Unlock potential of Generative AI by Conquering the Spooky Mountain | Nikhil Bhojwani | TEDxBoston

Summary

This script delves into the societal apprehensions surrounding AI, drawing parallels to historical resistance to new technologies like writing. It acknowledges AI's potential to improve governance and address biases and privacy issues but highlights the 'spooky' psychological barrier AI faces due to its human-like capabilities. The speaker emphasizes the importance of redesigning processes and organizations to integrate AI effectively, ensuring transparency, trust, and human-centricity to overcome fear and embrace AI's benefits in sectors like healthcare.

Takeaways

• 🤖 AI as a Double-Edged Sword: The script discusses the potential of AI to both improve and challenge human reliance on technology, similar to Socrates' concerns about writing in the 5th Century BC.
• 🧐 The Illusion of Knowledge: AI can create an illusion of knowledge without genuine understanding, blurring the lines between fact and fiction.
• 🔍 Societal Backlash: Every new technology, including AI, has faced societal backlash, which is a natural part of the adaptation process.
• 🛠 Addressing AI Criticism: AI criticism is often valid and leads to improved governance and performance, addressing issues like errors, bias, and privacy concerns.
• 🧠 Redefining Intelligence: AI challenges traditional notions of intelligence, as seen with achievements in games and creative tasks that were once considered uniquely human.
• 👥 The Uncanny Valley of AI: As AI becomes more human-like, it can evoke psychological fear and discomfort, which may hinder its acceptance and benefits.
• 🏥 Healthcare and AI: AI can assist in healthcare by taking on tasks like reading X-rays, but generative AI that engages in nuanced conversations raises ethical and practical questions.
• 🛑 The Spooky Mountain: The fear of AI encroaching on human territory, especially in areas like healthcare, is a significant barrier to its implementation.
• 🛑 Risk-Aversion in Healthcare: The potential for AI to cause harm in healthcare is a major concern due to the industry's high risk-aversion.
• 🌐 Accelerating AI Adoption: To address current healthcare challenges, there is a need to find ways to accelerate the adoption of AI, despite the psychological barriers.
• 🔄 Redesigning for AI: For AI to be effectively integrated, processes, organizations, and value chains must be redesigned to accommodate its capabilities and potential.
• 💡 Empathy in Change: Empathy is crucial in the change process involving AI, ensuring transparency, trust, and human-centric design to overcome the fear of AI.

Q & A

What is the main concern Socrates expressed about the information technology of his time, which he compared to modern AI?

Socrates was concerned that the information technology of his time, specifically writing, would lead people to rely on external sources instead of their own memory and understanding. He believed it created the illusion of knowledge without genuine understanding, similar to the concerns some have with modern AI.

Why does the speaker believe that generative AI presents a distinct challenge compared to other forms of AI?

The speaker believes generative AI presents a distinct challenge because it can engage in nuanced conversations, write complex content, and mimic human-like interactions, which can create psychological fear and unsettle people, as it treads on territory traditionally thought to belong to humans.

What is the 'uncanny valley' concept mentioned in the script, and how does it relate to AI?

The 'uncanny valley' is the idea that as robots or AI become more human-like, we find them more appealing until a point where they are almost human but not quite, causing an eerie feeling and a loss of trust. The speaker relates this to generative AI, which can mimic human interactions closely, potentially causing discomfort.

How does the speaker describe the potential impact of AI on healthcare?

The speaker describes AI as having the potential to help in areas such as reading X-rays, role-playing difficult conversations, and assisting clinicians with personalized care. However, the speaker also acknowledges the fear and resistance that may arise from AI's ability to perform tasks traditionally done by humans.

What does the speaker refer to as the 'spooky Mountain'?

The 'spooky Mountain' is a metaphor used by the speaker to describe the psychological fear and resistance that people may experience when AI starts to perform tasks that were traditionally the domain of humans, such as having nuanced conversations or making ethical decisions.

Why does the speaker suggest that the adoption of AI in healthcare could be slow despite its potential benefits?

The speaker suggests that the slow adoption of AI in healthcare could be due to the industry's risk-aversion, the psychological fear of AI's capabilities, and the time it takes for innovation to become part of routine clinical practice.

What are some of the challenges the speaker identifies in the current healthcare system that AI could potentially address?

The speaker identifies challenges such as insufficient capacity to meet patient needs, burned-out clinicians, high costs, and inconsistent quality. AI could help by improving efficiency, reducing administrative burdens, and enhancing personalized care.

What does the speaker propose as a way to overcome the psychological fear and resistance to AI?

The speaker proposes redesigning processes and organizations to incorporate AI effectively, ensuring transparency and trustworthiness, and infusing the change process with empathy to make it more human-centric and to keep humans in the driver's seat.

