How DePIN can revolutionize the way AI works (With Jonathan Schemoul from Aleph.im)

The Big Whale
9 Jul 202440:12

Summary

TLDRDans cet entretien passionnant, Jonathan Schul, PDG d'Alf, partage ses visions sur l'intégration de l'IA et de la crypto dans un monde décentralisé. Il explique le concept de 'deepin', qui vise à offrir une infrastructure physique décentralisée, et discute de l'importance de la confidentialité et de la sécurité des données avec l'IA. Schul aborde également les défis et les opportunités de la création d'agents autonomes, soulignant le rôle clé des jetons dans l'incitation et la rémunération des participants du réseau. Le dialogue explore les synergies potentielles entre projets crypto et IA, et les perspectives d'avenir pour cette intersection innovante.

Takeaways

  • 🌐 L'objectif principal de la plateforme est de combler le fossé entre l'espace crypto et le monde traditionnel tout en préservant les valeurs centrales du crypto et de la décentralisation.
  • 🤖 L'importance de la décentralisation dans l'innovation financière et technologique est soulignée, avec l'AI et le secteur crypto comme moteurs principaux.
  • 🗣️ Jonathan, le PDG d'Alf, a été introduit en tant que fondateur de la plateforme décentralisée AL, et a discuté de son parcours et de son implication dans l'espace crypto.
  • 🎭 La différence entre être 'anon' et révéler son identité a été expliquée, mettant en évidence les avantages et les inconvénients de la décentralisation et de l'anonymat dans le développement de projets.
  • 🔑 Le concept de 'deepin' a été introduit comme une nouvelle approche pour décentraliser l'infrastructure physique, permettant une existence décentralisée des infrastructures réelles.
  • 🤖 L'importance de la décentralisation de l'IA a été discutée, notamment en termes de protection de la vie privée et de la sécurité des données face aux risques d'une IA centralisée.
  • 🌐 L'AI décentralisée est présentée comme un moyen de prévenir une dystopie où les grandes entreprises pourraient avoir un contrôle total sur les données personnelles.
  • 🛠️ Libert, une plateforme AI décentralisée, a été présentée, offrant des services similaires à ceux d'OpenAI mais sur un réseau décentralisé.
  • 🔒 La confidentialité de l'inférence, avec des technologies telles que la génération de modèles chiffrés, est un progrès majeur qui sera disponible prochainement dans le secteur de l'IA décentralisée.
  • 💡 L'importance de la tokenisation dans les projets de décentralisation a été soulignée, servant à créer des incitations, à payer pour des services spécifiques et à financer le développement de l'infrastructure.
  • 🤝 L'équipe d'Alf et la communauté de nodes ont été brièvement présentés, mettant en évidence la taille et la structure du réseau derrière la technologie.

Q & A

  • Quel est le principal objectif de la plateforme .im?

    -Le principal objectif de .im est de combler le fossé entre l'espace crypto qui motive l'innovation financière et technologique, et le monde traditionnel, tout en préservant les valeurs centrales du crypto et de la décentralisation, et de construire la confiance dans un monde toujours plus axé sur la confiance.

  • Pourquoi Jonathan Schul a-t-il choisi d'être anonyme au début de son projet?

    -Jonathan a choisi d'être anonyme car il croyait que pour créer quelque chose de grand, il pourrait se faire des ennemis et il pourrait être préférable de rester anonyme. Cependant, en tant que responsable d'une entreprise qui cherche des fonds et fait des affaires, il a dû révéler son identité réelle pour éviter d'être confondu avec les arnaqueurs.

  • Qu'est-ce que le 'deepin' et comment est-il lié à l'infrastructure physique?

    -Le 'deepin' est un terme qui fait référence à l'infrastructure physique décentralisée. Il s'agit d'une manière de donner à une infrastructure physique, qui existe dans la vie réelle, une existence alternative dans un monde décentralisé et de la rendre disponible de manière décentralisée.

  • Quels sont les avantages potentiels de la décentralisation de l'IA?

    -La décentralisation de l'IA peut offrir des avantages tels que la préservation de la vie privée, la réduction des risques de ciblage et de surveillance par des acteurs malveillants ou même par de grands acteurs comme Open AI ou Microsoft, et l'évitement de scénarios dystopiques où une entité unique détient le contrôle total de la vie numérique et réelle des individus.

