September 29, 2025 • 21 mins
OpenAI Launches Pulse: Personalized Daily Reports and AI-Driven Insights for Pro Users
Microsoft adds Claude models to Copilot 365
OpenAI tests ChatGPT Agent upgrades powered by new models
OpenAI and Nvidia Forge $100 Billion Partnership for AI Infrastructure Development
OpenAI and Jony Ive to Develop Smart Speaker-Like Device by 2027
OpenAI readies Orders in ChatGPT for native checkout
Why Your Customer Support Still Has Humans
#AI, #CustomerSupport, #OpenAI, #Microsoft, #Nvidia, #ChatGPT, #Technology
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
Welcome to Innovation Pulse, your quick no-nonsense update on the latest in AI.
(00:09):
First, we will cover the latest news.
Open AI launches Pulse for personalized updates.
Microsoft enhances Co-Pilot with Anthropics AI.
New Alpha models hint at GPT-5 and a 100 billion NVIDIA partnership for AI infrastructure.
After this, we will dive deep into the challenges and opportunities of replacing human customer
(00:32):
service agents with AI.
Open AI has introduced Pulse, a new feature within chat GPT that creates personalized
reports for users while they sleep.
Pulse aims to encourage morning use, similar to checking social media or news apps.
It delivers 5-10 briefs, providing updates on topics like news or user-specific interests.
(00:58):
This feature is part of Open AI's effort to make chat GPT more proactive and assistive.
Initially available for $200 a month pro subscribers, Pulse allows users to connect
apps like Google Calendar and Gmail to pull relevant information.
It can also utilize chat GPT's memory features to tailor reports based on past interactions.
(01:22):
Although Pulse competes with news platforms, it cites sources similarly to chat GPT search.
Open AI plans to expand Pulse's capabilities, eventually enabling it to perform tasks like
making reservations, though such features are still in development.
Microsoft is expanding its AI capabilities by integrating Anthropics AI models into Microsoft
(01:49):
365 Co-Pilot.
Beta testers now have access to Anthropics Claude Opus 4.1 model alongside Open AI's
models like GPT-5, which continue to power Co-Pilot.
This integration is part of Microsoft's Frontier Program, an optional beta for AI previews.
(02:10):
Testers can now use Claude in Co-Pilot's Researcher Tool and Co-Pilot Studio, an AI
agent development tool. Users can toggle between Open AI and Anthropic models through a drop-down
menu. Although tensions between Microsoft and Open AI were speculated, both AI models are
currently utilized. Microsoft's Charles Lamanna indicated that Anthropic will enhance Co-Pilot's
(02:37):
capabilities further, promising more powerful experiences for users. This development is
expected to roll out broadly after the beta testing phase.
Up next, we're exploring transformative AI workflow advancements. Open AI is testing new
alpha models for Chat GPT, appearing briefly to users as Agent with Truncation and Agent with
(03:03):
Prompt Expansion. These models activate Agent mode, allowing Chat GPT to use a browser and
tools automatically. The terms suggest experiments with different system prompts or model architectures,
possibly paving the way for GPT-5-powered agents. These tests appeal to early adopters and researchers
(03:26):
interested in AI workflows. The models were quickly removed, pointing to an accidental release,
with no clear improvements or limitations known yet. Open AI CEO Sam Altman has hinted at major
announcements, and these tests may signal upcoming launches. If these models are released, they could
(03:46):
transform browsing and research tasks by managing context and prompts differently, aligning with
Open AI's goal of enhancing Chat GPT's capabilities against competitors like Co-Pilot.
Sam Altman, CEO of Open AI, and Jensen Huang, CEO of NVIDIA, have formed a $100 billion partnership
(04:08):
to develop AI infrastructure. This deal was finalized just before Altman's flight to Abilene,
Texas, and follows their involvement in a state visit to the UK with President Trump.
