 00:10:26.879
Who knows what?
00:10:27.039 --> 00:10:30.399
Who is stuck, what is moving, what is not.
00:10:30.720 --> 00:10:34.799
That map gets updated mostly by the questions that come in.
00:10:35.039 --> 00:10:41.919
When fewer questions come in, the map does not erase itself, it just stops updating.
00:10:43.039 --> 00:10:51.360
And in my experience, a leader reading an old map can feel just as confident as a leader reading a new one.
00:10:52.000 --> 00:10:54.879
That is the part I want leaders to hear.
00:10:55.120 --> 00:11:02.720
In my experience, confidence does not drop when information drops, it often stays right where it was.
00:11:03.039 --> 00:11:10.000
So a leader can be making decisions from a map from three months ago, the landscape changed.
00:11:10.639 --> 00:11:17.279
They feel grounded, they feel informed, but they are quietly behind.
00:11:17.919 --> 00:11:20.480
And under pressure, this gets worse.
00:11:22.799 --> 00:11:26.000
A leader under load has less attention to spare.
00:11:26.080 --> 00:11:34.080
When the calendar is full and the stakes are high, a quiet inbox feels or looks like a gift, not a gap.
00:11:34.320 --> 00:11:38.240
Nobody under strain goes looking for problems that did not knock.
00:11:38.480 --> 00:11:42.000
That is not a discipline problem, it's a load problem.
00:11:42.159 --> 00:11:48.240
The leader is not careless, the leader is busy, and silence does not ask for attention.
00:11:48.559 --> 00:11:50.720
Let me be fair on the other side too.
00:11:50.879 --> 00:11:52.799
Some of this quiet is pure game.
00:11:52.960 --> 00:11:54.559
People get unstuck faster.
00:11:54.720 --> 00:11:58.000
They stop waiting two hours for a reply to an easy question.
00:11:58.240 --> 00:11:59.519
They learn on their own.
00:11:59.679 --> 00:12:05.600
A team that can answer its own routine question in many ways is a stronger team.
00:12:05.840 --> 00:12:09.600
So the cost is not AI is bad for teams.
00:12:09.759 --> 00:12:11.360
That's not what I'm saying here.
00:12:11.600 --> 00:12:14.159
The cost is narrower and sharper.
00:12:14.399 --> 00:12:22.559
The leader can lose the small, odd, early signals, and those were never the routine ones.
00:12:23.039 --> 00:12:26.639
Now, why is this so easy to miss?
00:12:26.960 --> 00:12:32.159
Why would a smart, experienced leader not notice that something stopped arriving?
00:12:32.399 --> 00:12:35.919
There is one idea from behavioral science that helps here.
00:12:36.000 --> 00:12:39.279
It's called the feature positive effect.
00:12:39.840 --> 00:12:41.360
And here is what it is.
00:12:41.600 --> 00:12:49.759
Across many studies, people and animals too learn from what is present faster than from what is absent.
00:12:50.320 --> 00:12:56.320
They learn from what is present faster than what is absent.
00:12:57.039 --> 00:13:04.559
The classic work goes back to Newman, Wolfe, and Hearst in 1980, and it has been replicated since.
00:13:04.720 --> 00:13:09.039
In simple terms, when a signal shows up, we pick up on it.
00:13:09.279 --> 00:13:14.399
When a signal is missing, we are much slower to learn anything from that missing piece.
00:13:14.559 --> 00:13:16.799
It's actually one of the biases that we have.
00:13:16.960 --> 00:13:19.120
We call that the availability bias.
00:13:19.279 --> 00:13:27.440
Like, for example, when we ask a question, what is more dangerous of a job, being a cop or being a logger in the forest?
00:13:27.679 --> 00:13:31.600
Most people will answer very quickly a cop.
00:13:32.159 --> 00:13:36.080
Actually, the logger in the forest is a more dangerous job.
00:13:36.240 --> 00:13:39.600
But why they jump into being a cop of police officer?
00:13:39.759 --> 00:13:50.159
Because of the availability of the information that we always have about cops and police officers being in danger all the time on the news and everywhere we go.
00:13:50.320 --> 00:13:52.320
So it's called an availability bias.
00:13:52.480 --> 00:13:58.159
So think of it like this: if a smoke alarm goes off, everyone notices.
00:13:58.559 --> 00:14:04.159
If the sound of the dishwasher stops, it can take a long time for anyone to realize it's quiet.
00:14:04.639 --> 00:14:08.399
Presence grabs us, absence slips by.
00:14:08.879 --> 00:14:11.440
There is one more finding I find generally hopeful.
00:14:11.759 --> 00:14:18.320
A study by Rasen in 2014 found that simply alerting people to this tendency reduced it.
00:14:18.559 --> 00:14:23.360
Just alerting people to the then the tendency reduced it.
00:14:23.600 --> 00:14:26.799
That is part of why I'm telling you about it today.
00:14:26.960 --> 00:14:32.080
Now, the honest limit, and I wanted to say this out loud, these studies were lab learning tasks.
00:14:32.240 --> 00:14:36.639
They were not about leaders, they were not about teams, they were not about AI.
00:14:36.879 --> 00:14:46.000
Applying the feature positive effect to a leader whose team has started asking AI first is my inference, not their finding.
00:14:46.240 --> 00:14:51.200
And that is why, or here is why I think the inference is useful.
00:14:51.440 --> 00:14:53.679
A question that arrives is a feature.
00:14:53.840 --> 00:14:56.240
It's present, a leader can react to it.
00:14:56.480 --> 00:15:00.000
A question that does not arrive is an absence.
00:15:00.240 --> 00:15:02.240
There is nothing to react to.
00:15:02.480 --> 00:15:08.240
The research suggests that people tend to be slower to learn from that kind of absence.
00:15:08.480 --> 00:15:14.799
So a missing question may be exactly the kind of thing that is easy not to notice.
00:15:15.519 --> 00:15:20.320
That is the whole mechanism, or that's what all the mechanism needs to explain.
00:15:20.559 --> 00:15:23.360
It does not explain why a team went quiet.
00:15:23.440 --> 00:15:26.240
It does not tell us whether it's fear or convenience.
00:15:26.399 --> 00:15:32.639
It only explains why the leaders may not notice the quiet in the first place.
00:15:32.799 --> 00:15:38.399
And if the recent finding holds in real life, and again, that's a leap, okay?
00:15:38.720 --> 00:15:41.519
Then naming it is the first step.
00:15:41.919 --> 00:15:48.639
Leader who knows that absence is easy to miss can start looking for it on purpose.
00:15:48.960 --> 00:15:50.720
That's what the solution is built on.
00:15:50.879 --> 00:15:58.480
Not more monitoring, not surveillance of anyone's chat history, just deliberately looking for what is missing.
00:15:58.960 --> 00:16:00.879
So, what can a leader actually do?
00:16:01.120 --> 00:16:02.000
Four moves.
00:16:02.159 --> 00:16:11.600
Each one is small, each one involves AI directly, because AI is now part of how the team works and pretending otherwise is not a plan.