How does the speaker suggest redesigning processes to incorporate AI?

The speaker suggests thinking about where AI can do the same tasks differently, better, faster, or cheaper, and modifying processes accordingly. Additionally, designing new processes around AI's ability to do different things that were not possible before.

What role does the speaker see for humans in the AI-integrated systems of the future?

The speaker emphasizes that humans should always be in the loop and in charge, with AI serving as an assistant or tool to enhance human capabilities and decision-making, rather than replacing human judgment and interaction.

Why is it important to ensure that AI-integrated systems are rooted in purpose according to the speaker?

Ensuring that AI-integrated systems are rooted in purpose is important to take people along the journey of change, to address their concerns, and to ensure that the technology serves a meaningful and beneficial role in society, thereby conquering the 'spooky Mountain' of fear and resistance.

Outlines

00:00

📚 The Illusion of Knowledge and AI's Societal Impact

This paragraph discusses the historical skepticism towards new information technologies, like writing, which Socrates criticized for potentially leading people to rely on external sources instead of their own memory and understanding. The speaker draws a parallel to modern AI, noting that while it faces valid criticism, it also has the potential to improve governance and performance. The paragraph introduces the concept of 'generative AI' and its ability to mimic human-like interactions, which can be unsettling and may create psychological barriers to its adoption, a phenomenon referred to as the 'spooky mountain'.

05:02

🤖 Generative AI's Challenges and Opportunities in Healthcare

The second paragraph delves into the unique challenges and opportunities that generative AI presents in the healthcare sector. It contrasts structured AI applications, like analyzing medical images, with unstructured ones, such as role-playing complex conversations. The speaker emphasizes the importance of human involvement in AI processes and the need to address the fear and resistance that generative AI may provoke. The paragraph also highlights the potential benefits of AI in healthcare, such as improving efficiency, reducing costs, and enhancing patient care, while acknowledging the difficulty of integrating AI into risk-averse environments.

10:04

🛠️ Overcoming Fear and Embracing AI Transformation

The final paragraph focuses on the necessary transformations at various levels—processes, organizations, and value chains—to fully leverage AI's potential. It calls for a redesign of processes to better integrate AI, a restructuring of organizations to challenge traditional models, and a reevaluation of industry value chains to create more interconnected ecosystems. The speaker stresses the importance of empathy and effective communication in managing the change process, ensuring transparency, trustworthiness, and a human-centric approach to AI implementation. The goal is to overcome the psychological fear associated with AI and to create a future where AI supports and enhances human endeavors.

Keywords

💡Information Technology

Information Technology refers to the use of computers, software, and networks for the storage, retrieval, and transmission of information. In the video, it is discussed as a tool that can lead to reliance on external sources rather than personal memory and understanding, which echoes Socrates' concerns about writing in the 5th Century BC.

💡Wisdom

Wisdom is the deep understanding of knowledge, experience, and insight that involves questioning, challenging, and refining ideas. The video contrasts wisdom with the illusion of knowledge created by technology, suggesting that true wisdom comes from a genuine understanding rather than superficial access to information.

💡Generative AI

Generative AI refers to artificial intelligence systems that can create new content, such as text, images, or music. The video discusses generative AI as a distinct advancement that raises new challenges and opportunities, particularly in the context of creating a sense of unease or 'spooky' factor due to its human-like capabilities.

💡Uncanny Valley

The Uncanny Valley is a concept where human replicas, such as robots, elicit a response of unease or eeriness when they appear almost, but not quite, human. The video uses this concept to describe the discomfort people may feel when AI starts to mimic human conversation and creativity, blurring the line between human and machine.

💡Artificial Intelligence (AI)

Artificial Intelligence is the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions. The video discusses AI's evolving capabilities and societal fears, as well as its potential to revolutionize various fields, including healthcare.

💡Healthcare

Healthcare is the organized provision of medical services to individuals or communities through various health professionals and allied health fields. The video highlights the potential of AI to assist in healthcare by taking on tasks such as reading X-rays or role-playing difficult conversations, which can help address current challenges in the industry.

💡Innovation

Innovation refers to the process of translating an idea or invention into a good or service that creates value or for which customers will pay. The video discusses the slow adoption of AI in healthcare due to risk-aversion, and how overcoming this fear could lead to rapid innovation and improved patient care.

💡Psychological Fear

Psychological fear is an emotional response to the perception of danger or threat. The video identifies psychological fear as a significant barrier to the adoption of AI, particularly when AI begins to perform tasks traditionally associated with human intelligence and interaction.

💡Ethics

Ethics are the moral principles that govern a person's or group's behavior. In the context of AI, ethics involve considerations of how AI should be used responsibly, ensuring fairness, avoiding bias, and maintaining privacy. The video touches on the importance of ethics in AI, especially in sensitive areas like healthcare.

