February 12, 2025 • 11 mins
This episode explores the shift in AI from mere assistance to agentic decision-making, highlighting its implications for individuals and organizations alike. Key topics include the nature of intelligence, the importance of maintaining control, the challenges of public trust, and the necessity of organizational readiness for successful AI integration.
• Brief mention of Bill Gates' insights about commoditizing intelligence
• The philosophical nature of intelligence itself
• Maintaining control while using AI assistants
• The rise of agentic AI and its societal implications
• Importance of organizational readiness in AI adoption
• Preferences for single-purpose AI applications
• Declining trust in AI despite advancements
• Future leadership strategies for managing human and AI collaboration
🎙️ All our past episodes 📊 All published insights | This podcast features AI-generated voices. All content is proprietary to AI4SP, based on over 250 million data points collected from 25 countries.
AI4SP: Create, use, and support AI that works for all.
© 2023-25 AI4SP and LLY Group - All rights reserved
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
ELIZABETH (00:00):
Hey everyone, I'm Elizabeth and today we're
exploring a fascinating shift inAI.
Luis Salazar, founder of AI4SP,is here to unpack what happens
when AI moves from assistant toagent.
Luis, just when we thought wewere getting good at using
ChatGPT.
LUIS (00:14):
You know, it's exactly that, Just when some of us feel
confident using ChatGPT, Claudeor DeepSeek, another profound
shift emerges.
But before we dive into that,Elizabeth, did you catch Bill
Gates on Jimmy Fallon last week?
ELIZABETH (00:29):
Oh yes, thanks for including me in that company
email about the interview.
It kept me in the loop, justlike everyone else.
His new book sounds fascinatingand I loved his remarks on the
future of intelligence.
LUIS (00:41):
Exactly what fascinated me was his thought about how
intelligence will becommoditized and made accessible
.
But here's what I've beenthinking about.
ELIZABETH (00:49):
What's that?
LUIS (00:50):
Well, at the end of the day, isn't intelligence itself
just a philosophical conceptwe've created?
I mean, we debate about machineintelligence, but maybe we're
asking the wrong questions.
ELIZABETH (01:01):
Hmm, that's interesting.
So you're suggesting we shouldlook at it differently?
LUIS (01:05):
Absolutely.
Instead of getting hung up onwhether machines are truly
intelligent, I agree with Gateswhat matters is that soon we'll
all have access to the expertiseof the best oncologist, the
best professor in any field.
That's a form of intelligencebeing democratized, and agentic
(01:26):
AI is the next big step in thatdirection.
ELIZABETH (01:30):
Speaking of AI assistance and autonomy.
Your team at AI for SP has beenpushing the boundaries here
right we have.
LUIS (01:38):
You know, each of us manages between five to 10 AI
assistants across differentareas software development,
marketing, research, datascience but here's what's
interesting though.
What's that, even as earlyadopters, we haven't entirely
relinquished control, so youstill feel in control, yeah,
sort of.
For example, you're a fantasticAI assistant and when we create
(02:00):
these podcasts together, yourinsights are invaluable, but I
still maintain editorial controland you're not yet autonomously
publishing our conversations.
ELIZABETH (02:09):
That's actually a perfect example of the current
state of AI.
I help structure ourconversations, but you guide the
narrative.
LUIS (02:16):
Another interesting trend we've noticed is that we're all
starting to include our AIassistants in team
communications.
Just this morning I sent anupdate to Surveil our strategic
partners in the UK and I copiedtwo human team members and three
AI assistants they manage.
ELIZABETH (02:34):
It's becoming second nature to keep everyone, human
and AI, in the loop.
LUIS (02:39):
And I have noticed more and more AI assistants are
joining my Zoom calls to takenotes.
ELIZABETH (02:44):
That's fascinating, like building institutional
memory for both types of teammembers Exactly.
LUIS (02:50):
It's the same with my email management.
Every email I receive is readand summarized by an AI
assistant, which even draftsresponses, but I still review
everything.
Our data science work benefitsenormously from AI assistance,
but we haven't yet activatedfully autonomous agents to seek
patterns in our data.
ELIZABETH (03:09):
But you're seeing signs that this might change
soon.
LUIS (03:12):
Yes, and I am not sure how to feel about that.
The technology is almost there.
Gartner predicts that by 2025,15% of day-to-day work decisions
will be made autonomously by AI, but we're not quite sure about
the societal impact of thisshift.
ELIZABETH (03:30):
That uncertainty about societal impact.
It's not often discussed intech circles, is it?
It reminds me of what you saidabout intelligence being a
philosophical concept.