00:16:11.919 --> 00:16:13.120
Move number one.
00:16:13.840 --> 00:16:18.960
Once a week, write down three questions that used to reach you and did not this week.
00:16:19.120 --> 00:16:19.600
That's it.
00:16:19.759 --> 00:16:20.240
Three.
00:16:20.399 --> 00:16:23.759
On a paper or in a note on your laptop.
00:16:24.000 --> 00:16:27.279
Friday afternoon works for many leaders.
00:16:27.440 --> 00:16:28.720
Here is my thinking.
00:16:29.039 --> 00:16:33.759
If absence is easy to miss, the leader makes absence visible.
00:16:34.000 --> 00:16:37.759
Writing it down turns a missing thing into a present thing.
00:16:38.000 --> 00:16:40.159
Suddenly, there is a list to look at.
00:16:40.399 --> 00:16:41.919
What might be on that list?
00:16:42.159 --> 00:16:47.039
Something like nobody asked me how to handle the pricing exception this week.
00:16:47.279 --> 00:16:51.039
Or nobody checked with me before sending the client update.
00:16:51.279 --> 00:16:57.440
Or the new analyst used to ask me about the reporting format every Tuesday, not this week.
00:16:57.679 --> 00:17:00.080
Some weeks the list will be boring.
00:17:00.320 --> 00:17:00.960
Good.
00:17:01.519 --> 00:17:04.240
That may mean mostly routine questions moved.
00:17:04.400 --> 00:17:08.720
Other weeks, one item on the list will make the leader sit up.
00:17:08.960 --> 00:17:12.400
That is the week the habit paid for itself.
00:17:12.640 --> 00:17:19.279
A practical tip: if writing three feels hard, that itself is information.
00:17:19.519 --> 00:17:24.000
It may mean the leader's map is already a little old.
00:17:24.960 --> 00:17:26.160
Move number two.
00:17:26.799 --> 00:17:28.480
Ask one person plainly.
00:17:28.720 --> 00:17:32.240
What did you ask AI this week that you used to ask me?
00:17:32.559 --> 00:17:35.039
Say it with curiosity, not judgment.
00:17:35.200 --> 00:17:37.279
That is not a gotcha moment.
00:17:37.519 --> 00:17:39.119
It's not a compliance check.
00:17:39.200 --> 00:17:41.279
It's not an honest question.
00:17:41.519 --> 00:17:46.400
Here is what it can sound like word for word at the start of a one-on-one.
00:17:47.279 --> 00:17:48.640
Quick one before we start.
00:17:48.799 --> 00:17:50.559
I'm curious, not checking up.
00:17:50.720 --> 00:17:54.000
What did you ask AI this week that you used to ask me?
00:17:54.160 --> 00:18:03.200
I know you're trying to help me and relieve some of the work that I do, and I appreciate it, but I just want to make sure that we're not missing anything.
00:18:03.440 --> 00:18:04.480
Then stop talking.
00:18:04.640 --> 00:18:05.680
Let them answer.
00:18:05.920 --> 00:18:08.000
Often the answer will be easy and useful.
00:18:08.160 --> 00:18:14.559
All the formatting for the board deck or how the word, uh, how to word the reply to that supplier.
00:18:14.799 --> 00:18:15.200
Great.
00:18:15.359 --> 00:18:18.079
Now the leader knows what kinds of question moved.
00:18:18.240 --> 00:18:20.559
Sometimes the answer will be more interesting.
00:18:20.720 --> 00:18:25.119
Like, I asked it how to tell a client we were going to miss the date.
00:18:25.440 --> 00:18:27.759
That is a question with weight.
00:18:28.000 --> 00:18:31.119
That is a question the leader may want to know about.
00:18:31.279 --> 00:18:35.920
And now the leader does just because they asked.
00:18:36.240 --> 00:18:37.680
One more thing about this move.
00:18:37.839 --> 00:18:42.720
Asking it tells the person that bringing questions to the leader, it's still welcome.
00:18:43.039 --> 00:18:44.960
The question itself sends a message.
00:18:45.119 --> 00:18:46.880
I still want to hear from you.
00:18:47.200 --> 00:18:49.359
Move number three.
00:18:49.599 --> 00:18:52.319
Listen for which kind of quiet it is.
00:18:52.480 --> 00:18:55.440
Remember the two readings convenience or caution.
00:18:55.599 --> 00:18:57.680
Convenience sounds like this.
00:18:58.079 --> 00:18:59.200
It was faster.
00:18:59.440 --> 00:19:02.480
Or I did not want to bother you with something that small.
00:19:02.720 --> 00:19:04.480
Or it was 10 at night.
00:19:04.720 --> 00:19:06.799
That is the tool doing its job.
00:19:07.039 --> 00:19:10.000
AI is fast, patient, always there.
00:19:10.160 --> 00:19:11.440
That is a good thing.
00:19:11.680 --> 00:19:15.119
When it's convenience, respond with thanks and curiosity.
00:19:15.279 --> 00:19:17.839
Something like, that's great, honestly.
00:19:18.160 --> 00:19:20.079
What did you learn from it?
00:19:20.400 --> 00:19:22.000
Anything surprising?
00:19:22.319 --> 00:19:29.680
The leader stays connected to what the person is learning without pulling them back into asking permission for small things.
00:19:30.000 --> 00:19:31.839
Caution sounds different.
00:19:32.079 --> 00:19:35.440
Something like, I wasn't sure how you would take it.
00:19:35.759 --> 00:19:38.960
Or I figured it was something I should already know.
00:19:39.519 --> 00:19:43.119
Or it felt safer to just check first.
00:19:43.279 --> 00:19:48.960
That is a question that went on a machine because the machine felt less risky than the person.
00:19:49.359 --> 00:19:54.160
When it's a caution, the move is to make the next question easier to bring.
00:19:55.119 --> 00:20:01.680
Not with a speech, with a simple, specific invitation like the following.
00:20:02.480 --> 00:20:05.680
Bring me the have form ones too.
00:20:05.920 --> 00:20:12.079
I would rather hear a messy question early than a polished one late.
00:20:12.640 --> 00:20:14.240
That sentence does a lot.
00:20:14.319 --> 00:20:15.759
It lowers the bar.
00:20:16.079 --> 00:20:25.759
It says unfinished thinking is welcome and it gives the person a concrete permission, not a vague my door is always open kind of permission.
00:20:26.000 --> 00:20:29.440
Which in my experience, many people quietly do not believe.
00:20:29.599 --> 00:20:33.519
Notice that the leader is not deciding in advance which kind of quiet it is.
00:20:33.680 --> 00:20:36.960
The leader is listening for it, one person at a time.
00:20:37.200 --> 00:20:43.920
Some people on the same team will be convenience, some may be caution, both can be true in one room.
00:20:44.079 --> 00:20:45.440
Now, move number four.
00:20:45.680 --> 00:20:48.960
Keep one channel that only a person can fill.