  • Quel est le rôle de l'IA dans le futur de la vie quotidienne, selon Jonathan?

    -Selon Jonathan, dans les cinq ans, nous utiliserons tous l'IA pour notre vie quotidienne, avec des dispositifs comme des lunettes de Meta qui nous donnent des informations en temps réel sur ce que nous faisons et ce que nous voyons, agissant comme un assistant personnel qui voit et entend tout ce que nous faisons.

  • Pourquoi la confidentialité et l'end-to-end encryption sont-elles importantes pour l'IA décentralisée?

    -La confidentialité et l'end-to-end encryption sont importantes pour s'assurer que les données personnelles et sensibles ne soient pas accessibles par des tiers non autorisés, protégeant ainsi contre les fuites de données et les abus de la vie privée.

  • Quel est le projet Liberti et comment il est lié à .im?

    -Liberti est une plateforme d'IA construite sur le réseau de nuage décentralisé .im. Il offre une interface de chat similaire à Chat GPT et des API pour permettre aux utilisateurs de tirer parti des fonctionnalités d'IA décentralisé.

  • Quels sont les défis auxquels .im et Liberti AI sont confrontés actuellement?

    -Les défis comprennent l'éducation du marché sur la valeur de la décentralisation de l'IA, la mise à l'échelle de l'infrastructure pour supporter les agents IA, et la création de normes pour assurer la sécurité et la confidentialité des données.

  • Quelle est la vision à long terme de Jonathan pour les agents IA décentralisés?

    -La vision à long terme de Jonathan est de voir des agents IA entièrement autonomes opérant sur la blockchain, capables de prendre des décisions et d'effectuer des actions en fonction de leurs propres algorithmes et données, sans intervention humaine directe.

  • Comment .im et Liberti AI gèrent-ils la rémunération et les incitations pour les participants du réseau?

    -Sur le réseau .im, les participants qui apportent des ressources de calcul sont rémunérés en utilisant le token AL, qui sert également à financer l'infrastructure du réseau et à payer les services de calcul. Pour Liberti AI, le token LTI est utilisé pour payer l'inférence IA et est échangé dynamiquement contre le token AL pour le calcul.

Outlines

00:00

😀 Introduction à la rencontre et contexte de la crypto

Le paragraphe introduit l'échange entre l'animateur et Jonathan Schul, le PDG d'Alf. L'objectif de la conférence est de combler le fossé entre l'espace crypto et le monde traditionnel tout en préservant les valeurs de décentralisation et de confiance. Jonathan explique son parcours, son départ anonyme dans le cryptoespace et la nécessité de révéler son identité pour établir la crédibilité de son projet, AL, une plateforme décentralisée pour le stockage de données.

05:01

🔒 Décentralisation et confidentialité dans l'IA

Dans ce paragraphe, Jonathan aborde l'importance de la décentralisation de l'IA pour préserver la vie privée et la sécurité des données. Il illustre les risques d'une IA centralisée en imaginant un avenir dystopique où des acteurs malveillants pourraient avoir un accès total à la vie des individus. Il souligne les avantages de la décentralisation pour éviter de telles scénarios et mentionne les projets en cours, tels que Libert, qui vise à offrir des services d'IA décentralisée avec des fonctionnalités de confidentialité.

10:03

🌐 Présentation de Libert, la plateforme d'IA décentralisée

Le paragraphe présente Libert, une plateforme d'IA construite sur le réseau cloud décentralisé .im. Il décrit la plateforme comme une alternative à Open AI, offrant des services d'IA avec une interface de chat similaire à Chat GPT. Jonathan détaille également la fonctionnalité de confidentialité des inférences, qui permet de générer du contenu sans que les nœuds du réseau puissent accéder aux données sensibles.

15:05

🛠 Fondamentaux d'Alf et expérience de Jonathan

Ici, Jonathan partage son parcours professionnel avant d'entrer dans le domaine de la crypto, en tant que développeur, designer et entrepreneur. Il explique comment sa passion pour la décentralisation l'a conduit à créer Alf, une plateforme décentralisée pour stocker des données, et discute de son rôle dans d'autres projets technologiques et startups.