💡Organizational Change

Organizational change refers to the process of making significant transformations in the way an organization operates. The video calls for a redesign of processes and restructuring of organizations to accommodate AI, emphasizing the need for transparency, trust, and human-centered design in these changes.

💡Empathy

Empathy is the ability to understand and share the feelings of others. The video suggests that empathy should be a key component in the change processes involving AI, ensuring that the transition is human-centric and addresses the emotional responses of those affected by the changes.

Highlights

Technology can lead to reliance on external sources rather than personal memory and understanding.

Wisdom is derived from questioning and refining ideas, unlike the illusion of knowledge created by technology.

Socrates' critique of writing as an information technology in the 5th Century BC is analogous to modern concerns about AI.

Every new technology, from writing to smartphones, has faced societal backlash, and AI is no exception.

AI criticism is often valid and leads to improved governance and better performance, addressing errors, bias, and privacy concerns.

Generative AI presents a new challenge in implementation, distinct from traditional AI applications.

AI's ability to mimic human intelligence, such as playing games or creating art, challenges our understanding of what constitutes intelligence.

The 'uncanny valley' effect is applied to AI, where it becomes unsettlingly human-like, causing psychological fear.

The term 'spooky Mountain' is introduced to describe the fear and resistance to AI that mimics human abilities.

AI's potential in healthcare includes reading X-rays and classifying images, but also raises ethical and moral questions.

Generative AI in healthcare could role-play difficult conversations, requiring a nuanced understanding of human emotion and morality.

The fear of AI taking over human roles, especially in sensitive fields like healthcare, is a significant barrier to adoption.

Health systems worldwide are facing challenges that AI could help address, such as capacity, costs, and quality of care.

AI has the potential to provide personalized care, reduce administrative burden, and revitalize the healthcare system.

To fully leverage AI, we must be willing to redesign processes, organizations, and value chains, challenging traditional models.

The role of empathy in change processes is crucial when integrating AI to ensure transparency, trust, and human oversight.

AI's impact on value chains and industry structures requires rethinking boundaries and relationships within ecosystems.

Overcoming the psychological fear of AI and ensuring human-centric design are key to successful integration and adoption.