00:20:49.279 --> 00:20:50.799
AI can answer a lot.
00:20:50.960 --> 00:20:56.240
It can explain a policy, draft an email, check a formula, summarize a meeting.
00:20:56.480 --> 00:21:05.279
What it cannot do is tell a leader what a specific person is worried about unless that person says it to the leader.
00:21:05.759 --> 00:21:09.279
So protect one question that AI cannot answer for them.
00:21:09.440 --> 00:21:12.960
A simple version at the end of every one-on-one.
00:21:13.519 --> 00:21:14.480
Ask this.
00:21:24.640 --> 00:21:28.240
What worries me is a human question.
00:21:28.559 --> 00:21:34.160
It invites the odd thing, the small thing, the thing that does not fit in a project record.
00:21:34.400 --> 00:21:36.400
Here is how that might go.
00:21:36.559 --> 00:21:40.240
The leader asks, before we wrap, what worries you right now?
00:21:40.400 --> 00:21:43.279
The person says, honestly, nothing big.
00:21:43.599 --> 00:21:50.799
Maybe that the new process is going to double the work for the support team, but I'm probably overthinking it.
00:21:51.119 --> 00:21:59.759
That last line, probably overthinking it, is exactly the kind of early signal that used to arrive in a hallway question.
00:22:00.799 --> 00:22:07.920
Now it has a home, a regular, predictable place where only one person can put it.
00:22:08.079 --> 00:22:12.079
So let me put the four moves together so they are easy to carry.
00:22:12.240 --> 00:22:16.079
One, weakly write down three questions that did not arrive.
00:22:16.319 --> 00:22:21.759
Two, ask one person what they asked AI that they used to ask the leader.
00:22:21.920 --> 00:22:26.960
Three, listen for convenience or caution and answer each one differently.
00:22:27.119 --> 00:22:32.400
Four, keep one human-only channel like what worries you at the end of every one-on-one.
00:22:33.759 --> 00:22:36.480
None of this is actually anti-AI.
00:22:36.720 --> 00:22:41.519
In fact, it takes AI for what it is, a real part of how work happens now.
00:22:41.680 --> 00:22:46.960
The team is going to keep asking AI first for a lot of things, and in many cases they should.
00:22:47.200 --> 00:22:49.519
The goal is not to pull those questions back.
00:22:49.680 --> 00:22:54.079
The goal is to make sure the leader is still hearing the one that matters.
00:22:54.319 --> 00:22:58.319
In my opinion, that is what a grounded leader does with AI in the room.
00:22:58.480 --> 00:23:06.000
Not compete with it, not ignore it, just stay curious about what is now carrying the cues to come to them.
00:23:06.400 --> 00:23:10.880
Now you might be thinking, good, fewer interruptions, isn't that exactly what I wanted?
00:23:11.119 --> 00:23:13.839
That's a fair objection and it's a strong one.
00:23:14.079 --> 00:23:19.680
Last week's piece was about how answering everyone's question can swallow a leader's day.
00:23:19.839 --> 00:23:23.119
So when the questions slow down, shouldn't a leader celebrate?
00:23:23.359 --> 00:23:24.319
Partly yes.
00:23:24.559 --> 00:23:27.039
For the routine questions, absolutely.
00:23:27.279 --> 00:23:37.599
If AI is handling what the temp what is the template for this and how do I format that, the leader gets time back and the team gets answers faster.
00:23:37.759 --> 00:23:40.799
That is a win, and I'm not going to pretend otherwise.
00:23:41.039 --> 00:23:42.400
But here is the problem.
00:23:42.640 --> 00:23:45.839
The questions did not leave in a neat order.
00:23:46.160 --> 00:23:50.960
AI did not take only the boring ones and leave the important ones behind.
00:23:51.279 --> 00:23:57.200
People choose where to send each question, one at a time, often in a hurry.
00:23:57.440 --> 00:24:05.200
Some of the small odd ones went too, and those small odd ones are the ones that used to carry early warnings.
00:24:05.599 --> 00:24:09.279
Is it normal that this customer emailed three times this week?
00:24:09.440 --> 00:24:11.039
That sounds routine.
00:24:11.440 --> 00:24:16.720
It might be routine, or it might be the first sign of an account in trouble.
00:24:16.960 --> 00:24:20.400
And this is where the feature positive effect comes back in.
00:24:20.559 --> 00:24:24.640
If the routine questions disappear, a leader notices the relief.
00:24:24.799 --> 00:24:27.279
Relief is present, it's easy to feel.
00:24:27.519 --> 00:24:33.519
But if one of the odd early warning questions also disappears, the leader feels nothing.
00:24:34.000 --> 00:24:36.640
There is no gap to feel.
00:24:36.880 --> 00:24:39.279
Absence is hard to notice.
00:24:39.519 --> 00:24:44.240
The research suggests people tend to learn more slowly from what is missing.
00:24:44.799 --> 00:24:50.319
And while applying that to leaders, in my inference, it is a reasonable warning.
00:24:50.799 --> 00:24:59.759
So the honest answer to is not this what I wanted is a leader wants fewer interruptions, yes, not fewer warnings.
00:25:00.079 --> 00:25:03.759
And from the inside, those two look exactly the same.
00:25:03.920 --> 00:25:06.160
That is why the weekly list matters.
00:25:06.240 --> 00:25:10.559
That is how a leader keeps the relief and still checks the gap.
00:25:10.880 --> 00:25:15.200
Today was about what stops reaching the leader, but there is a mirror question.
00:25:15.440 --> 00:25:23.599
On Friday, we asked why a team can't hear the leader in a perfect college AI written message.
00:25:24.720 --> 00:25:28.720
If this conversation is lending for you, I want to invite you to one thing.
00:25:29.039 --> 00:25:31.359
Register for the pressure room.
00:25:31.519 --> 00:25:36.559
It's a free live session for executives, 45 minutes on Zoom.
00:25:36.640 --> 00:25:41.759
The next one is Thursday, October 29 at 11 Pacific a.m.
00:25:42.160 --> 00:25:49.359
The session title is Why Does a Small Problem Suddenly Feels Like an Emergency Under Pressure?
00:25:49.599 --> 00:25:55.680
You can register at saharandradi.com slash pressure dash rule.
00:25:55.920 --> 00:25:58.799
So this week look for what did not arrive.
00:25:59.039 --> 00:26:03.119
Write down three questions, ask one person what they asked.
00:26:03.440 --> 00:26:06.799
AI, listen for which kind of quite it is.
00:26:07.200 --> 00:26:10.559
The quite is information too.
00:26:10.799 --> 00:26:12.480
It just does not knock.
00:26:12.720 --> 00:26:14.480
Leadership does not fail.
00:26:14.799 --> 00:26:17.839
Systems, nervous systems do.
00:26:18.240 --> 00:26:22.319
As usual, before I wrap up, I want to ask you to show me some love.
00:26:22.720 --> 00:26:26.000
Rate, comment, save, share the podcast.
00:26:26.160 --> 00:26:31.359
Thank you for your support for making us one of the top 2% global podcasts that exist.