The partnership is significant in the tech industry, linking NVIDIA's powerful GPUs with
Open AI's generative AI expertise. NVIDIA will invest in Open AI and supply processors for new
(04:34):
data centers. The companies plan to build AI supercomputing facilities with the first site
expected to go online next year. While working closely with NVIDIA, Open AI maintains relationships
with other partners, including Microsoft and Oracle. This venture is part of Open AI's broader
Stargate project, aiming to expand AI infrastructure globally. Open AI may eventually offer its own
(05:01):
cloud services, further advancing its AI capabilities. For now, let's focus on the
upcoming smart device. Open AI is reportedly developing a new AI device with Apple's former
chief design officer, Johnny Ive. This device is said to resemble a smart speaker without a display.
(05:23):
Open AI has secured a contract with Luxure and approached GURTECH, both Apple assemblers,
for components like speaker modules. In addition to the speaker, Open AI is exploring other products
like smart glasses, a digital voice recorder, and a wearable pin, aiming for a release in
2026 or 2027. CEO Sam Altman envisions a family of pocket-sized, screen-free, contextually aware
(05:53):
devices. Open AI is leveraging Apple's supply chain in China, with Luxure and GURTECH involved.
Many Apple employees have moved to Open AI, attracted by promises of less bureaucracy and
more collaboration. Open AI's first device is not expected to be smart glasses, though this could
follow later. Open AI is enhancing chat GPT with new features, signalling a shift in user interaction.
(06:22):
The latest update integrates voice mode directly into the main app window,
though screen sharing for voice mode is not progressing. A new orders setting is being
prepared, potentially enabling native checkout and order tracking, with users adding credit
cards for seamless purchases within the app. This feature is likely to debut in the US,
(06:45):
assisting frequent users in buying plugins and services more efficiently. Additionally,
Open AI is working on introducing parental controls to the chat GPT web interface,
addressing concerns about safe usage for younger audiences. These updates align with industry
trends to transform AI chat platforms into comprehensive tools for productivity,
(07:08):
shopping, and communication. Open AI is laying the groundwork for these enhancements,
aiming to make chat GPT a daily utility. And now, pivot our discussion towards the
main AI topic.
All right everybody, welcome back to Innovation Pulse. I'm Alex, and today we're diving into
(07:34):
something that honestly baffled me when I first started looking into it. We've got all this
incredible AI technology, right? Chat GPT can write essays, AI can diagnose diseases,
robots are doing backflips. So why the heck am I still waiting on hold to talk to a human
being at my cable company? Ha, great question, Alex. I'm Yakov Lasker, and I've been researching
(07:57):
this exact paradox. And let me tell you, the answer is way more fascinating than you'd think.
It's not that companies aren't trying to go full AI. Many have, and a lot of them spectacularly
failed. Oh, I need to hear this. Give me the T on these AI disasters. Okay, so remember Air Canada,
their AI chatbot told a customer they could get a bereavement discount on a flight.
(08:19):
Turns out, totally made up. The customer sued, and here's the kicker. The court ruled that Air
Canada was responsible for what their chatbot said, even though it was wrong. The airline
actually tried to argue they shouldn't be held accountable for their own AI. Wait, they tried
to say our chatbot isn't us? That's like saying sorry if my left hand doesn't speak for my right
(08:41):
hand. Exactly. And then there's DPD, the UK delivery company, their chatbot started swearing
at customers and writing poems about how terrible the company was after a system update went wrong.
No way. The AI roasted its own company? Oh yeah, and McDonald's their AI drive-through added
butter and ketchup to ice cream orders, and once kept adding chicken nuggets until the order hit
(09:03):
over 2000 nuggets. The customer was just laughing, trying to get it to stop. Okay, but those are like
extreme fails, right? What about the fundamental question? Can AI actually replace human customer
service agents? That's where it gets really interesting. So here's the data. Between 52 and
(09:25):
71% of customers prefer interacting with human agents over AI chatbots. And get this, 81% of
customers would rather wait to speak with a human than get instant help from AI. Wow, even with
instant service people still want humans? Yes, and 90% of users prefer human assistance over
chatbots. It's not even close. But here's the thing, it's not just stubborn customers refusing
(09:48):
to adapt. There are real legitimate reasons AI can't handle it all. Lay it on me. What's AI missing?
Three big things. First, the empathy gap. AI chatbots fundamentally lack emotional intelligence.
