Is Pressure a Performance Problem or a Physiology Problem? | AI for Executives

Send us Fan Mail

Why leaders misread a physiological state as a skills gap when an AI model disagrees with their judgment

Pressure under AI-speed decisions is not a performance problem. It's a physiology problem: the body reaches a threat state before the mind finishes evaluating the data, AI-generated or otherwise. 

Sahar Andrade, MB.BCh, Neuroleadership Coach and Forbes Coaches Council member, breaks down why leaders misread a physiological state as a skills gap, and what actually steadies judgment when an algorithm hands you a call you don't have time to fully vet.

A 2026 Deloitte survey found 60 percent of leaders now use AI in some part of their decision-making, but only 5 percent say they manage that shift well. That gap isn't a training problem. It's physiological: the amygdala threat response can't tell the difference between an actual predator and an AI recommendation that contradicts a leader's own read, delivered under a clock, with an audience watching.

In this episode, Sahar walks through what is actually happening in that four-minute window, what it costs a leadership team when it repeats every week, and the one shift, taught through her B.R.A.I.N.™ framework, that changes the outcome before the meeting even starts.

Register for The Pressure Room, the free monthly live session where leaders work through exactly this kind of moment:

 https://www.saharandrade.com/pressure-room 

 #AIForExecutives #ExecutiveLeadership #NeuroscienceLeadership #AIStrategy #Neuroleadership #AITransformation #LeadershipUnderPressure  #HumancenteredAI #thepressureroom #steadybeforestrategy #HRleaderburnout #neuroleadershipcoach #nervoussystemregulationforleaders   #CHROleadershipdevelopment #executiveburnout  #humancenteredleadership  #executivecoaching #AICAFEConversationspodcast   #AIStrategy  #AITransformation #RegulatedLeadership   #Artificialintelligence #executivepresence  #neuroscienceinleadership #Futureofwork  #AIpodcast #AICaféConversations    #AInotechrequired 
 #neuroscience   #notechrequired #AItoolsforexecutives #AIandleadership #executivecoachingwithAI #AIcoachingtoolsforexecutives  #NeuroscienceOfLeadership #AIAdoption #ChangeManagement #LeadershipDevelopment   #AIAdaptation #Neuroleadership #AILeadership #LeadershipUnderPressure #ExecutiveDecisionMaking #AIIntegration #NeuroscienceInAI #neuroscienceofpressure #leadershipunderpressure #Thepressureroom

●      Is pressure a performance problem or a physiology problem?

●      Why do leaders freeze when an AI recommendation contradicts their own judgment?

●      What is the amygdala threat response in leadership?

●      Why do only 5 percent of leaders manage AI-assisted decisions well?

●      What helps a leader stay steady when a model disagrees with their read?

●      How does AI adoption affect executive decision-making under pressure?

Support the show

--- 

AI Cafe Conversations: Neuroscience-based AI leadership for executives. Hosted by Sahar (The AI Whisperer) | New episodes Wed & Fri 

🔗 Connect: https://www.linkedin.com/in/saharandradespeaker/

📧 Work with me: [e-mail]

🌐 Website: https://www.saharconsulting.com/

 📧 Instagram: https://www.instagram.com/saharthereinventcoach

Episode Transcript

SPEAKER_00 (00:11):
You just got a forecast from an AI model that
says do the opposite of whatyour gut says.
You have four minutes before themeeting.
Is that a performance problem?
Or is physiology making the callbefore judgment gets a vote?
Before we go into this, here ismy standing line.

(00:34):
Everything I share comes from myexperience and my opinion as a
coach.
It's not a diagnosis and it'snot a fact about you.
You decide what fits.
Welcome to the AI CafeConversation Podcast, the only
podcast where neuroscience, AI,and leadership intersect.

(00:54):
My name is Samar Andradi and Iam a neuroleadership coach.
Three weeks ago, a CFO and coachopened her laptop to a dashboard
that had already run thenumbers.
The AI forecast tool flagged asupply contract as high risk and
recommending pulling out.

(01:16):
Her gut said the opposite.
She had built that vendorrelationship for six years.
She had four minutes before shehad to answer in the room.
That is not a hypotheticalanymore.
In 2026, the Lloyd survey foundthat 60% of leaders now use AI
in some part of their decisionmaking, and only 5% say they

(01:40):
manage that shift well.
The tools moved faster than thejudgment did.
Here is what nobody tells youabout that moment.
The four minutes she had werenot really a thinking problem.
They were a physiology problem.
The instant a machine has aneedle recommendation that

(02:01):
contradicts the read on therule, the brain treats it the
same way it treats any otherunresolved threat.
A decision with unclear stakesmade under a clock with an
audience watching for theanswer.
I hear versions of this storyevery month now.
A VP got an AI-generated churnprediction that said to cut a