00:26:31.440 --> 00:26:32.880
I really appreciate you.
00:26:33.039 --> 00:26:36.880
Share it so we can reach more people that need to hear that.
00:26:37.119 --> 00:26:38.559
This is Sahara Andradi.
00:26:38.799 --> 00:26:43.440
Till we meet again at AI Cafe Conversation Podcast.
00:26:43.680 --> 00:26:44.640
Peace out.
00:00:13.199 --> 00:00:15.279
Picture a familiar moment.
00:00:15.519 --> 00:00:17.120
It's Thursday afternoon.
00:00:17.440 --> 00:00:22.000
A leader looks up from the screen and notices something strange.
00:00:22.320 --> 00:00:24.160
Nobody has knocked today.
00:00:24.399 --> 00:00:28.640
Nobody has sent a message that starts with quick question.
00:00:28.960 --> 00:00:35.359
Nobody has leaned into the doorway with that half sentence, hey, do you have a second?
00:00:35.679 --> 00:00:39.600
For about 10 minutes, it feels wonderful.
00:00:40.079 --> 00:00:42.479
Finally, room to think.
00:00:42.719 --> 00:00:44.560
Then a second thought arrives.
00:00:44.799 --> 00:00:47.520
Quite hard to shake.
00:00:48.000 --> 00:00:50.880
Where did all the questions go?
00:00:51.359 --> 00:00:53.119
Because the work did not stop.
00:00:53.280 --> 00:00:57.920
The projects are moving, the deadlines are real, the questions still exist.
00:00:58.320 --> 00:01:01.359
They are just landing somewhere else now.
00:01:01.600 --> 00:01:09.280
In a chat window, in a tool that never sighs, never looks at the clock, never says, Can we talk about this tomorrow?
00:01:09.519 --> 00:01:13.680
And the leader is left holding a question nobody asked out loud.
00:01:14.079 --> 00:01:20.799
What I am no longer hearing because AI is answering first.
00:01:21.200 --> 00:01:24.000
Welcome to the AI Cafe Conversations Podcast.
00:01:24.079 --> 00:01:25.519
I'm Sahar Andradi.
00:01:25.680 --> 00:01:32.159
I am a neuroleadership coach, and this is where the neuroscience of pressure meets the future of leadership.
00:01:32.400 --> 00:01:36.239
Everything I share comes from my experience and my opinion as a coach.
00:01:36.400 --> 00:01:39.920
It's not a diagnosis and it's not a fact about you.
00:01:40.159 --> 00:01:41.680
You decide what fits.
00:01:41.920 --> 00:01:46.400
Today's question is simple to say and hard to sit with.
00:01:46.640 --> 00:01:52.879
When a team asks AI first, what stops reaching the leader?
00:01:53.200 --> 00:01:57.760
Last week I wrote about the day that turns into answering everyone's questions.
00:01:58.000 --> 00:02:01.439
The day that disappears into gut a minute or just one thing.
00:02:01.599 --> 00:02:04.640
That's a real problem, and many leaders are living it.
00:02:04.799 --> 00:02:07.040
Today is the opposite problem.
00:02:07.280 --> 00:02:10.319
Today is about the day the questions stop arriving.
00:02:10.479 --> 00:02:13.759
Let me share some numbers and let me be careful with them.
00:02:14.000 --> 00:02:26.400
In July of 2026, tech.co reported on an Adobe business survey of more than a thousand full-time US employees who use AI every week.
00:02:26.639 --> 00:02:33.439
When those employees had an obvious question, 68% said they turned to an AI agent for the answer.
00:02:33.759 --> 00:02:37.439
4% said they ask their manager.
00:02:37.919 --> 00:02:39.520
Sit with that for a second.
00:02:39.759 --> 00:02:42.000
68% to 4.
00:02:42.159 --> 00:02:45.120
Now, this is a survey of reported preferences.
00:02:45.280 --> 00:02:47.759
It tells us what people say they do.
00:02:48.159 --> 00:02:51.680
It doesn't follow anyone around the office with a clipboard.
00:02:52.240 --> 00:02:55.919
And it's about obvious questions, the kind with a clear answer.
00:02:56.080 --> 00:03:07.840
Still, if even part of that picture is true on a given team, a lot of small questions that once walked through a leader's door are now doing or going to a machine.
00:03:08.080 --> 00:03:09.520
Here is a second survey.
00:03:09.840 --> 00:03:17.840
MyIQ ran a global survey from June to July 2026, reported by FM magazine in August.
00:03:18.080 --> 00:03:24.879
More than 22,000 employees in 10 regions, almost regular AI users.
00:03:25.120 --> 00:03:32.400
74% said that they now direct questions to a chatbot that they would previously have asked a colleague.
00:03:38.800 --> 00:03:44.000
And 48% reported fewer spontaneous conversations during the workday.
00:03:44.240 --> 00:03:52.000
The report itself says its findings do not establish that AI use Wiccans workplace relationships.
00:03:52.319 --> 00:03:54.719
Fewer spontaneous conversations.
00:03:54.960 --> 00:03:57.360
That one stays with me.
00:03:57.680 --> 00:04:02.960
Because spontaneous conversation is where a lot of leadership information used to live.
00:04:03.199 --> 00:04:08.560
The hallway, the end of the call, the oh, one more thing on the way out.
00:04:08.800 --> 00:04:10.560
And here is a third piece.
00:04:10.800 --> 00:04:20.720
In October 2026, RGP released a study of 500 professionals who actively used AI, surveyed in August and September.
00:04:21.040 --> 00:04:28.959
For complex questions, not just obvious ones, 51% of Gen Z said they turned to AI first.
00:04:29.199 --> 00:04:34.560
Compare that with 19% of millennials and 11% of Gen X.
00:04:34.800 --> 00:04:36.000
And the C-suite?
00:04:36.720 --> 00:04:41.199
65% of C-suite's executives said they turn first to a colleague.
00:04:41.439 --> 00:04:43.279
7% turn to AI.
00:04:43.439 --> 00:04:44.720
So picture the gap.
00:04:44.959 --> 00:04:48.959
The people at the top still walk down the hall and ask a human.
00:04:49.199 --> 00:04:51.839
Parts of the team open a chat window instead.
00:04:52.000 --> 00:04:56.160
Two different habits, at least in what people report.
00:04:56.639 --> 00:04:58.319
Again, honesty first.
00:04:58.720 --> 00:05:00.639
These are reported preferences.
00:05:00.800 --> 00:05:03.600
They do not prove that leaders are losing information.
00:05:03.759 --> 00:05:05.199
Nobody measured that.
00:05:05.439 --> 00:05:09.759
But they point at a question every leader can check for themselves.
00:05:10.000 --> 00:05:12.319
Now let me bring in the human voice.
00:05:12.560 --> 00:05:20.639
A manager in a public online forum wrote this, and I want to read it exactly, verbatim, as written.
00:05:24.160 --> 00:05:27.519
They are managing around me now.