They can't truly empathize with a frustrated customer who's been on hold for an hour or
someone dealing with a sensitive issue. Second, complex problem solving. Human agents can think
(10:10):
outside the box, handle ambiguous situations, offer creative solutions. AI is still very much
pattern matching. And the third thing? Linguistic adaptability. There's actually research showing
that human agents demonstrate higher linguistic similarity to customers than chatbots do.
Meaning humans are better at matching how customers actually talk with all the variations,
(10:32):
slang and dynamic language changes. And here's a stat that should scare companies.
60% of customers say chatbots often fail to understand their issue.
So it's not working. But companies keep trying, right? There must be some success stories.
Oh, there are. That's what makes this so complex. Companies see massive cost savings.
(10:55):
The average cost per customer interaction has dropped from $4.60 to $1.45 after AI
implementation. That is a 68% reduction. Some companies are reporting $3.50 earned for every
dollar invested in AI customer service, with leading organizations achieving up to 8x ROI.
(11:16):
Okay, those are huge numbers. So why not go all in?
Because look what happened to Clarna. This is my favorite story. Their CEO,
Sebastian Semyatkowski, went all in on AI. He publicly stated that AI can already do all of
the jobs that we as humans do. Their chatbot was supposedly doing the work of 700 customer service
(11:37):
agents, resolving cases nine minutes faster. Sounds like a success. That's what they thought.
But then this year, Clarna completely reversed course. They're now recruiting human customer
service workers again. The CEO literally said, from a brand perspective, a company perspective,
I just think it's so critical that you are clear to your customer that there will always be a human
(11:59):
if you want. Their new motto is basically, AI gives us speed. Talent gives us empathy.
That's a complete 180. What changed their mind?
Reality. Customers weren't happy. The brand was suffering. And they're not alone.
Even Salesforce's Mark Benioff, who cut 4,000 customer service jobs and replaced them with AI,
(12:21):
now admits that 50% of interactions are with AI and 50% with humans. He used to say,
AI wouldn't replace jobs, just augment them. Now he's more, let's say, realistic about the
displacement happening. This sounds like companies are stuck between a rock and a hard place.
The economics push them toward AI, but customers push them toward humans.
(12:45):
Precisely. And there's more complexity. Implementation is really hard. The main obstacles
include cost, lengthy implementation times, complex systems, and ineffective training.
Despite all the investment, only 25% of call centers have successfully integrated
AI automation into their daily operations. Only 25%? That's shockingly low.
(13:05):
Right. And then there's the regulatory nightmare. AI regulations are evolving rapidly,
but are inconsistent across regions. The EU has the AI Act with risk-based rules.
In the US, it's a patchwork of state and local laws. California just finalized
regulations on automated decision-making technology. Companies are navigating this
(13:26):
complex landscape while trying to innovate. So where does this leave us? What's the actual future here?
The hybrid model. Every expert I looked into, and I mean every single one,
points to the same solution. AI handles routine tasks. Humans handle complex high-value interactions.
It's not about replacement. It's about strategic collaboration.
(13:47):
What does that look like in practice? Think of it as the 80-20 rule.
AI handles about 80% of routine inquiries. Password resets, order tracking, basic FAQs.
Humans focus on the 20% that requires creativity, empathy, and complex problem solving.
ServiceNow's CEO, Bill McDermott, actually argues that AI won't reduce jobs at all,
(14:09):
because there's a massive tech talent deficit globally. Millions of unfilled positions.
Wait, so AI could create jobs? His argument is that AI fills the talent gap,
allowing companies to do things they couldn't before, because they lacked the workforce.
Humans get re-skilled for more interesting work. Workday's CEO, Carl Eschenbach, said it well.
(14:31):
There will be jobs that AI can do much better than humans, no doubt.
But that doesn't mean that a human is no longer needed. Humans are going to be re-skilled.
Okay, but I want to get practical here. What should companies actually do?
And what should we as customers expect? For companies, the advice from experts is clear.
AI should augment human agents, not replace them. Automate the routine to drive efficiency,
(14:56):
but always ensure customers have a clear, easy path to a human, especially when emotions or
complexity come into play. One expert, Shashank Singh from Backstage.d. AI, said AI should act as
an assistant, not a gatekeeper. That's a great distinction. AI is assistant, not gatekeeper.
