Exploring the Value of Knowledge in Web3 | #Consensus2024 AI Summit Recap

Summary

The video script discusses the challenges of data privacy and monetization in the AI era, emphasizing the importance of decentralized AI. Shashank Shara, co-founder of GYET, explains how decentralized AI allows individuals to own their data and contribute to learning models without reliance on large corporations. He advocates for a system where people can monetize their intellectual property through AI agents, facilitated by web 3 technology, to prevent monopolies and empower creators. The conversation highlights the need for transparency, attribution, and fair compensation in AI development.

Takeaways

• 😐 The current challenge with AI is the issue of data privacy and monetization, where individuals give away their data for free, which is then used and monetized by others.
• 🤔 Decentralized AI is presented as a solution to data ownership, allowing individuals to control their data and get paid for its use, contrasting with centralized AI where large corporations control the data.
• 👥 The concept of AI being more than just a tool and forming a relationship with humans, akin to a parent-child dynamic, is discussed as a way to build trust and interaction.
• 💡 The importance of transparency in AI data sources and usage is highlighted as a way to make people comfortable with AI and to encourage its adoption.
• 💰 The discussion suggests that the incentive for decentralization in AI comes from the potential for financial gain, similar to how Bitcoin gained momentum.
• 🌐 Web 3 is identified as a crucial component in the decentralization of AI, providing a framework for attribution, validation, and monetization of data.
• 📚 The script touches on the devaluation of intellectual property in the internet era and suggests that decentralized AI could restore value to knowledge and intellectual property.
• 🚀 The potential for a new class of knowledge workers who can monetize their data and intellectual property through AI agents is proposed.
• 🛡️ The script warns against the rise of new monopolies in the AI space and emphasizes the need for open-source and web 3 technologies to prevent this.
• 🌟 The discussion concludes with the idea that an 'arms race' in AI should be about outperforming centralized models, with the goal of creating a more equitable and decentralized AI ecosystem.

Q & A

What is the main problem with AI and data monetization discussed in the script?

The script discusses the issue of individuals giving away their data for free, which is then monetized by others without their consent or benefit. This was a significant problem in the web 2 era and is expected to intensify in the AI-driven world.

What is the definition of decentralized AI as per the script?

Decentralized AI is defined as the ability for individuals to own their own data. It involves everyone contributing to a large learning model, rather than a single giant corporation controlling and training the model in secret.

How does the script relate the concept of decentralization in AI to Bitcoin and Ethereum?

The script draws a parallel between the decentralization in cryptocurrency networks like Bitcoin and Ethereum, where nodes are run by various participants, and the concept of decentralized AI, where data contribution and model training are distributed among individuals.

What is the vision of AI presented in the script, and how does it differ from current AI models like GPT?

The script presents a vision of AI that fosters a deep relationship with humans, where AI aids and enhances human capabilities rather than replacing them. This contrasts with current models like GPT, which are sometimes seen as monolithic and less personalized.

How does the script suggest incentivizing decentralization in AI?

The script suggests that incentivizing decentralization in AI can be achieved by aligning human good with human greed, similar to impact investing. By allowing individuals to own their AI agents and nodes, and by monetizing their data, people can be motivated to contribute to decentralized AI systems.

What role does the script suggest for Web 3.0 in the future of AI and intellectual property?

The script posits that Web 3.0, with its emphasis on blockchain technology and decentralized systems, can play a crucial role in attributing, validating, and monetizing intellectual property in AI, thus restoring the value of knowledge and preventing monopolization.

How does the script address the issue of AI and the potential devaluation of intellectual property?

The script discusses the devaluation of intellectual property due to the internet and suggests that decentralized AI, combined with Web 3.0 technologies, can help restore value by ensuring proper attribution and monetization of contributions.

What is the script's perspective on the relationship between humans and AI?

The script envisions a relationship between humans and AI that is akin to a friendship or a parent-child relationship, where AI is individually trained and trusted, rather than a top-down, prescribed relationship.

How does the script view the role of open source in decentralized AI?

The script views open source as essential for decentralized AI, as it allows for the development of AI agents and nodes that individuals and institutions can own and control, preventing the rise of another monopolist organization.

What is the script's stance on the importance of transparency in AI and data usage?

The script emphasizes the importance of transparency in AI, stating that for AI to be truly beneficial and trusted by humans, it must be clear where the data is coming from, how it is being used, and how individuals can benefit from it.

What challenges does the script identify in the current AI landscape regarding data and intellectual property?

The script identifies challenges such as the monopolization of data by large corporations, the devaluation of intellectual property due to the internet, and the lack of transparency and control over how data is used in AI models.