00:05:28.319 --> 00:05:30.160
Read that one more time in your head.
00:05:30.319 --> 00:05:32.800
They are managing around me now.
00:05:33.360 --> 00:05:36.319
One way to hear that sentence, they don't trust you.
00:05:36.560 --> 00:05:39.439
They don't trust me anymore or trust you anymore.
00:05:39.680 --> 00:05:42.800
They are scared or they are avoiding me.
00:05:42.959 --> 00:05:46.399
And that may be true for some teams, it's a real possibility.
00:05:46.639 --> 00:05:52.399
When asking a person feels risky, asking a machine feels safe.
00:05:52.800 --> 00:05:56.399
A chatbot does not remember that you asked a basic question.
00:05:56.560 --> 00:05:58.399
It does not raise an eyebrow.
00:05:58.639 --> 00:06:01.360
It does not bring it up in the performance review.
00:06:01.600 --> 00:06:04.480
But here is the misread I want to name today.
00:06:04.720 --> 00:06:07.120
Quiet does not always mean fear.
00:06:07.279 --> 00:06:09.759
Sometimes quite means convenience.
00:06:10.079 --> 00:06:15.439
AI is fast, AI is patient, AI is there at 11 at night when the leader is asleep.
00:06:15.759 --> 00:06:18.160
AI does not need a calendar invite.
00:06:18.399 --> 00:06:23.519
For an obvious question, of course, people go to the thing that answers in four seconds.
00:06:23.839 --> 00:06:26.480
That is not a judgment of the leader.
00:06:26.720 --> 00:06:29.199
That is just the shortest path.
00:06:29.439 --> 00:06:33.360
So there are two readings of the same silence.
00:06:33.759 --> 00:06:41.839
One reading says the team is afraid to ask, the other says the team found a faster way to get the easy answers.
00:06:42.079 --> 00:06:45.519
I'm not going to tell you which one is true for any particular team.
00:06:45.600 --> 00:06:47.519
I cannot know that from here.
00:06:47.600 --> 00:06:54.879
And honestly, a leader often cannot know it from the insider either, at least not right away.
00:06:55.120 --> 00:07:02.319
What I can say is this: in my experience, either way, the leader's early signal can dry up.
00:07:02.560 --> 00:07:08.000
Because the questions were never really just questions, they were information.
00:07:08.560 --> 00:07:22.560
A question tells a leader what someone is working on, where they are stuck, or what confuses them, or even what they are worried about, what they think the leader cares about.
00:07:22.879 --> 00:07:27.920
A steady stream of questions is a live feed of the team's reality.
00:07:28.319 --> 00:07:31.680
When that feed goes quiet, the leader doesn't get an alert.
00:07:31.839 --> 00:07:32.959
Nothing pinks.
00:07:33.120 --> 00:07:37.920
There is no notification that says three questions you would have needed did not arrive this week.
00:07:38.240 --> 00:07:43.519
There is just a slightly quieter inbox and a slightly calmer afternoon.
00:07:43.839 --> 00:07:45.279
That is the crisis.
00:07:45.519 --> 00:07:48.480
Not that AI answers question, that's useful.
00:07:48.720 --> 00:07:55.920
The crisis is that the leader can lose a source of early warning without ever noticing it left.
00:07:56.560 --> 00:07:58.480
What does that actually cost?
00:07:58.720 --> 00:08:00.480
Let me walk you through it.
00:08:00.720 --> 00:08:04.319
Outward, the way it tends to show up in organizations.
00:08:04.800 --> 00:08:06.879
The first cost is timing.
00:08:07.120 --> 00:08:11.519
In my experience, small questions often carry early warnings.
00:08:11.839 --> 00:08:15.040
Is it normal that the vendor change the delivery date?
00:08:15.279 --> 00:08:17.600
Should this number look like this?
00:08:17.920 --> 00:08:21.120
Who owns this now that the project moved?
00:08:21.439 --> 00:08:23.279
None of those sound like a crisis.
00:08:23.360 --> 00:08:25.120
They sound like housekeeping.
00:08:25.360 --> 00:08:31.920
But sometimes they are the first visible edge of a problem that will be big in six weeks.
00:08:32.240 --> 00:08:36.480
When those questions go to AI, the person may get a reasonable answer.
00:08:36.720 --> 00:08:39.919
The tool may say, delivery days change often.
00:08:40.159 --> 00:08:42.080
Here is how to update the plan.
00:08:42.480 --> 00:08:43.120
Fine.
00:08:43.759 --> 00:08:45.039
Accurate even.
00:08:45.360 --> 00:08:51.759
But the leader who knows that this vendor changed it twice, already discorder, never hears it.
00:08:52.000 --> 00:08:53.759
The leader had the contacts.
00:08:54.000 --> 00:08:56.159
The machine had the question.
00:08:56.559 --> 00:08:57.759
They never met.
00:08:58.159 --> 00:09:00.320
So the cost is not a wrong answer.
00:09:00.480 --> 00:09:05.200
The cost is that the right person never saw the right question at the right time.
00:09:05.600 --> 00:09:07.679
The second cost is texture.
00:09:07.919 --> 00:09:12.879
Leaders learn a lot from how people ask, not just what they ask.
00:09:13.200 --> 00:09:17.679
The hesitation, the this might be a dumb question.
00:09:17.919 --> 00:09:21.360
The person who asks the same kind of question three weeks in a row.
00:09:21.679 --> 00:09:30.559
A leader who hears these things starts to see who is stretched, who is new to something, who is quietly carrying more than the raw.
00:09:31.120 --> 00:09:35.200
That texture is hard to get from a dashboard.
00:09:35.440 --> 00:09:37.919
It mostly comes from the questions.
00:09:38.240 --> 00:09:41.120
The third cost is the second opinion.
00:09:41.360 --> 00:09:50.879
Remember that my IQ figure, 59% of regular AI users said they asked colleagues for second opinions less often.
00:09:51.360 --> 00:09:57.279
A second opinion is not only about getting a better answer, it's how a team checks itself.
00:09:57.679 --> 00:10:03.039
It's how one person's blind spot gets caught by another person's experience.
00:10:03.279 --> 00:10:08.720
If more of those checks go to a machine, the team may still get answers.
00:10:08.960 --> 00:10:13.919
What it may lose, in my opinion, is the habit of checking each other.
00:10:14.399 --> 00:10:21.360
The fourth cost is the lowest and the hardest to see is the leader's map.
00:10:21.679 --> 00:10:25.519
Every leader carries a map of the team in their head.
00:10:25.840 --> 00:10:26.879
Who knows what?
00:10:27.039 --> 00:10:30.399
Who is stuck, what is moving, what is not.
00:10:30.720 --> 00:10:34.799
That map gets updated mostly by the questions that come in.
00:10:35.039 --> 00:10:41.919
When fewer questions come in, the map does not erase itself, it just stops updating.
00:10:43.039 --> 00:10:51.360
And in my experience, a leader reading an old map can feel just as confident as a leader reading a new one.