Exactly. And for customers, here's what you should expect. More and more of your initial
(15:18):
interactions will be with AI. By some estimates, 95% of customer interactions will involve AI
in some form by now. But, and this is crucial, you should always have an easy option to escalate
to a human. If a company is hiding that option or making it difficult, that's a red flag.
What about the quality of service? Are we getting better or worse service with this hybrid approach?
(15:41):
Actually, better in many cases. Companies using this model report some impressive stats.
First response times have been cut by up to 74% within the first year. Resolution times are down
87% in some cases. And customer satisfaction is often higher, because AI handles the boring stuff
quickly, and humans are available for when things get complicated or emotional. So it's not AI versus
(16:06):
humans. It's AI and humans together. That's the big takeaway. Tom Eggmeier, Zendesk's CEO,
says we're advancing toward a world where 100% of customer interactions involve AI in some form.
But here's the key stat. 75% of CX leaders see AI as a force for amplifying human intelligence,
not replacing it. I love that framing. Amplifying, not replacing. And there's a human element we
(16:31):
can't ignore. Industries with strong human touch requirements, healthcare, complex financial
services, anything with sensitive personal issues, will always need that human judgment.
You can't have an AI tell someone their insurance claim is denied, or walk them through a traumatic
situation. The empathy just isn't there. Right, and probably won't be for a long time, if ever.
(16:53):
So for listeners wondering about their own jobs in customer service,
what should they be thinking about? Upskill. The jobs aren't disappearing entirely,
but they're changing. The World Economic Forum predicts that automation will displace 85 million
jobs by 2030, but create 97 million new ones. In customer service specifically, 37% of those new
(17:15):
jobs will be in the customer care industry. The key is developing skills that complement AI.
Critical thinking, complex problem solving, emotional intelligence, creativity. Those are
exactly the things AI struggles with. Exactly. Human agents need to become experts at the complex
stuff, the edge cases, the situations where genuine human judgment and empathy matter. Think of it
(17:40):
less as customer service agent, and more as customer success specialist or complex issue resolver.
Before we wrap up, I want to touch on something you mentioned earlier, the failures. Should companies
be more transparent about when AI messes up? Absolutely. One of the big lessons from all
this research is that transparency builds trust. Customers need to know they're talking to AI.
(18:04):
They need to know how to reach a human. And when AI makes mistakes, which it will,
companies need to own it, not try to disclaim responsibility like Air Canada did.
It's interesting because the technology is incredible, but the implementation is where
everything falls apart. That's exactly it. The technology is amazing. AI can process information
(18:25):
at superhuman speeds, work 24-7, handle multiple languages, learn from interactions,
but without proper guardrails, proper training, and proper integration with human agents.
It becomes a liability instead of an asset. So bottom line for our listeners, if you're running
a business, don't go all in on AI without careful planning and always maintain that human safety
(18:50):
net. If you're a customer, demand access to humans when you need it. And if you work in customer
service, focus on becoming really, really good at the things AI can't do.
Perfect summary, and I'd add one more thing. Watch this space. The technology is evolving
rapidly. The hybrid models we're seeing today might look completely different in two or three years.
(19:11):
But one thing seems certain. Humans aren't going anywhere. We're just changing roles.
From doing everything to doing the things that matter most.
Exactly. And honestly, that's probably better for everyone. Customers get faster service on
routine stuff and better service on complex stuff. Companies save money while maintaining
quality and human agents get to do more interesting fulfilling work.
(19:34):
Well, folks, next time you're on hold waiting to talk to a human, maybe you'll feel a little
differently about it. There's actually a good reason that person is there, and they're probably
handling something your AI assistant couldn't touch. And if you're lucky, you won't get 2,000
chicken nuggets added to your order in the process. Still can't believe that actually happened.
(19:57):
All right, everyone, that's our deep dive into why humans are still manning the customer support
lines. Thanks for joining us on innovation pulse. If you learned something today, share this episode
with someone who's curious about the AI revolution and what it really means for jobs and service.
And remember, the future isn't AI or humans, it's AI and humans working together until next time.