Outlines

00:00

🤖 The Challenge of Data Monetization in AI

The first paragraph discusses the complex issue of how individuals freely share their data, which is then monetized by others without their consent. It highlights the difference between web 2.0 and the potential amplified risks in the AI-driven future. The speaker introduces the concept of decentralized AI as a solution, emphasizing the importance of data ownership and the role of AI in building trust and relationships, rather than replacing humans. The conversation also touches on the challenges of incentivizing decentralization and the need for AI to be transparent and beneficial to its users.

05:01

🌐 Embracing Decentralization in AI for Intellectual Property Rights

The second paragraph delves into the practical implementation of decentralized AI, using the example of Gyana, a platform that allows individuals to set up AI agents and nodes. It discusses the importance of aligning human good with human greed, ensuring that creators and researchers are rewarded for their contributions to AI models. The paragraph also addresses the devaluation of intellectual property in the internet era and the potential for a new class of knowledge workers who can monetize their data through AI agents, reflecting on the societal and economic implications of these changes.

10:02

🛡️ Combating Monopolies with Decentralized AI and Web 3.0

In the third paragraph, the discussion shifts to the role of web 3.0 in attributing and validating contributions to AI models, which is crucial for monetization and preventing copyright violations. The speaker argues that the combination of open-source AI and web 3.0 can create a fair system for intellectual property rights, contrasting it with the monopolistic practices of web 2.0 companies. The paragraph also touches on the importance of human involvement in training AI models and the potential for a more equitable distribution of wealth in the AI industry.

15:05

🔄 The Future of Decentralized AI and Its Impact on Society

The final paragraph wraps up the conversation by emphasizing the necessity of a decentralized system to handle the vast amount of work involved in AI development fairly and efficiently. It stresses the importance of preventing monopolies and the role of open-source AI and web 3.0 in creating a more equitable and transparent AI ecosystem. The speaker also reflects on the societal impact of AI, suggesting that a decentralized approach can lead to a more inclusive and democratic use of technology.

Keywords

💡Decentralized AI

Decentralized AI refers to a model where data ownership and AI learning models are not centralized within a single entity but are distributed among multiple contributors. In the video, this concept is presented as a solution to the problem of data privacy and monetization, emphasizing the importance of individuals owning and benefiting from their data. The script mentions that in a decentralized AI system, everyone acts as a 'village' to contribute to the learning model, contrasting with the current model where large corporations control and profit from the data.

💡Monetization

Monetization in this context refers to the process of converting data into revenue. The script discusses how individuals can monetize their data by owning it and contributing to AI models, which is a departure from the current scenario where data is often given away for free and others profit from it. The video suggests that decentralized AI can allow individuals to get paid for their data, thus changing the dynamic of who benefits from the use of personal information.

💡Data Privacy

Data privacy is a central theme in the video, highlighting the increasing sensitivity and privacy concerns around personal data. The script points out the risks of giving away personal data for free and the potential misuse by others. Decentralized AI is presented as a solution to enhance data privacy by allowing individuals to control and validate their data, thus preventing unauthorized use.

💡Centralized AI

Centralized AI is contrasted with decentralized AI in the script, referring to AI systems controlled by a single entity or a few large corporations. The video suggests that centralized AI can lead to issues such as data being used without the owner's consent or knowledge, and the profits being monopolized by the controlling entity. Centralized AI is depicted as less transparent and less equitable in terms of data ownership and profit distribution.

💡Intellectual Property

Intellectual property (IP) in the video is discussed in the context of the value and ownership of creative works and research. The script mentions the devaluation of knowledge due to the internet and the challenges faced by creators and researchers in protecting and monetizing their IP. Decentralized AI is posited as a way to empower individuals to own their IP and get paid for its use in AI models.

💡Web 3.0

Web 3.0, or simply 'web 3,' is mentioned as a crucial component in the future of AI and data ownership. It refers to the next generation of the internet, often associated with blockchain technology, decentralized applications, and a more participatory online environment. In the video, web 3 is seen as a means to attribute, validate, and monetize contributions to AI models, providing a transparent and fair system for intellectual property rights.

💡Open Source

Open source in the context of the video refers to the practice of making software or systems freely available for anyone to use, modify, and distribute. The script discusses the benefits of open-source AI systems, which allow for greater transparency, collaboration, and the prevention of monopolies. Open-source AI is presented as a way to empower individuals to set up their own AI agents and nodes, contributing to a more decentralized and democratic AI ecosystem.