00:10:52.000 --> 00:10:54.879
That is the part I want leaders to hear.
00:10:55.120 --> 00:11:02.720
In my experience, confidence does not drop when information drops, it often stays right where it was.
00:11:03.039 --> 00:11:10.000
So a leader can be making decisions from a map from three months ago, the landscape changed.
00:11:10.639 --> 00:11:17.279
They feel grounded, they feel informed, but they are quietly behind.
00:11:17.919 --> 00:11:20.480
And under pressure, this gets worse.
00:11:22.799 --> 00:11:26.000
A leader under load has less attention to spare.
00:11:26.080 --> 00:11:34.080
When the calendar is full and the stakes are high, a quiet inbox feels or looks like a gift, not a gap.
00:11:34.320 --> 00:11:38.240
Nobody under strain goes looking for problems that did not knock.
00:11:38.480 --> 00:11:42.000
That is not a discipline problem, it's a load problem.
00:11:42.159 --> 00:11:48.240
The leader is not careless, the leader is busy, and silence does not ask for attention.
00:11:48.559 --> 00:11:50.720
Let me be fair on the other side too.
00:11:50.879 --> 00:11:52.799
Some of this quiet is pure game.
00:11:52.960 --> 00:11:54.559
People get unstuck faster.
00:11:54.720 --> 00:11:58.000
They stop waiting two hours for a reply to an easy question.
00:11:58.240 --> 00:11:59.519
They learn on their own.
00:11:59.679 --> 00:12:05.600
A team that can answer its own routine question in many ways is a stronger team.
00:12:05.840 --> 00:12:09.600
So the cost is not AI is bad for teams.
00:12:09.759 --> 00:12:11.360
That's not what I'm saying here.
00:12:11.600 --> 00:12:14.159
The cost is narrower and sharper.
00:12:14.399 --> 00:12:22.559
The leader can lose the small, odd, early signals, and those were never the routine ones.
00:12:23.039 --> 00:12:26.639
Now, why is this so easy to miss?
00:12:26.960 --> 00:12:32.159
Why would a smart, experienced leader not notice that something stopped arriving?
00:12:32.399 --> 00:12:35.919
There is one idea from behavioral science that helps here.
00:12:36.000 --> 00:12:39.279
It's called the feature positive effect.
00:12:39.840 --> 00:12:41.360
And here is what it is.
00:12:41.600 --> 00:12:49.759
Across many studies, people and animals too learn from what is present faster than from what is absent.
00:12:50.320 --> 00:12:56.320
They learn from what is present faster than what is absent.
00:12:57.039 --> 00:13:04.559
The classic work goes back to Newman, Wolfe, and Hearst in 1980, and it has been replicated since.
00:13:04.720 --> 00:13:09.039
In simple terms, when a signal shows up, we pick up on it.
00:13:09.279 --> 00:13:14.399
When a signal is missing, we are much slower to learn anything from that missing piece.
00:13:14.559 --> 00:13:16.799
It's actually one of the biases that we have.
00:13:16.960 --> 00:13:19.120
We call that the availability bias.
00:13:19.279 --> 00:13:27.440
Like, for example, when we ask a question, what is more dangerous of a job, being a cop or being a logger in the forest?
00:13:27.679 --> 00:13:31.600
Most people will answer very quickly a cop.
00:13:32.159 --> 00:13:36.080
Actually, the logger in the forest is a more dangerous job.
00:13:36.240 --> 00:13:39.600
But why they jump into being a cop of police officer?
00:13:39.759 --> 00:13:50.159
Because of the availability of the information that we always have about cops and police officers being in danger all the time on the news and everywhere we go.
00:13:50.320 --> 00:13:52.320
So it's called an availability bias.
00:13:52.480 --> 00:13:58.159
So think of it like this: if a smoke alarm goes off, everyone notices.
00:13:58.559 --> 00:14:04.159
If the sound of the dishwasher stops, it can take a long time for anyone to realize it's quiet.
00:14:04.639 --> 00:14:08.399
Presence grabs us, absence slips by.
00:14:08.879 --> 00:14:11.440
There is one more finding I find generally hopeful.
00:14:11.759 --> 00:14:18.320
A study by Rasen in 2014 found that simply alerting people to this tendency reduced it.
00:14:18.559 --> 00:14:23.360
Just alerting people to the then the tendency reduced it.
00:14:23.600 --> 00:14:26.799
That is part of why I'm telling you about it today.
00:14:26.960 --> 00:14:32.080
Now, the honest limit, and I wanted to say this out loud, these studies were lab learning tasks.
00:14:32.240 --> 00:14:36.639
They were not about leaders, they were not about teams, they were not about AI.
00:14:36.879 --> 00:14:46.000
Applying the feature positive effect to a leader whose team has started asking AI first is my inference, not their finding.
00:14:46.240 --> 00:14:51.200
And that is why, or here is why I think the inference is useful.
00:14:51.440 --> 00:14:53.679
A question that arrives is a feature.
00:14:53.840 --> 00:14:56.240
It's present, a leader can react to it.
00:14:56.480 --> 00:15:00.000
A question that does not arrive is an absence.
00:15:00.240 --> 00:15:02.240
There is nothing to react to.
00:15:02.480 --> 00:15:08.240
The research suggests that people tend to be slower to learn from that kind of absence.
00:15:08.480 --> 00:15:14.799
So a missing question may be exactly the kind of thing that is easy not to notice.
00:15:15.519 --> 00:15:20.320
That is the whole mechanism, or that's what all the mechanism needs to explain.
00:15:20.559 --> 00:15:23.360
It does not explain why a team went quiet.
00:15:23.440 --> 00:15:26.240
It does not tell us whether it's fear or convenience.
00:15:26.399 --> 00:15:32.639
It only explains why the leaders may not notice the quiet in the first place.
00:15:32.799 --> 00:15:38.399
And if the recent finding holds in real life, and again, that's a leap, okay?
00:15:38.720 --> 00:15:41.519
Then naming it is the first step.
00:15:41.919 --> 00:15:48.639
Leader who knows that absence is easy to miss can start looking for it on purpose.
00:15:48.960 --> 00:15:50.720
That's what the solution is built on.
00:15:50.879 --> 00:15:58.480
Not more monitoring, not surveillance of anyone's chat history, just deliberately looking for what is missing.
00:15:58.960 --> 00:16:00.879
So, what can a leader actually do?
00:16:01.120 --> 00:16:02.000
Four moves.
00:16:02.159 --> 00:16:11.600
Each one is small, each one involves AI directly, because AI is now part of how the team works and pretending otherwise is not a plan.
00:16:11.919 --> 00:16:13.120
Move number one.
00:16:13.840 --> 00:16:18.960
Once a week, write down three questions that used to reach you and did not this week.
00:16:19.120 --> 00:16:19.600
That's it.
00:16:19.759 --> 00:16:20.240
Three.
00:16:20.399 --> 00:16:23.759
On a paper or in a note on your laptop.