(02:25):
product line his team hadchampioned for a year.
A hospital administrator got anAI staffing model that
contradicted what her chargenurses were telling her on the
floor.
A school district superintendentgot an AI enrollment forecast
that recommended closing aprogram three community members

(02:48):
had just publicly praised.
Different industries, same fourminutes, same physiology.
Here is the mechanism thoughtoutward, not aimed at anyone
listener.
The amygdala threat response isthe brain's fast danger
detection system.
It does not wait for a fullanalysis, it flags a threat

(03:12):
first and asks questions later.
Take hostages first.
Just kidding.
Because for most of humanhistory, waiting for the full
analysis got you eaten.
It cannot tell the differencebetween an actual predator and
an unresolved decision with aclock running and an audience

(03:32):
watching.
This is not only a boardroomphenomenon.
I see the same four minutesphysiology in a plant manager
reading an AI predictivemaintenance alert that
contradicts 20 years of knowingthe sound of his own machines.
And in marketing director,watching an AI attribution model

(03:54):
reassign credit away from thecampaign her team believes
actually move the number.
The industries change the clockand the threat response do not.
When an AI system hands a leadera recommendation that conflicts
with their own read, that'sexactly the kind of unresolved
threat the amygdala, our threatcenter, was built to catch.

(04:19):
Not because the AI is dangerous,because the moment carries every
ingredient the threat system istuned for.
Uncertainty, time pressure, andsocial stains.
The brain does not evaluate thequality of the forecast.
It reacts to the shape of themoment.

(04:40):
And the shape of an AIdisagreement, machine says one
thing, years of burn judgmentsays another, is one of the
sharpest shapes a modern leaderwill meet in a given week.
Read the Deloitte gap carefullyagain.
60% adoption, 5% will manage itwell.
That is not a training gap.

(05:02):
Most of these leaders have beenbriefed on the tools.
Some have sat through hours ofAI literacy sessions.
It's a physiology gap.
Nobody trained the amygdala toslow down just because the
threat now arrives on adashboard instead of in the
grass.

(05:22):
And no solid deck about promptengineering touches that.
When a leader is caught in athreat response, the prefrontal
cortex, the part of the brainresponsible for weighing nuance
and holding multiple scenariosat once, gets less blood flow

(05:43):
relative to the faster, morereactive surface.
That is not a flaw.
It's the design, it's theanatomy, it's the physiology.
Under a real physical threat,you do not want nuance, you want
speed.
But a board-drown decision abouta vendor contract is not a
physical threat.

(06:04):
And the brain does not know thatin the moment.
So what actually happens underthat four-minute clock is a
narrowing.

Options collapse to two (06:14):
trust the AI completely or override it
completely.
The middle path, the one where aleader asks better questions of
the model, checks it againstsomething the AI cannot see, and
makes a generally integratedcall, requires exactly the kind

(06:34):
of steady, white-lens thinkingthat a threat response shuts
down first.
Remember, AI suggests humandisciples.
Do you know how many discussionsI had with my chief of staff on
Clogged?
Back and forth, back and forth.
You know, sometimes AI winsbecause it makes sense.

(06:58):
Sometimes I win because it makesbetter sense.
And that's what AI was there.
AI is just here to support us,not to believe it 100%, but at
least to learn from it and useit as a support mechanism.
So I have watched leaders madeboth mistakes in the same week.

(07:26):
One CEO overwrote every AIrecommendation on principle,
calling it instant, and missedtwo generally good calls the
model got right.
Another followed AIrecommendation without question,
called the trigger, and missedthe one time her team's
floor-level knowledge shouldhave overridden the dashboard.

(07:48):
That's exactly what I was sayinga minute ago.
Neither was wrong about wantingto lead well.
Both were caught in the samephysiological narrowing, just
running in opposite directions.
Multiply this across the core,across a year.
A leadership team that treatsevery AI assisted decision as a

(08:10):
threat to survive rather thaninformation to integrate does
not get faster.
It gets brittle.
Decisions get made in thefour-minute window instead of
the four-hour window that wasactually available.
Because the buddy alreadydecided the clock was shorter
than it was.

(08:31):
The EY Ernest Young 2026research on this is blunt.
52% of departments are alreadyusing AI without real oversight.
You cannot do that.
And 78% of organizations sayadoption is outpacing the risk

(08:52):
management.
That is not a technology problemwaiting on better tools.
That is a physiology problemwaiting on steadier leaders.
And the cost is not only thedecision itself.
A leader will resolve every oneof these four-minute moments
from threat instead of judgmenttrains their own team to do the

(09:12):
same.
The team watches how the leaderhandles the model's disagreement
more closely than they watch themodel itself.
Steadiness at the top iscontagious.
So is the narrowing as well.
A room that watches its leaderflinch at a dashboard learns to

(09:33):
flinch at dashboards too.
And that habit outliveswhichever tool triggered it
first.
This is a longer cost as well,one that shows up slower and
matters more.
Leaders who spend a yearresolving AI disagreements from
threat instead of judgment startto trust their own read less,

(09:57):
not more.
Some overcorrect into defaultingto the model every time because
defaulting feels safer thansitting inside the discomfort of
disagreement.
Others overcorrect intorejecting the tools altogether
and lose real value the AI wasactually offering.