(20:21):
Thanks for tuning in, folks. Stay curious, stay informed, and we'll catch you on the next episode
of innovation pulse. That's a wrap for today's podcast, where we explored open AI's latest
innovations and discussed the balance between AI and human roles in customer service. Don't forget
(20:44):
to like, subscribe and share this episode with your friends and colleagues so they can also
stay updated on the latest news and gain powerful insights. Stay tuned for more updates.
Welcome to Innovation Pulse, your quick no-nonsense update on the latest in AI.
(00:09):
First, we will cover the latest news.
Open AI launches Pulse for personalized updates.
Microsoft enhances Co-Pilot with Anthropics AI.
New Alpha models hint at GPT-5 and a 100 billion NVIDIA partnership for AI infrastructure.
After this, we will dive deep into the challenges and opportunities of replacing human customer
(00:32):
service agents with AI.
Open AI has introduced Pulse, a new feature within chat GPT that creates personalized
reports for users while they sleep.
Pulse aims to encourage morning use, similar to checking social media or news apps.
It delivers 5-10 briefs, providing updates on topics like news or user-specific interests.
(00:58):
This feature is part of Open AI's effort to make chat GPT more proactive and assistive.
Initially available for $200 a month pro subscribers, Pulse allows users to connect
apps like Google Calendar and Gmail to pull relevant information.
It can also utilize chat GPT's memory features to tailor reports based on past interactions.
(01:22):
Although Pulse competes with news platforms, it cites sources similarly to chat GPT search.
Open AI plans to expand Pulse's capabilities, eventually enabling it to perform tasks like
making reservations, though such features are still in development.
Microsoft is expanding its AI capabilities by integrating Anthropics AI models into Microsoft
(01:49):
365 Co-Pilot.
Beta testers now have access to Anthropics Claude Opus 4.1 model alongside Open AI's
models like GPT-5, which continue to power Co-Pilot.
This integration is part of Microsoft's Frontier Program, an optional beta for AI previews.
(02:10):
Testers can now use Claude in Co-Pilot's Researcher Tool and Co-Pilot Studio, an AI
agent development tool. Users can toggle between Open AI and Anthropic models through a drop-down
menu. Although tensions between Microsoft and Open AI were speculated, both AI models are
currently utilized. Microsoft's Charles Lamanna indicated that Anthropic will enhance Co-Pilot's
(02:37):
capabilities further, promising more powerful experiences for users. This development is
expected to roll out broadly after the beta testing phase.
Up next, we're exploring transformative AI workflow advancements. Open AI is testing new
alpha models for Chat GPT, appearing briefly to users as Agent with Truncation and Agent with
(03:03):
Prompt Expansion. These models activate Agent mode, allowing Chat GPT to use a browser and
tools automatically. The terms suggest experiments with different system prompts or model architectures,
possibly paving the way for GPT-5-powered agents. These tests appeal to early adopters and researchers
(03:26):
interested in AI workflows. The models were quickly removed, pointing to an accidental release,
with no clear improvements or limitations known yet. Open AI CEO Sam Altman has hinted at major
announcements, and these tests may signal upcoming launches. If these models are released, they could
(03:46):
transform browsing and research tasks by managing context and prompts differently, aligning with
Open AI's goal of enhancing Chat GPT's capabilities against competitors like Co-Pilot.
Sam Altman, CEO of Open AI, and Jensen Huang, CEO of NVIDIA, have formed a $100 billion partnership
(04:08):
to develop AI infrastructure. This deal was finalized just before Altman's flight to Abilene,
Texas, and follows their involvement in a state visit to the UK with President Trump.
The partnership is significant in the tech industry, linking NVIDIA's powerful GPUs with
Open AI's generative AI expertise. NVIDIA will invest in Open AI and supply processors for new
(04:34):
data centers. The companies plan to build AI supercomputing facilities with the first site
expected to go online next year. While working closely with NVIDIA, Open AI maintains relationships
with other partners, including Microsoft and Oracle. This venture is part of Open AI's broader
Stargate project, aiming to expand AI infrastructure globally. Open AI may eventually offer its own
(05:01):
cloud services, further advancing its AI capabilities. For now, let's focus on the
upcoming smart device. Open AI is reportedly developing a new AI device with Apple's former
chief design officer, Johnny Ive. This device is said to resemble a smart speaker without a display.