00:16:24.000 --> 00:16:27.279
Friday afternoon works for many leaders.
00:16:27.440 --> 00:16:28.720
Here is my thinking.
00:16:29.039 --> 00:16:33.759
If absence is easy to miss, the leader makes absence visible.
00:16:34.000 --> 00:16:37.759
Writing it down turns a missing thing into a present thing.
00:16:38.000 --> 00:16:40.159
Suddenly, there is a list to look at.
00:16:40.399 --> 00:16:41.919
What might be on that list?
00:16:42.159 --> 00:16:47.039
Something like nobody asked me how to handle the pricing exception this week.
00:16:47.279 --> 00:16:51.039
Or nobody checked with me before sending the client update.
00:16:51.279 --> 00:16:57.440
Or the new analyst used to ask me about the reporting format every Tuesday, not this week.
00:16:57.679 --> 00:17:00.080
Some weeks the list will be boring.
00:17:00.320 --> 00:17:00.960
Good.
00:17:01.519 --> 00:17:04.240
That may mean mostly routine questions moved.
00:17:04.400 --> 00:17:08.720
Other weeks, one item on the list will make the leader sit up.
00:17:08.960 --> 00:17:12.400
That is the week the habit paid for itself.
00:17:12.640 --> 00:17:19.279
A practical tip: if writing three feels hard, that itself is information.
00:17:19.519 --> 00:17:24.000
It may mean the leader's map is already a little old.
00:17:24.960 --> 00:17:26.160
Move number two.
00:17:26.799 --> 00:17:28.480
Ask one person plainly.
00:17:28.720 --> 00:17:32.240
What did you ask AI this week that you used to ask me?
00:17:32.559 --> 00:17:35.039
Say it with curiosity, not judgment.
00:17:35.200 --> 00:17:37.279
That is not a gotcha moment.
00:17:37.519 --> 00:17:39.119
It's not a compliance check.
00:17:39.200 --> 00:17:41.279
It's not an honest question.
00:17:41.519 --> 00:17:46.400
Here is what it can sound like word for word at the start of a one-on-one.
00:17:47.279 --> 00:17:48.640
Quick one before we start.
00:17:48.799 --> 00:17:50.559
I'm curious, not checking up.
00:17:50.720 --> 00:17:54.000
What did you ask AI this week that you used to ask me?
00:17:54.160 --> 00:18:03.200
I know you're trying to help me and relieve some of the work that I do, and I appreciate it, but I just want to make sure that we're not missing anything.
00:18:03.440 --> 00:18:04.480
Then stop talking.
00:18:04.640 --> 00:18:05.680
Let them answer.
00:18:05.920 --> 00:18:08.000
Often the answer will be easy and useful.
00:18:08.160 --> 00:18:14.559
All the formatting for the board deck or how the word, uh, how to word the reply to that supplier.
00:18:14.799 --> 00:18:15.200
Great.
00:18:15.359 --> 00:18:18.079
Now the leader knows what kinds of question moved.
00:18:18.240 --> 00:18:20.559
Sometimes the answer will be more interesting.
00:18:20.720 --> 00:18:25.119
Like, I asked it how to tell a client we were going to miss the date.
00:18:25.440 --> 00:18:27.759
That is a question with weight.
00:18:28.000 --> 00:18:31.119
That is a question the leader may want to know about.
00:18:31.279 --> 00:18:35.920
And now the leader does just because they asked.
00:18:36.240 --> 00:18:37.680
One more thing about this move.
00:18:37.839 --> 00:18:42.720
Asking it tells the person that bringing questions to the leader, it's still welcome.
00:18:43.039 --> 00:18:44.960
The question itself sends a message.
00:18:45.119 --> 00:18:46.880
I still want to hear from you.
00:18:47.200 --> 00:18:49.359
Move number three.
00:18:49.599 --> 00:18:52.319
Listen for which kind of quiet it is.
00:18:52.480 --> 00:18:55.440
Remember the two readings convenience or caution.
00:18:55.599 --> 00:18:57.680
Convenience sounds like this.
00:18:58.079 --> 00:18:59.200
It was faster.
00:18:59.440 --> 00:19:02.480
Or I did not want to bother you with something that small.
00:19:02.720 --> 00:19:04.480
Or it was 10 at night.
00:19:04.720 --> 00:19:06.799
That is the tool doing its job.
00:19:07.039 --> 00:19:10.000
AI is fast, patient, always there.
00:19:10.160 --> 00:19:11.440
That is a good thing.
00:19:11.680 --> 00:19:15.119
When it's convenience, respond with thanks and curiosity.
00:19:15.279 --> 00:19:17.839
Something like, that's great, honestly.
00:19:18.160 --> 00:19:20.079
What did you learn from it?
00:19:20.400 --> 00:19:22.000
Anything surprising?
00:19:22.319 --> 00:19:29.680
The leader stays connected to what the person is learning without pulling them back into asking permission for small things.
00:19:30.000 --> 00:19:31.839
Caution sounds different.
00:19:32.079 --> 00:19:35.440
Something like, I wasn't sure how you would take it.
00:19:35.759 --> 00:19:38.960
Or I figured it was something I should already know.
00:19:39.519 --> 00:19:43.119
Or it felt safer to just check first.
00:19:43.279 --> 00:19:48.960
That is a question that went on a machine because the machine felt less risky than the person.
00:19:49.359 --> 00:19:54.160
When it's a caution, the move is to make the next question easier to bring.
00:19:55.119 --> 00:20:01.680
Not with a speech, with a simple, specific invitation like the following.
00:20:02.480 --> 00:20:05.680
Bring me the have form ones too.
00:20:05.920 --> 00:20:12.079
I would rather hear a messy question early than a polished one late.
00:20:12.640 --> 00:20:14.240
That sentence does a lot.
00:20:14.319 --> 00:20:15.759
It lowers the bar.
00:20:16.079 --> 00:20:25.759
It says unfinished thinking is welcome and it gives the person a concrete permission, not a vague my door is always open kind of permission.
00:20:26.000 --> 00:20:29.440
Which in my experience, many people quietly do not believe.
00:20:29.599 --> 00:20:33.519
Notice that the leader is not deciding in advance which kind of quiet it is.
00:20:33.680 --> 00:20:36.960
The leader is listening for it, one person at a time.
00:20:37.200 --> 00:20:43.920
Some people on the same team will be convenience, some may be caution, both can be true in one room.
00:20:44.079 --> 00:20:45.440
Now, move number four.
00:20:45.680 --> 00:20:48.960
Keep one channel that only a person can fill.
00:20:49.279 --> 00:20:50.799
AI can answer a lot.
00:20:50.960 --> 00:20:56.240
It can explain a policy, draft an email, check a formula, summarize a meeting.
00:20:56.480 --> 00:21:05.279
What it cannot do is tell a leader what a specific person is worried about unless that person says it to the leader.
00:21:05.759 --> 00:21:09.279
So protect one question that AI cannot answer for them.
00:21:09.440 --> 00:21:12.960
A simple version at the end of every one-on-one.