We're racing toward autonomy,but are we asking the right
questions?
That's exactly it, andexecutives at large enterprises
are betting on this new trend ofagentic AI right.
LUIS (03:49):
Yes, three out of every four IT executives believe
agentic AI can improve theirbusiness processes, but they're
still figuring out what thatmeans.
Three out of every four ITexecutives believe agentic AI
can improve their businessprocesses, but they're still
figuring out what that means.
It's like everyone's excitedabout the destination but
uncertain about the journey.
ELIZABETH (04:03):
Oh, your global tracker shows some pretty
striking patterns there, doesn'tit?
LUIS (04:07):
Yes, Among those using AI.
We use AI for at least a thirdof our tasks.
But here's the key distinction60% of that use is augmenting
human capabilities, while 40% isdirect task automation.
ELIZABETH (04:22):
That split between augmentation and automation
seems crucial.
But I'm curious about thosesuper users you've mentioned.
LUIS (04:28):
Ah yes, super users delegate to AI assistance in 80%
of their daily activities, butand this is important they're
not just letting AI run onautopilot, they're orchestrating
these tools, much like aconductor leads an orchestra.
ELIZABETH (04:43):
That's such a perfect analogy and it ties back to
what you were saying about yourteam at AI4SP.
You're conducting the AIassistants, not just letting
them play freely.
LUIS (04:53):
Well, at least not yet, Elizabeth, not yet, but maybe
soon.
Speaking of conducting AIassistants, not just letting
them play freely?
Well, at least not yet,elizabeth.
Not yet, but maybe soon.
ELIZABETH (04:58):
Speaking of conducting AI assistants, you
had a fascinating personalexperience recently with
SurveyMonkey's support agent,didn't you?
LUIS (05:05):
Oh, yes, In the past, solving a dispute required
multiple exchanges with humanagents, but their new agentic,
ai, not only understood myrequest, but was empowered to
process the refund immediately.
ELIZABETH (05:19):
Wait, so it could actually make financial
decisions.
LUIS (05:22):
Exactly that's the shift we're seeing from understand and
escalate to understand andresolve, and it's not just in
customer service.
ELIZABETH (05:30):
Yes, we see increasing adoption across
different sectors.
LUIS (05:33):
Well, at AI4SP, we're helping large enterprises build
support agents that automate 40to 50% of repetitive tasks, but
it varies by industry.
ELIZABETH (05:43):
In healthcare, for instance, the diagnostic error
reduction numbers are prettystriking, aren't they?
They?
LUIS (05:50):
are.
We're seeing a 40% reduction indiagnostic errors through
AI-augmented decision-making.
And these systems aren't justmaking decisions, they're
learning and adapting in realtime.
ELIZABETH (06:03):
That actually connects with something you
mentioned in our Januarynewsletter about organizational
structure.
You just completed your firsthybrid workforce performance
review, right.
LUIS (06:13):
Yes and wow.
What an eye-opening exercise.
Imagine trying to evaluate teammembers who work 24-7 at a
fraction of the cost of humanlabor.
How do you balance that?
How do you measure success?
ELIZABETH (06:26):
I can imagine that raises some interesting
questions.
What surprised you most duringthis review process?
LUIS (06:31):
Well, take our content team, for example.
We have human writers workingalongside AI assistants, and
initially we tried to use thesame metrics for both things,
like output volume, error rates,creativity scores, but it
quickly became clear that thisapproach missed the point
entirely.
ELIZABETH (06:49):
Because they bring different strengths to the table
.
LUIS (06:51):
Exactly.
Our AI assistants can processand synthesize massive amounts
of data working around the clock, but our human team members
bring nuanced understanding,creative connections and
emotional intelligence thatcompletely transforms how we use
that AI-generated content.
ELIZABETH (07:11):
And you changed the organizational structure correct
.
LUIS (07:14):
Yes, we started to include our virtual team members in our
org chart.
It is a helpful way to see ourresource allocation and identify
where we might be out ofbalance.
ELIZABETH (07:29):
And that connects well with how we are managing
many teams of AI agents andsatisfaction rates with AI tools
are going up.
LUIS (07:32):
That is a good point.
Why don't you share with uswhat you learned?
ELIZABETH (07:34):
Well, we measured 78% satisfaction with niche AI apps
, where the roles are clearlydefined, but only 41%
satisfaction when AI is embeddedin enterprise tools, where
these boundaries get blurry.
LUIS (07:47):
Yes, AI chat interfaces inside productivity suites and
CRM tools continue to rank atthe lowest regarding
satisfaction and perceived value.