(10:19):
I would rather leaders sit inthat discomfort a little longer
than resolve it either directionjust to make the feeling stop.
Both are physiology decisionswearing a strategy costume.
So if this is a physiologyproblem, the fix is not a
smarter dashboard.

(10:40):
It's a steadier leader lookingat the dashboard.
Here is what changes the fourminutes.
Before the meeting, before themodel's recommendation even
loads, a leader who is groundedhas access to the full range of
their thinking.
A leader already in a threatresponse only has access to the

(11:01):
narrow range.
The work is not reacting fasterto the AI, it's arriving
steadier before the AI everspeaks.
Brain is the framework I builtfor exactly this kind of moment.
The space between a fast-movinginput, whether that input is a

(11:22):
person, a market, a model, andthe leader's real response.
It gives leaders a way to catchthe narrowing before it makes
the decision for them.
This matters more, not less.
As AI moves into more of thedecision layer, the tools are
not going to slow down.

(11:44):
A 2026 CHRO association surveyfound AI adoption is now the top
concern for 91% of HR.
91%.
Not because the technology isbad, but because the humans
making the calls around it havenot been given anything to

(12:05):
steady themselves with.

(12:36):
Some of the best calls I havewatched leaders make this year
came from taking an AIrecommendation seriously and
then asking one more question itcould not answer.
That question only shows up whenthe leader is steady enough to
ask it instead of reacting tothis agreement itself.

(12:58):
What this looks like in practiceis smaller than people expect.
It's not a retreat, not ameditation app, not a two-week
course.
It's a leader who has trainedone specific pause into the
seconds before a hard call.
This way, the nervous systemarrives to the decision already

(13:20):
steady instead of arriving thereand then trying to come down in
front of the room.
It will not happen.
That gap, steady first versussteady after, is the entire
difference between a leader whointegrates an AI disagreement
well and one who either carvesto it or fights it.

(13:44):
Steadiness comes beforestrategy.
Always steady before strategy.
That has always been true, it'sjust more visible now because
the clock got shorter and theinputs got faster.
I can hear some of you already.
I don't have four minutes to getsteady.
The model already flagged it.

(14:06):
I have to decide now.

Here is the thing (14:08):
you don't need four minutes to get steady.
You need about 90 seconds.
And most leaders spend that timerereading the dashboard instead.
The nervous system does not needa retreat.
It needs one breath, low enough,to reach the part of the brain
that can actually weigh thecall.

(14:30):
Taken before the door opens, notduring the meeting.
The leaders who get this wrongare not the ones who are too
busy.
They are the ones who mistakerereading the data for getting
steady.
Those are two different actions.
One keeps a leader in the threatresponse, cycling the same

(14:51):
numbers without adding realjudgment.
The other gets a leader out ofit.
And the meeting that followsplays out what completely
different, even when the finalanswer ends up the same.
I can hear a second objectiontoo.
What if I get steady and themodel was right and I lose time
I didn't have?

(15:11):
Getting steady doesn't meansecond guessing every
recommendation into paralysis.
It means the 90 seconds you takeeither confirm the model fast
because a clear head seesclearly, or surface the one real
objection your bleed wasstriking.
Either way, the decision thatfollows is a decision, not a

(15:36):
flinch.
If this is the moment you keephitting the four minutes where
the model says one thing and youread says another, this is
exactly what we work throughlife in the pressure.
One session, one model, oneprotocol, free.

(15:57):
Register for the pressure room.
The next one is on Thursday,August 27th at 11 a.m.
Pacific.
I will leave the link in thenotes.
If you have ever sat in thatfour-minute gap between a
machine's NSOP and your own,this is built for that exact
seat.
Leadership doesn't fail, NervousSystems 2.

(16:21):
Study before strategy.
This is Saharn Andradi.
I am a neuroleadership coach.
I specialize in the neuroscienceof pressure and leadership under
pressure.
This was our Friday Forbesarticle podcast, a short one.
Before I leave, show me somelove.
Register, rate, share, commenton our podcast.

(16:46):
Thank you for your support andmaking us one of the top 2%
global podcasts.
Thank you so much for being herefor listening to the podcast.
Till I see you on our regularpodcast on Wednesday.
Peace out.