00:21:13.519 --> 00:21:14.480
Ask this.
00:21:24.640 --> 00:21:28.240
What worries me is a human question.
00:21:28.559 --> 00:21:34.160
It invites the odd thing, the small thing, the thing that does not fit in a project record.
00:21:34.400 --> 00:21:36.400
Here is how that might go.
00:21:36.559 --> 00:21:40.240
The leader asks, before we wrap, what worries you right now?
00:21:40.400 --> 00:21:43.279
The person says, honestly, nothing big.
00:21:43.599 --> 00:21:50.799
Maybe that the new process is going to double the work for the support team, but I'm probably overthinking it.
00:21:51.119 --> 00:21:59.759
That last line, probably overthinking it, is exactly the kind of early signal that used to arrive in a hallway question.
00:22:00.799 --> 00:22:07.920
Now it has a home, a regular, predictable place where only one person can put it.
00:22:08.079 --> 00:22:12.079
So let me put the four moves together so they are easy to carry.
00:22:12.240 --> 00:22:16.079
One, weakly write down three questions that did not arrive.
00:22:16.319 --> 00:22:21.759
Two, ask one person what they asked AI that they used to ask the leader.
00:22:21.920 --> 00:22:26.960
Three, listen for convenience or caution and answer each one differently.
00:22:27.119 --> 00:22:32.400
Four, keep one human-only channel like what worries you at the end of every one-on-one.
00:22:33.759 --> 00:22:36.480
None of this is actually anti-AI.
00:22:36.720 --> 00:22:41.519
In fact, it takes AI for what it is, a real part of how work happens now.
00:22:41.680 --> 00:22:46.960
The team is going to keep asking AI first for a lot of things, and in many cases they should.
00:22:47.200 --> 00:22:49.519
The goal is not to pull those questions back.
00:22:49.680 --> 00:22:54.079
The goal is to make sure the leader is still hearing the one that matters.
00:22:54.319 --> 00:22:58.319
In my opinion, that is what a grounded leader does with AI in the room.
00:22:58.480 --> 00:23:06.000
Not compete with it, not ignore it, just stay curious about what is now carrying the cues to come to them.
00:23:06.400 --> 00:23:10.880
Now you might be thinking, good, fewer interruptions, isn't that exactly what I wanted?
00:23:11.119 --> 00:23:13.839
That's a fair objection and it's a strong one.
00:23:14.079 --> 00:23:19.680
Last week's piece was about how answering everyone's question can swallow a leader's day.
00:23:19.839 --> 00:23:23.119
So when the questions slow down, shouldn't a leader celebrate?
00:23:23.359 --> 00:23:24.319
Partly yes.
00:23:24.559 --> 00:23:27.039
For the routine questions, absolutely.
00:23:27.279 --> 00:23:37.599
If AI is handling what the temp what is the template for this and how do I format that, the leader gets time back and the team gets answers faster.
00:23:37.759 --> 00:23:40.799
That is a win, and I'm not going to pretend otherwise.
00:23:41.039 --> 00:23:42.400
But here is the problem.
00:23:42.640 --> 00:23:45.839
The questions did not leave in a neat order.
00:23:46.160 --> 00:23:50.960
AI did not take only the boring ones and leave the important ones behind.
00:23:51.279 --> 00:23:57.200
People choose where to send each question, one at a time, often in a hurry.
00:23:57.440 --> 00:24:05.200
Some of the small odd ones went too, and those small odd ones are the ones that used to carry early warnings.
00:24:05.599 --> 00:24:09.279
Is it normal that this customer emailed three times this week?
00:24:09.440 --> 00:24:11.039
That sounds routine.
00:24:11.440 --> 00:24:16.720
It might be routine, or it might be the first sign of an account in trouble.
00:24:16.960 --> 00:24:20.400
And this is where the feature positive effect comes back in.
00:24:20.559 --> 00:24:24.640
If the routine questions disappear, a leader notices the relief.
00:24:24.799 --> 00:24:27.279
Relief is present, it's easy to feel.
00:24:27.519 --> 00:24:33.519
But if one of the odd early warning questions also disappears, the leader feels nothing.
00:24:34.000 --> 00:24:36.640
There is no gap to feel.
00:24:36.880 --> 00:24:39.279
Absence is hard to notice.
00:24:39.519 --> 00:24:44.240
The research suggests people tend to learn more slowly from what is missing.
00:24:44.799 --> 00:24:50.319
And while applying that to leaders, in my inference, it is a reasonable warning.
00:24:50.799 --> 00:24:59.759
So the honest answer to is not this what I wanted is a leader wants fewer interruptions, yes, not fewer warnings.
00:25:00.079 --> 00:25:03.759
And from the inside, those two look exactly the same.
00:25:03.920 --> 00:25:06.160
That is why the weekly list matters.
00:25:06.240 --> 00:25:10.559
That is how a leader keeps the relief and still checks the gap.
00:25:10.880 --> 00:25:15.200
Today was about what stops reaching the leader, but there is a mirror question.
00:25:15.440 --> 00:25:23.599
On Friday, we asked why a team can't hear the leader in a perfect college AI written message.
00:25:24.720 --> 00:25:28.720
If this conversation is lending for you, I want to invite you to one thing.
00:25:29.039 --> 00:25:31.359
Register for the pressure room.
00:25:31.519 --> 00:25:36.559
It's a free live session for executives, 45 minutes on Zoom.
00:25:36.640 --> 00:25:41.759
The next one is Thursday, October 29 at 11 Pacific a.m.
00:25:42.160 --> 00:25:49.359
The session title is Why Does a Small Problem Suddenly Feels Like an Emergency Under Pressure?
00:25:49.599 --> 00:25:55.680
You can register at saharandradi.com slash pressure dash rule.
00:25:55.920 --> 00:25:58.799
So this week look for what did not arrive.
00:25:59.039 --> 00:26:03.119
Write down three questions, ask one person what they asked.
00:26:03.440 --> 00:26:06.799
AI, listen for which kind of quite it is.
00:26:07.200 --> 00:26:10.559
The quite is information too.
00:26:10.799 --> 00:26:12.480
It just does not knock.
00:26:12.720 --> 00:26:14.480
Leadership does not fail.
00:26:14.799 --> 00:26:17.839
Systems, nervous systems do.
00:26:18.240 --> 00:26:22.319
As usual, before I wrap up, I want to ask you to show me some love.
00:26:22.720 --> 00:26:26.000
Rate, comment, save, share the podcast.
00:26:26.160 --> 00:26:31.359
Thank you for your support for making us one of the top 2% global podcasts that exist.
00:26:31.440 --> 00:26:32.880
I really appreciate you.
00:26:33.039 --> 00:26:36.880
Share it so we can reach more people that need to hear that.
00:26:37.119 --> 00:26:38.559
This is Sahara Andradi.
00:26:38.799 --> 00:26:43.440
Till we meet again at AI Cafe Conversation Podcast.
00:26:43.680 --> 00:26:44.640
Peace out.