💡AI Relationship

The term 'AI relationship' is used in the video to describe the dynamic between humans and artificial intelligence. It suggests that AI should not replace humans but rather aid and assist them, forming a relationship based on trust and transparency. The video emphasizes the importance of AI being trained and contextualized in a way that reflects human values and social interactions.

💡Monopolies

Monopolies are discussed in the video as entities that control a market or industry to the extent that they can limit competition and set prices without constraint. The script warns against the rise of monopolies in the AI space, which could lead to the exploitation of data and the stifling of innovation. Decentralized AI is presented as a counterbalance to prevent the formation of new monopolies and ensure fair competition.

💡Knowledge Workers

Knowledge workers are individuals who create value through their intellectual or creative contributions, such as researchers, writers, or artists. The video suggests that decentralized AI and web 3.0 can empower knowledge workers by allowing them to monetize their intellectual property and data. This empowerment is expected to create a new class of well-earning individuals who can benefit from the AI economy.

💡Copyright

Copyright in the video is discussed in the context of the legal protection of original works of authorship. The script explores the challenges of copyright in the digital age, where traditional protections have been eroded by the ease of copying and distributing content online. The video suggests that decentralized AI and web 3.0 can offer new ways to protect and monetize copyrighted material by providing clear attribution and tracking of usage.

Highlights

The current challenge of data privacy in AI where individuals give away their data for free while others monetize it.

Decentralized AI as a solution to data privacy issues, allowing individuals to own and control their data.

Decentralized AI defined as a system where everyone contributes to a large learning model rather than a single corporation controlling it.

The importance of transparency in data handling and the potential for individuals to get paid for their data contributions.

The analogy of AI as a life form that builds trust with humans, rather than a top-down relationship.

The role of AI as an aid to humans, not a replacement, and the need for a deep human-AI relationship.

The necessity for AI transparency to ensure user comfort and contribution to AI systems.

The comparison between the arms race in Bitcoin and the incentivization needed for decentralized AI adoption.

The idea of aligning human good with human greed in AI development, similar to impact investing.

The concept of open-source AI agents and nodes allowing individuals to monetize their intellectual property.

The potential societal impact of decentralized AI in creating a new class of knowledge workers.

The role of web 3.0 in restoring copyright and intellectual property rights through blockchain technology.

The discussion on the future of copyright in the context of AI and the internet's impact on devaluing knowledge.

The importance of avoiding monopolies in AI to prevent exploitation and to ensure fair compensation for data contributions.

The potential of decentralized AI to empower individuals to monetize their data and intellectual property.

The need for open-source and web 3.0 to work together to prevent monopolistic control in AI.

The human involvement in training AI models and the potential for fair compensation through decentralized systems.