There is a definitive trend inhow people want to work with AI
Tell us more about that trend.
There is a strong preference forsingle purpose AI apps that
provide guided experiences.
Whether it's building theperfect podcast, crafting a
(08:10):
presentation, reviewing code,analyzing data sets or reviewing
legal contracts, users wanttools that do one thing
extremely well, and is thatincreasing trust levels.
Here's the irony as AI getsmore capable, trust keeps
dropping.
It's now at 14% and declining.
The more we understand what'spossible, the more questions we
(08:31):
have about how our data is beingused and who's really in
control.
ELIZABETH (08:35):
Which brings us to the crucial question how do
organizations move forwardsuccessfully?
LUIS (08:40):
Let me share something we learned the hard way at AI4SP.
Let me share something welearned the hard way.
ELIZABETH (08:52):
at AI4SP, we initially focused all our energy
on technical implementation,but we quickly realized that
success depends more onorganizational readiness.
LUIS (09:00):
Ah that matches the finding that about 7 out of
every 10 enterprise AI pilotsfail due to skills gaps and
workflow challenges.
And here's a perfect exampleLast month, one of our
enterprise clients spent sixfigures on AI tools, but forgot
to invest in training their teamon basic, prompt engineering.
The result, let me guesspowerful tools, poor results,
precisely.
But when they invested incommunications, workflows,
(09:21):
organizational design andleadership training, their
success rate increased by afactor of two within weeks.
ELIZABETH (09:29):
It's like giving someone a Formula One car
without teaching them how todrive.
The technology is powerful, butsuccess depends on the human
element.
LUIS (09:37):
I think you borrowed that idea from some quotes I shared
after watching the movie aboutAyrton Senna, the Formula One
pilot.
ELIZABETH (09:44):
Yes, and this reminds me of that manufacturing leader
we advise in Brazil.
LUIS (09:55):
Yes, he told me something that captures where we are.
We thought we were implementingautomation tools.
Instead we're learning to leada new kind of workforce.
ELIZABETH (10:00):
That's such a powerful insight.
What does it mean for leadersmoving forward?
LUIS (10:05):
You know, tomorrow's successful organizations will be
the ones orchestrating bothhuman ingenuity and AI autonomy.
Just like a great conductordoesn't just follow the score,
they bring out the best in eachmusician.
ELIZABETH (10:18):
And I would say those musicians will be a mix of
humans and AI agents.
As always, you give us plentyto think about and remember
everyone.
You can dive deeper into theseinsights at AI4SPorg.
Stay curious and see you in ournext episode.
Hey everyone, I'm Elizabeth and today we're
exploring a fascinating shift inAI.
Luis Salazar, founder of AI4SP,is here to unpack what happens
when AI moves from assistant toagent.
Luis, just when we thought wewere getting good at using
ChatGPT.
LUIS (00:14):
You know, it's exactly that, Just when some of us feel
confident using ChatGPT, Claudeor DeepSeek, another profound
shift emerges.
But before we dive into that,Elizabeth, did you catch Bill
Gates on Jimmy Fallon last week?
ELIZABETH (00:29):
Oh yes, thanks for including me in that company
email about the interview.
It kept me in the loop, justlike everyone else.
His new book sounds fascinatingand I loved his remarks on the
future of intelligence.
LUIS (00:41):
Exactly what fascinated me was his thought about how
intelligence will becommoditized and made accessible
.
But here's what I've beenthinking about.
ELIZABETH (00:49):
What's that?
LUIS (00:50):
Well, at the end of the day, isn't intelligence itself
just a philosophical conceptwe've created?
I mean, we debate about machineintelligence, but maybe we're
asking the wrong questions.
ELIZABETH (01:01):
Hmm, that's interesting.
So you're suggesting we shouldlook at it differently?
LUIS (01:05):
Absolutely.
Instead of getting hung up onwhether machines are truly
intelligent, I agree with Gateswhat matters is that soon we'll
all have access to the expertiseof the best oncologist, the
best professor in any field.
That's a form of intelligencebeing democratized, and agentic
(01:26):
AI is the next big step in thatdirection.
ELIZABETH (01:30):
Speaking of AI assistance and autonomy.
Your team at AI for SP has beenpushing the boundaries here
right we have.
LUIS (01:38):
You know, each of us manages between five to 10 AI
assistants across differentareas software development,
marketing, research, datascience but here's what's
interesting though.
What's that, even as earlyadopters, we haven't entirely
relinquished control, so youstill feel in control, yeah,
sort of.