Transcripts

foreign

so the problem with this information

technology is that it could lead people

to rely on external sources instead of

their own memory and understanding

and in fact while wisdom comes from

questioning challenging and refining

ideas

this technology creates the illusion of

knowledge without genuine understanding

and in fact it's just as good in

producing fiction as fact

and I could be talking about chat GPT or

any of the large language models

but I'm not

these words are from Socrates in the 5th

Century BC

and he is talking about an information

technology that we call writing

and why do we know that

is because his student Plato wrote it

down

so every new technology since gave out

from writing to smartphones has faced

societal backlash

and AI is no different

why should it be

and granted a lot of the criticism that

AI has been receiving

is for good reason in fact it is going

to result in better performance

improved governance and that's going to

address some of the risks around errors

bias or privacy concerns now it's not

going to be easy but we can see a

pathway for technology

and governance to take us in that

direction we've got private companies

government

coalitions working on that

but I believe that generative AI is has

raised a new a distinct and different

impairment

to implementation after all AI is not

new we are comfortable using it to

choose a movie on Netflix

we're happy to get our cell phones and

our computers give us autocorrect at

least most of the time

but that's because we've always been

able to think about intelligence as

whatever machines have not done yet

and you know think about it when uh when

deep blue defeated Kasparov we said

that's not intelligence that's just

Brute Force analytics let's see if a

computer can beat us at goal

and then when alphago came up with a

program that defeated Lee settle

we said well that's just a machine that

learned how to play really well by

playing against itself many times that's

not intelligence

but we can't hide behind Tesla's theorem

anymore not if you've played with chat

GPT or you've used mid-journey or Dolly

to create art

here's why

we all know about the uncanny valley

right it's that idea that

when robots become more and more

human-like in appearance we find them

more

appealing to us but only up to a point

there comes to a point when they're

almost human but not quite and then they

just look eerie and our survival

Instinct kicks in and we lose our trust

in the robot

now think of what's happening with AI

now in generative AI in particular

we have an AI that can engage you know

nuanced conversation with you you have

ai that can write complex emails and you

have ai that can come up with Dad jokes

well bad ones trust me I'm an expert and

this one says why don't scientists trust

atoms

well because they make up everything

now you know chat GPD could have been

talking about itself

but that's not the problem I'm talking

about here the problem I'm talking about

is that when a computer can do this

it's really unsettling it may not be as

visceral as a response to that image

but the AI is treading on territory that

we thought belonged to us humans that

creates psychological fear and I believe

that that fear is going to be a

significant impediment to our being able

to gain the benefits of AI in the future

I like to call this the spooky Mountain

and it is particularly important

in areas such as Healthcare that are

highly human centered

so you know the AI that we're not afraid

of the Netflix kind we can use in

healthcare so we use it to read an x-ray

so it's a defined problem let's say

we've got to figure out or classify

mammograms into high risk versus low

risk we can train the model using

machine learning on a set of images that

either have or don't have abnormalities

and are appropriately annotated so

that's pretty simple and the Machine

gives us an answer there's no morality

or ethics or anything else built into

that answer

and of course we can have a discussion

about

the advantages and drawbacks of using an

AI in that situation but that's really

about the capabilities

and costs

now when you move to generative AI that

allows you to use the AI in a completely

different type of use case for example

to role play a difficult conversation

between a patient and a doctor

so this is obviously an unstructured

problem that requires to be trained on a

general model

but more importantly it's collaborative

and the AI now needs to converse with

you in a human-like way

and it's not just about logic it needs

to listen to your emotion and it needs

to mimic emotion and Morality In what it

gives back to you

that is spooky AI

and

we find it threatening just imagine that

you're the Doctor Who who would have had

that conversation what does it mean what

is your role when a computer can have

that conversation for you if you're a

researcher what's your role if a

computer can not only invent but also

file the patent so this is obviously

very hard to deal with as humans but

most importantly in healthcare

what if that AI causes harm so we are

risk-averse in healthcare and for very

good reason

but because of that partly because of

that sometimes it can take more than a

decade for Innovation to find its way

into routine into broad routine clinical

practice

but Health Systems around the world

today are suffering we have insufficient

capacity to meet the needs of all our

patients

we have burned out clinicians which only

makes that capacity problem worse we

have high costs and huge administrative

burden and we have inconsistent quality

AI can help in every one of those places

so we are hurting patients today if we

don't figure out a way to accelerate the

adoption of AI

and if we can

just imagine a world in which you or

your loved ones can gain high quality

personalized care

from an AI assisted clinician whenever

you need it for as long as you need it

knowing that a doctor can step in the

right doctor can step in at the right

time

imagine that you are that doctor and now

you have time to focus on what's most

important for your patients

knowing that you have a tireless AI

assistant to back you up on your

decisions but also to take care of all

the drudgery or you're a public health

professional and you have the tools to

pinpoint population risk and Target your

interventions to prevent the next

pandemic and I could go on with examples

like this but suffice it to say that in

this world you'd have rapid Innovation

dramatically reduced trade of burden and

therefore costs and a revitalized

education system that's better suited to

the needs of tomorrow

but to get there

we really have to change everything and

be willing to change everything

so first we need to redesign our

processes let's think about where is it

that AI can do the same thing

differently by that I mean better faster

cheaper and let's modify our processes

to do that

but then let's also think about where AI

can do different things and design new

processes around that

we should also restructure our

organizations why should our

organizations of the future be designed

on models that were developed in the

20th century let's challenge our ideas

around who should be on an executive

team how do we do governance what is the

structure of an organization today

that's based on Legacy function and

location which matters less let's also

think about the role of the individual

the role of a team and how we

collaborate

and how we supervise I mean just think

about it why should

we use outdated Norms around how many

people a manager can can manage

so-called span of control when AI is

going to change the very nature of work

and output and supervision so let's

question all of that

but it's not enough just to change

within organizations AI is going to

affect

entire value chains and if you think

about it the structure of our Industries

is based on those value chains and so

how sectors within an industry

interrelate so in the case of healthcare

that would be payers providers life

sciences and so forth we need to

question those boundaries between the

firms and redraw them and think about

how these organizations interrelate

within an ecosystem

easy right

Okay so

change is not easy and this is going to

be really hard on people particularly

because of what I said earlier about the

psychological fear of change that all

this AI is going to bring

so what do we do

well

first

we make sure that we

are passionate about

effectively communicating the vision at

the other end of this

but even as we do that

as we change our processes organizations

and systems to incorporate AI

we need to dig into that most human of

emotions or characteristics empathy

and Infuse our change processes with

that so what do I mean by this so when I

said change processes earlier

I bet a lot of you are thinking

efficiency Effectiveness

Etc but no we have to make sure that our

new processes that incorporate AI are

transparent and trustworthy

we have to make sure that we when we

design organizations that use AI that

humans are always in the loop actually

not just in the loop but humans are

always in charge in the driver's seat

and as we redesign our systems let's

pause and think deeply about making sure

that they are

rooted in purpose

because only if we do that only if we do

that can we take people with us

and conquer the spooky Mountain

thank you

[Applause]