Transcripts

one of the trickiest most vexing

problems facing AI right now right now

is how we all give away our data for

free and then others monetize it and use

it and do God knows what with it that

was already a massive problem in web 2

and it could be on steroids in the AI

world right it's one thing when we give

away what type of sneakers we buy online

or where we live but imagine the data

getting more sensitive and more private

it's going to get trickier and trickier

so to dive deep in this issue and to see

how decentralized AI can be a solution

we have no one bowly the breakup and

Shashank Shara co-founder of gyet thanks

gentlemen enjoy thanks everyone thanks

for being here have a seat Shashank have

a seat all right great so indeed you

know the the spirit of the whole event

of course is decentralization now we

understand what decentralization means

in terms of Bitcoin and ethereum and you

know nodes and who can run them and

who's propagating the network we all get

that but decentralized AI is kind of

different what what is it what is

decentralized AI versus centralized AI

yeah I think uh so decentralized AI is a

lot of buzzwords obviously um we we're

defining it as being able to own your

own data uh Ai and large learning models

uh are trained on all kinds of data

they're contextualized and inferenced

all kinds of ways uh but decentralized

AI uh we think is kind being able to own

your data everyone uh acts as a village

to contribute to that large learning

model it's no longer some giant

Corporation that's trained secretly and

now they're ready to go I don't want to

name any names but you all know who it

is and for us it's being able to take

charge of that data be able to uh I

figure out where it's coming from

attribute it validate it and then get

paid for it hopefully now when I think

about why it matters I think about the

relationship we have with AI in a sense

because it's kind of like it's

artificial intelligence so we build

relationships with it you know Bitcoin

is about securing commercial

relationships and being able to transact

in a way that we can have more trust

right because we know the network is

decentralized now you know I wrote a

book on AI and Bitcoin and it was all

about this idea of us having an eyeball

an a eyeball and indeed in my story

they're each individually trained and

we're able to trust where the

information is coming from and we're

able to make these kind of friendships

with them almost like a or even a parent

child relationship does that Vision make

sense to you was was my story on its

right track instead of instead of some

Professor top down relationship that's

being prescribed to you you're able to

raise this sort of life form and

of course I mean I mean think about AI

right I mean our our recent relationship

with AI has been through chat GPT right

and that was released to you maybe you

guys interacted with it starting about a

year and a half ago uh there's others

who maybe are in the test net now when

it was released it was presented as uh

Skynet right it was like Hey we're ready

to go but that's not what AI is supposed

to be or at least we don't believe that

the future of AI we think that AI needs

to be uh have a deep relationship with

humans it is here to help humans not to

replace humans it's to Aid not to uh

replace so how do you do that well you

have to go back to kind of how humans

even interact right humans are uh social

people uh as you said a family bond uh

you're the way to make people feel

comfortable uh with AI and then

contribute to Ai and utilize it for

their own lives is to have uh make sure

that it's transparent on where that data

is coming from how that data is working

and then have a stake in actually

getting paid from it using it and kind

of benefiting from it well now that's

all well and good but we're in an arms

race here right in the same way that

Bitcoin was able to gain momentum in the

arms race it was able to do so because

people made money it wasn't because we

nagged a bunch of people to love

decentralization that it worked out

right what we were able to do was make

more money than the people who didn't

get decentralization and then eventually

they're like oh oh we should do it that

way now how do we win the arms race how

do we incentivize decentralize because

right now I use chat TBT it empowers my

own writing it empowers my own

communication but there's things about

it I don't like I have to trick it

constantly I have to hypnotize it to do

what I want I've got to motivate it I've

got to you know there's all these crazy

things that I have to do to get it to do

what I want so how do we do this with

training these models how do we get it

so individuals want to train the models

and and really cultivate this

relationship yeah I mean I love for

example I I love uh impact investing

right I love a and the way we I always

approach that is aligning human good uh

with uh human greed with doing good uh I

think that's the same for uh AI as well

uh so over Guyana what we're doing is uh

allowing folk to open source set up

their own uh AI agents their own nodes

and we were our firm was born out of uh

project at Berkeley our professor our

CTO Dr Allen Yang he he started to build

virtual Tas at Berkeley and the

reasoning behind that was you know he

wanted he had a hard time hiring people

to be honest for TAS but he couldn't go

to

GPT because they would take his data

they it's expensive it's slow uh when so

he started building his own node and

then permissioning access and for a

researcher that's a big step you cannot

the moment year he he put 20 years of

his research on a node that's his life's

work and so obviously he should he can

and should get paid for it um and really

I think it it pointed towards uh uh a

kind of a wrong that even the internet

perpetrated on intellectual property uh

the internet was has been a really great

don't get me wrong it proliferated

knowledge but at the same time it

devalued that knowledge right and when

you devalue the knowledge uh that's when

you start to see a lot of societal

problems you see newspapers shift to

more uh clickbait uh uh material you

start to see uh ad you know kind of a

more ad-based Revenue just a survive and

so many newspapers went bankrupt and now

you see the the Pinnacle of

it uh these news newspapers and IP

holders are starting to sue open AI now

because open AI has and others have

trained on a hundred years of their

newspaper workor and other IP to get to

where they're at but it's not a specific

model it doesn't really like you said

you have to train along uh we believe

fine-tune learning models uh fed by uh

nodes that are

uh individuals of uh companies

institutions who have worked hard to

create their knowledge uh is the way

forward so in our Ideal World the last

generation was you know YouTube allowed

creators to proliferate and have a

middle class set of creators unlike

before in Hollywood where it was the

halves and the Have Nots you have a

thousands of YouTubers making good money

we think the next stage of allowing

people to own their data and monetize it