(05:23):
Open AI has secured a contract with Luxure and approached GURTECH, both Apple assemblers,
for components like speaker modules. In addition to the speaker, Open AI is exploring other products
like smart glasses, a digital voice recorder, and a wearable pin, aiming for a release in
2026 or 2027. CEO Sam Altman envisions a family of pocket-sized, screen-free, contextually aware
(05:53):
devices. Open AI is leveraging Apple's supply chain in China, with Luxure and GURTECH involved.
Many Apple employees have moved to Open AI, attracted by promises of less bureaucracy and
more collaboration. Open AI's first device is not expected to be smart glasses, though this could
follow later. Open AI is enhancing chat GPT with new features, signalling a shift in user interaction.
(06:22):
The latest update integrates voice mode directly into the main app window,
though screen sharing for voice mode is not progressing. A new orders setting is being
prepared, potentially enabling native checkout and order tracking, with users adding credit
cards for seamless purchases within the app. This feature is likely to debut in the US,
(06:45):
assisting frequent users in buying plugins and services more efficiently. Additionally,
Open AI is working on introducing parental controls to the chat GPT web interface,
addressing concerns about safe usage for younger audiences. These updates align with industry
trends to transform AI chat platforms into comprehensive tools for productivity,
(07:08):
shopping, and communication. Open AI is laying the groundwork for these enhancements,
aiming to make chat GPT a daily utility. And now, pivot our discussion towards the
main AI topic.
All right everybody, welcome back to Innovation Pulse. I'm Alex, and today we're diving into
(07:34):
something that honestly baffled me when I first started looking into it. We've got all this
incredible AI technology, right? Chat GPT can write essays, AI can diagnose diseases,
robots are doing backflips. So why the heck am I still waiting on hold to talk to a human
being at my cable company? Ha, great question, Alex. I'm Yakov Lasker, and I've been researching
(07:57):
this exact paradox. And let me tell you, the answer is way more fascinating than you'd think.
It's not that companies aren't trying to go full AI. Many have, and a lot of them spectacularly
failed. Oh, I need to hear this. Give me the T on these AI disasters. Okay, so remember Air Canada,
their AI chatbot told a customer they could get a bereavement discount on a flight.
(08:19):
Turns out, totally made up. The customer sued, and here's the kicker. The court ruled that Air
Canada was responsible for what their chatbot said, even though it was wrong. The airline
actually tried to argue they shouldn't be held accountable for their own AI. Wait, they tried
to say our chatbot isn't us? That's like saying sorry if my left hand doesn't speak for my right
(08:41):
hand. Exactly. And then there's DPD, the UK delivery company, their chatbot started swearing
at customers and writing poems about how terrible the company was after a system update went wrong.
No way. The AI roasted its own company? Oh yeah, and McDonald's their AI drive-through added
butter and ketchup to ice cream orders, and once kept adding chicken nuggets until the order hit
(09:03):
over 2000 nuggets. The customer was just laughing, trying to get it to stop. Okay, but those are like
extreme fails, right? What about the fundamental question? Can AI actually replace human customer
service agents? That's where it gets really interesting. So here's the data. Between 52 and
(09:25):
71% of customers prefer interacting with human agents over AI chatbots. And get this, 81% of
customers would rather wait to speak with a human than get instant help from AI. Wow, even with
instant service people still want humans? Yes, and 90% of users prefer human assistance over
chatbots. It's not even close. But here's the thing, it's not just stubborn customers refusing
(09:48):
to adapt. There are real legitimate reasons AI can't handle it all. Lay it on me. What's AI missing?
Three big things. First, the empathy gap. AI chatbots fundamentally lack emotional intelligence.
They can't truly empathize with a frustrated customer who's been on hold for an hour or
someone dealing with a sensitive issue. Second, complex problem solving. Human agents can think
(10:10):
outside the box, handle ambiguous situations, offer creative solutions. AI is still very much
pattern matching. And the third thing? Linguistic adaptability. There's actually research showing
that human agents demonstrate higher linguistic similarity to customers than chatbots do.