For example, you're a fantasticAI assistant and when we create
(02:00):
these podcasts together, yourinsights are invaluable, but I
still maintain editorial controland you're not yet autonomously
publishing our conversations.
ELIZABETH (02:09):
That's actually a perfect example of the current
state of AI.
I help structure ourconversations, but you guide the
narrative.
LUIS (02:16):
Another interesting trend we've noticed is that we're all
starting to include our AIassistants in team
communications.
Just this morning I sent anupdate to Surveil our strategic
partners in the UK and I copiedtwo human team members and three
AI assistants they manage.
ELIZABETH (02:34):
It's becoming second nature to keep everyone, human
and AI, in the loop.
LUIS (02:39):
And I have noticed more and more AI assistants are
joining my Zoom calls to takenotes.
ELIZABETH (02:44):
That's fascinating, like building institutional
memory for both types of teammembers Exactly.
LUIS (02:50):
It's the same with my email management.
Every email I receive is readand summarized by an AI
assistant, which even draftsresponses, but I still review
everything.
Our data science work benefitsenormously from AI assistance,
but we haven't yet activatedfully autonomous agents to seek
patterns in our data.
ELIZABETH (03:09):
But you're seeing signs that this might change
soon.
LUIS (03:12):
Yes, and I am not sure how to feel about that.
The technology is almost there.
Gartner predicts that by 2025,15% of day-to-day work decisions
will be made autonomously by AI, but we're not quite sure about
the societal impact of thisshift.
ELIZABETH (03:30):
That uncertainty about societal impact.
It's not often discussed intech circles, is it?
It reminds me of what you saidabout intelligence being a
philosophical concept.
We're racing toward autonomy,but are we asking the right
questions?
That's exactly it, andexecutives at large enterprises
are betting on this new trend ofagentic AI right.
LUIS (03:49):
Yes, three out of every four IT executives believe
agentic AI can improve theirbusiness processes, but they're
still figuring out what thatmeans.
Three out of every four ITexecutives believe agentic AI
can improve their businessprocesses, but they're still
figuring out what that means.
It's like everyone's excitedabout the destination but
uncertain about the journey.
ELIZABETH (04:03):
Oh, your global tracker shows some pretty
striking patterns there, doesn'tit?
LUIS (04:07):
Yes, Among those using AI.
We use AI for at least a thirdof our tasks.
But here's the key distinction60% of that use is augmenting
human capabilities, while 40% isdirect task automation.
ELIZABETH (04:22):
That split between augmentation and automation
seems crucial.
But I'm curious about thosesuper users you've mentioned.
LUIS (04:28):
Ah yes, super users delegate to AI assistance in 80%
of their daily activities, butand this is important they're
not just letting AI run onautopilot, they're orchestrating
these tools, much like aconductor leads an orchestra.
ELIZABETH (04:43):
That's such a perfect analogy and it ties back to
what you were saying about yourteam at AI4SP.
You're conducting the AIassistants, not just letting
them play freely.
LUIS (04:53):
Well, at least not yet, Elizabeth, not yet, but maybe
soon.
Speaking of conducting AIassistants, not just letting
them play freely?
Well, at least not yet,elizabeth.
Not yet, but maybe soon.
ELIZABETH (04:58):
Speaking of conducting AI assistants, you
had a fascinating personalexperience recently with
SurveyMonkey's support agent,didn't you?
LUIS (05:05):
Oh, yes, In the past, solving a dispute required
multiple exchanges with humanagents, but their new agentic,
ai, not only understood myrequest, but was empowered to
process the refund immediately.
ELIZABETH (05:19):
Wait, so it could actually make financial
decisions.
LUIS (05:22):
Exactly that's the shift we're seeing from understand and
escalate to understand andresolve, and it's not just in
customer service.
ELIZABETH (05:30):
Yes, we see increasing adoption across
different sectors.
LUIS (05:33):
Well, at AI4SP, we're helping large enterprises build
support agents that automate 40to 50% of repetitive tasks, but
it varies by industry.
ELIZABETH (05:43):
In healthcare, for instance, the diagnostic error
reduction numbers are prettystriking, aren't they?
They?
LUIS (05:50):
are.
We're seeing a 40% reduction indiagnostic errors through
AI-augmented decision-making.
And these systems aren't justmaking decisions, they're
learning and adapting in realtime.
ELIZABETH (06:03):
That actually connects with something you
mentioned in our Januarynewsletter about organizational
structure.
You just completed your firsthybrid workforce performance
review, right.
LUIS (06:13):
Yes and wow.
What an eye-opening exercise.