will create another class of well-

earning knowledge workers it could be an

individual just blogging at home writing

at home writing poetry writing songs

whatever that that IP is they can

monetize it and that monetization looks

like in the form of an AI agent that

contains their knowledge and they can

permission who gets to access it and for

how much well you know I'll be the

devil's advocate for a second because my

own ideas have to change when I when I

graduated from law school my thesis I

wrote it on the sort of future of

copyright a lot of issues there was Gene

patenting at the time that was looking

terrible right that that that fell apart

thankfully the Supreme Court struck that

down um but my prediction in the paper

which did well I got an A probably the

only a I got in law school um the theory

was that the the copyright came in with

the printing press and was about to go

out with the printing press now this was

before there was AI it was before there

was Bitcoin um now I still don't know

what the future of copyright is uh but I

think things have changed can it go back

can can copyright become an institution

again like it was because it's melting

away right I mean you look at music

sampling no one it's not offensive now

to copy you know 30 years ago a cover

song was like a cover band it was like

yeah I want the real thing right

nowadays everyone's so used to it it

doesn't matter right that that whole

thing became wishy-washy you know how do

we

or do we find new ways to to make money

off of of intellectual property I'm not

sure what the answer is how do how do

you fit into this and how ises AI fit

into yeah so I think that's where kind

of web 3 actually fits into it uh AI I

mean we look on our side we're just

developing really great toolkits to help

you own your data the web 3 component is

absolutely crucial to restoring the

copyright use mention because the

problem is right now once again with web

2 and what's happening is you don't know

where

different components were used in the

case of a music sample you know they mix

too many bits up to properly track right

so that's where web 3 is crucial to the

story you need attribution you need

validation to lead to monetization right

so if you can attribute if everything's

on chain and you know who contributed

what if New York Times is starting to

permission a specific newspaper headline

for to to feed a Al algorithm or a

learning model then it they'll know how

many times it was uh inferenced and uh

what they should get paid uh you put

everything on chain you can't mess it up

and basically if New York Times thinks

there was a VI violation then you can

use our normal court system you can go

to through the normal Court copyright

courts patent law and uh and sue them

and you'll know exactly how many

violations were so if you can at least

track it I think that's a solution to it

yeah perhaps because my my theory was

the old Canadian media slogan the media

is the message and when the internet

came the message was the physical

isolation or the physical um you know

one of one or whatever the that it

wasn't in in bits and bites

automatically meant that the medium

paper books printing had value of course

we took that away now the way you

described it goes back to what I said

earlier this is an arms race yeah right

just like in Bitcoin there's an arms

race here and in Bitcoin I see the arms

racist trying to get richer than the

Fiat people so that people will say well

those are the ones I want to follow it's

not an intellectual thing right yeah so

is that is that what's going on are we

do we find a way to make all these AI

people stand out and if you use

copyright this way you're just going to

be able to lead the way because you're

successful yeah I mean it should be an

arms race uh but the Army the armies are

you know centralization versus

decentralization I mean for example

we're an open source decentralized

project you guys can go download us and

we're not trying to take money from that

um but the Army here is is you guys it's

it's people who actually exist in in

real life uh it's not just some

ephemeral uh training bot uh so yeah you

need to arm up and I think you know the

there's many open source llms now uh

that fed by uh your nodes hopefully

whether use us or set up your own uh AI

agents I think that becomes the Army but

uh the key here is we cannot allow

another monopolist organization to be

between us and the end product uh we

like it we it love it yeah we love the

Applause thank you thank you thank you

uh the key here is the web to has seen

the death of so many companies so many

small and midsize potential businesses

because of the monopolist and I it's

great look we I use every Fang stock

ever i' I've I'm invested in Facebook

Netflix Google Amazon like don't get me

wrong it's made life really easy but

when you're talking about owning your

data and then getting paid for it uh and

contributing to it you we need to remove

the the the centralized player you

cannot do it in web 2 because it'll lead

to the rise of another monopolist and as

you see in Amazon and the others they

will wait till their time or just like

uber they will wait till their time and

charge you three times five times more

and and you'll have to pay it at that

point they've driven the taxi company

out of business you have nowhere else to

go and the last thing I'll say is that

uh you need open source and you need uh

web 3 to work together to combat that

yeah because you know something I

learned at this show in another AI talk

and and I didn't know this that that the

big breakthrough that chat GPT had was

they figured out that human sorder

were a big part of the language training

models they pay people in the

Philippines to say dog cat dog cat until

for hours and hours and hours yeah um

and there you know it's a job like you

know it's not bad uh and I heard that

chat GPT spent you know they're going to

spend a billion dollars this year on

people just humans

going so indeed you know we need humans

to put this data to to tag it to to

upload it into a node and and hopefully

make more than the $1 or2 hour the folks

in the Philippines are making for doing

this stuff yeah you know that would be

great yeah I mean I think that's where I

think web 3's already solved that issue

by by staking by uh you know validating

someone's work um that's like at the

core of web 3 so uh yeah I mean in

funnily enough I mean it's funny we uh

it's it's two big buzzword you know we a

lot of people interested AI web 3 how

does it fit together are we just driving

up a bunch of money uh and a bunch of

people hopping on it it it's but when

you actually go under the hood of uh how

humans uh interact uh to form Ai and

form learning models and neural networks

uh it's crucial to uh have a

decentralized system from the Monopoly

uh monopolist angle but also just pure

there can't be a centralized system

can't handle that kind of work uh and

not in a fair way so this centralized

system of vent production can handle the

work of going overboard please please a

warm Round of Applause for Shashank

everyone thank you thank you