Meaning humans are better at matching how customers actually talk with all the variations,
(10:32):
slang and dynamic language changes. And here's a stat that should scare companies.
60% of customers say chatbots often fail to understand their issue.
So it's not working. But companies keep trying, right? There must be some success stories.
Oh, there are. That's what makes this so complex. Companies see massive cost savings.
(10:55):
The average cost per customer interaction has dropped from $4.60 to $1.45 after AI
implementation. That is a 68% reduction. Some companies are reporting $3.50 earned for every
dollar invested in AI customer service, with leading organizations achieving up to 8x ROI.
(11:16):
Okay, those are huge numbers. So why not go all in?
Because look what happened to Clarna. This is my favorite story. Their CEO,
Sebastian Semyatkowski, went all in on AI. He publicly stated that AI can already do all of
the jobs that we as humans do. Their chatbot was supposedly doing the work of 700 customer service
(11:37):
agents, resolving cases nine minutes faster. Sounds like a success. That's what they thought.
But then this year, Clarna completely reversed course. They're now recruiting human customer
service workers again. The CEO literally said, from a brand perspective, a company perspective,
I just think it's so critical that you are clear to your customer that there will always be a human
(11:59):
if you want. Their new motto is basically, AI gives us speed. Talent gives us empathy.
That's a complete 180. What changed their mind?
Reality. Customers weren't happy. The brand was suffering. And they're not alone.
Even Salesforce's Mark Benioff, who cut 4,000 customer service jobs and replaced them with AI,
(12:21):
now admits that 50% of interactions are with AI and 50% with humans. He used to say,
AI wouldn't replace jobs, just augment them. Now he's more, let's say, realistic about the
displacement happening. This sounds like companies are stuck between a rock and a hard place.
The economics push them toward AI, but customers push them toward humans.
(12:45):
Precisely. And there's more complexity. Implementation is really hard. The main obstacles
include cost, lengthy implementation times, complex systems, and ineffective training.
Despite all the investment, only 25% of call centers have successfully integrated
AI automation into their daily operations. Only 25%? That's shockingly low.
(13:05):
Right. And then there's the regulatory nightmare. AI regulations are evolving rapidly,
but are inconsistent across regions. The EU has the AI Act with risk-based rules.
In the US, it's a patchwork of state and local laws. California just finalized
regulations on automated decision-making technology. Companies are navigating this
(13:26):
complex landscape while trying to innovate. So where does this leave us? What's the actual future here?
The hybrid model. Every expert I looked into, and I mean every single one,
points to the same solution. AI handles routine tasks. Humans handle complex high-value interactions.
It's not about replacement. It's about strategic collaboration.
(13:47):
What does that look like in practice? Think of it as the 80-20 rule.
AI handles about 80% of routine inquiries. Password resets, order tracking, basic FAQs.
Humans focus on the 20% that requires creativity, empathy, and complex problem solving.
ServiceNow's CEO, Bill McDermott, actually argues that AI won't reduce jobs at all,
(14:09):
because there's a massive tech talent deficit globally. Millions of unfilled positions.
Wait, so AI could create jobs? His argument is that AI fills the talent gap,
allowing companies to do things they couldn't before, because they lacked the workforce.
Humans get re-skilled for more interesting work. Workday's CEO, Carl Eschenbach, said it well.
(14:31):
There will be jobs that AI can do much better than humans, no doubt.
But that doesn't mean that a human is no longer needed. Humans are going to be re-skilled.
Okay, but I want to get practical here. What should companies actually do?
And what should we as customers expect? For companies, the advice from experts is clear.
AI should augment human agents, not replace them. Automate the routine to drive efficiency,
(14:56):
but always ensure customers have a clear, easy path to a human, especially when emotions or
complexity come into play. One expert, Shashank Singh from Backstage.d. AI, said AI should act as
an assistant, not a gatekeeper. That's a great distinction. AI is assistant, not gatekeeper.
Exactly. And for customers, here's what you should expect. More and more of your initial
(15:18):
interactions will be with AI. By some estimates, 95% of customer interactions will involve AI
in some form by now. But, and this is crucial, you should always have an easy option to escalate
to a human. If a company is hiding that option or making it difficult, that's a red flag.