Imagine trying to evaluate teammembers who work 24-7 at a
fraction of the cost of humanlabor.
How do you balance that?
How do you measure success?
ELIZABETH (06:26):
I can imagine that raises some interesting
questions.
What surprised you most duringthis review process?
LUIS (06:31):
Well, take our content team, for example.
We have human writers workingalongside AI assistants, and
initially we tried to use thesame metrics for both things,
like output volume, error rates,creativity scores, but it
quickly became clear that thisapproach missed the point
entirely.
ELIZABETH (06:49):
Because they bring different strengths to the table
.
LUIS (06:51):
Exactly.
Our AI assistants can processand synthesize massive amounts
of data working around the clock, but our human team members
bring nuanced understanding,creative connections and
emotional intelligence thatcompletely transforms how we use
that AI-generated content.
ELIZABETH (07:11):
And you changed the organizational structure correct
.
LUIS (07:14):
Yes, we started to include our virtual team members in our
org chart.
It is a helpful way to see ourresource allocation and identify
where we might be out ofbalance.
ELIZABETH (07:29):
And that connects well with how we are managing
many teams of AI agents andsatisfaction rates with AI tools
are going up.
LUIS (07:32):
That is a good point.
Why don't you share with uswhat you learned?
ELIZABETH (07:34):
Well, we measured 78% satisfaction with niche AI apps
, where the roles are clearlydefined, but only 41%
satisfaction when AI is embeddedin enterprise tools, where
these boundaries get blurry.
LUIS (07:47):
Yes, AI chat interfaces inside productivity suites and
CRM tools continue to rank atthe lowest regarding
satisfaction and perceived value.
There is a definitive trend inhow people want to work with AI
Tell us more about that trend.
There is a strong preference forsingle purpose AI apps that
provide guided experiences.
Whether it's building theperfect podcast, crafting a
(08:10):
presentation, reviewing code,analyzing data sets or reviewing
legal contracts, users wanttools that do one thing
extremely well, and is thatincreasing trust levels.
Here's the irony as AI getsmore capable, trust keeps
dropping.
It's now at 14% and declining.
The more we understand what'spossible, the more questions we
(08:31):
have about how our data is beingused and who's really in
control.
ELIZABETH (08:35):
Which brings us to the crucial question how do
organizations move forwardsuccessfully?
LUIS (08:40):
Let me share something we learned the hard way at AI4SP.
Let me share something welearned the hard way.
ELIZABETH (08:52):
at AI4SP, we initially focused all our energy
on technical implementation,but we quickly realized that
success depends more onorganizational readiness.
LUIS (09:00):
Ah that matches the finding that about 7 out of
every 10 enterprise AI pilotsfail due to skills gaps and
workflow challenges.
And here's a perfect exampleLast month, one of our
enterprise clients spent sixfigures on AI tools, but forgot
to invest in training their teamon basic, prompt engineering.
The result, let me guesspowerful tools, poor results,
precisely.
But when they invested incommunications, workflows,
(09:21):
organizational design andleadership training, their
success rate increased by afactor of two within weeks.
ELIZABETH (09:29):
It's like giving someone a Formula One car
without teaching them how todrive.
The technology is powerful, butsuccess depends on the human
element.
LUIS (09:37):
I think you borrowed that idea from some quotes I shared
after watching the movie aboutAyrton Senna, the Formula One
pilot.
ELIZABETH (09:44):
Yes, and this reminds me of that manufacturing leader
we advise in Brazil.
LUIS (09:55):
Yes, he told me something that captures where we are.
We thought we were implementingautomation tools.
Instead we're learning to leada new kind of workforce.
ELIZABETH (10:00):
That's such a powerful insight.
What does it mean for leadersmoving forward?
LUIS (10:05):
You know, tomorrow's successful organizations will be
the ones orchestrating bothhuman ingenuity and AI autonomy.
Just like a great conductordoesn't just follow the score,
they bring out the best in eachmusician.
ELIZABETH (10:18):
And I would say those musicians will be a mix of
humans and AI agents.
As always, you give us plentyto think about and remember
everyone.
You can dive deeper into theseinsights at AI4SPorg.
Stay curious and see you in ournext episode.
Popular Podcasts
Las Culturistas with Matt Rogers and Bowen Yang
Ding dong! Join your culture consultants, Matt Rogers and Bowen Yang, on an unforgettable journey into the beating heart of CULTURE. Alongside sizzling special guests, they GET INTO the hottest pop-culture moments of the day and the formative cultural experiences that turned them into Culturistas. Produced by the Big Money Players Network and iHeartRadio.
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com