What about the quality of service? Are we getting better or worse service with this hybrid approach?
(15:41):
Actually, better in many cases. Companies using this model report some impressive stats.
First response times have been cut by up to 74% within the first year. Resolution times are down
87% in some cases. And customer satisfaction is often higher, because AI handles the boring stuff
quickly, and humans are available for when things get complicated or emotional. So it's not AI versus
(16:06):
humans. It's AI and humans together. That's the big takeaway. Tom Eggmeier, Zendesk's CEO,
says we're advancing toward a world where 100% of customer interactions involve AI in some form.
But here's the key stat. 75% of CX leaders see AI as a force for amplifying human intelligence,
not replacing it. I love that framing. Amplifying, not replacing. And there's a human element we
(16:31):
can't ignore. Industries with strong human touch requirements, healthcare, complex financial
services, anything with sensitive personal issues, will always need that human judgment.
You can't have an AI tell someone their insurance claim is denied, or walk them through a traumatic
situation. The empathy just isn't there. Right, and probably won't be for a long time, if ever.
(16:53):
So for listeners wondering about their own jobs in customer service,
what should they be thinking about? Upskill. The jobs aren't disappearing entirely,
but they're changing. The World Economic Forum predicts that automation will displace 85 million
jobs by 2030, but create 97 million new ones. In customer service specifically, 37% of those new
(17:15):
jobs will be in the customer care industry. The key is developing skills that complement AI.
Critical thinking, complex problem solving, emotional intelligence, creativity. Those are
exactly the things AI struggles with. Exactly. Human agents need to become experts at the complex
stuff, the edge cases, the situations where genuine human judgment and empathy matter. Think of it
(17:40):
less as customer service agent, and more as customer success specialist or complex issue resolver.
Before we wrap up, I want to touch on something you mentioned earlier, the failures. Should companies
be more transparent about when AI messes up? Absolutely. One of the big lessons from all
this research is that transparency builds trust. Customers need to know they're talking to AI.
(18:04):
They need to know how to reach a human. And when AI makes mistakes, which it will,
companies need to own it, not try to disclaim responsibility like Air Canada did.
It's interesting because the technology is incredible, but the implementation is where
everything falls apart. That's exactly it. The technology is amazing. AI can process information
(18:25):
at superhuman speeds, work 24-7, handle multiple languages, learn from interactions,
but without proper guardrails, proper training, and proper integration with human agents.
It becomes a liability instead of an asset. So bottom line for our listeners, if you're running
a business, don't go all in on AI without careful planning and always maintain that human safety
(18:50):
net. If you're a customer, demand access to humans when you need it. And if you work in customer
service, focus on becoming really, really good at the things AI can't do.
Perfect summary, and I'd add one more thing. Watch this space. The technology is evolving
rapidly. The hybrid models we're seeing today might look completely different in two or three years.
(19:11):
But one thing seems certain. Humans aren't going anywhere. We're just changing roles.
From doing everything to doing the things that matter most.
Exactly. And honestly, that's probably better for everyone. Customers get faster service on
routine stuff and better service on complex stuff. Companies save money while maintaining
quality and human agents get to do more interesting fulfilling work.
(19:34):
Well, folks, next time you're on hold waiting to talk to a human, maybe you'll feel a little
differently about it. There's actually a good reason that person is there, and they're probably
handling something your AI assistant couldn't touch. And if you're lucky, you won't get 2,000
chicken nuggets added to your order in the process. Still can't believe that actually happened.
(19:57):
All right, everyone, that's our deep dive into why humans are still manning the customer support
lines. Thanks for joining us on innovation pulse. If you learned something today, share this episode
with someone who's curious about the AI revolution and what it really means for jobs and service.
And remember, the future isn't AI or humans, it's AI and humans working together until next time.
(20:21):
Thanks for tuning in, folks. Stay curious, stay informed, and we'll catch you on the next episode
of innovation pulse. That's a wrap for today's podcast, where we explored open AI's latest
innovations and discussed the balance between AI and human roles in customer service. Don't forget
(20:44):
to like, subscribe and share this episode with your friends and colleagues so they can also
stay updated on the latest news and gain powerful insights. Stay tuned for more updates.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
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
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.