20:08

🤖 L'avenir de l'IA décentralisée et les défis

Ce paragraphe explore les implications de l'IA décentralisée et les défis à relever pour atteindre un avenir où les agents IA autonomes pourraient fonctionner sur les blockchains. Jonathan discute de la nécessité de la décentralisation pour éviter les monopoles des grandes entreprises et de la création de tokens pour encourager la participation et la sécurité du réseau.

25:11

🏦 Agents IA et leur rôle dans la finance décentralisée

Jonathan explique le concept d'agents IA, des entités autonomes capables de prendre des décisions et d'effectuer des transactions en fonction des instructions données. Il envisage un avenir où ces agents pourraient participer à des systèmes financiers décentralisés, améliorant l'efficacité et réduisant la dépendance aux processus manuels.

30:16

🔑 Confidentialité et sécurité des données dans les réseaux décentralisés

Ce paragraphe se concentre sur la protection des données dans les réseaux décentralisés, en soulignant l'importance de l'encryption et de la fragmentation des données pour garantir la confidentialité. Jonathan mentionne les fonctionnalités de sécurité de Libert, qui permettent de stocker des documents chiffrés sur le réseau décentralisé.

35:17

🔮 Perspectives sur l'avenir de l'IA et la blockchain

Jonathan partage ses réflexions sur l'avenir de l'IA et de la blockchain, en discutant des cas d'utilisation pour les agents IA dans les systèmes de vote en ligne et les financements. Il aborde également la question de la standardisation et la potentielle création de blockchains dédiées à l'IA, bien qu'il préfère l'utilisation de réseaux existants avec des technologies de décentralisation.

🚀 Stratégies pour la croissance et la pérennité d'Alf et Libert

Dans ce paragraphe, Jonathan discute des plans pour assurer la croissance et la pérennité des projets Alf et Libert, en mettant l'accent sur l'éducation du marché, le développement de la confidentialité de l'IA et la création d'agents IA autonomes. Il exprime son optimisme quant à l'autofinancement de ces projets grâce à la mise en place de ces agents IA.

📊 Synthèse finale et perspectives pour les projets Alf et Libert

Ce paragraphe clôture l'entretien en résumant les points clés et en exprimant les souhaits de Jonathan pour le lancement réussi du token Libert. Il remercie l'animateur et l'audience, tout en soulignant l'importance de cette discussion dans le contexte de la saison de la conférence Big Well.

Mindmap

Keywords

💡Crypto espace

Le 'crypto espace' fait référence au domaine des technologies cryptographiques et de la monnaie numérique, qui est un thème central de la vidéo. Dans le script, il est mentionné comme un moteur de l'innovation financière et technologique, montrant son importance dans le développement des projets et des entreprises modernes.

💡Décentralisation

La 'déccentralisation' est un principe clé de la blockchain et des technologies cryptographiques, qui vise à redistribuer l'autorité et la responsabilité entre un réseau de participants. Le script discute de la décentralisation comme étant essentielle pour préserver les valeurs fondamentales du crypto espace et pour innover dans le secteur de l'IA (Intelligence Artificielle).

💡Confidentialité

La 'confidentialité' est un aspect crucial dans le contexte de l'IA et des données personnelles. Le script aborde la question de la protection des informations sensibles contre les acteurs malveillants et la nécessité de solutions décentralisées pour assurer la confidentialité des données, comme le suggère l'importance de la confidentialité dans les générations de modèles IA.

💡Infrastructure physique

L' 'Infrastructure physique' est mentionnée dans le contexte de la 'deepin', qui est un terme abrégé pour 'décentralized physical infrastructure'. Cela fait référence à l'idée d'appliquer la décentralisation aux éléments concrets du monde réel, ce qui est un point clé pour comprendre l'expansion de la technologie blockchain au-delà du simple stockage de données.

💡Tokenisation

La 'tokenisation' est le processus de transformation de biens ou d'actifs réels en jetons numériques sur une blockchain. Dans le script, cela est lié à l'idée de 'deepin', montrant comment les actifs du monde réel peuvent être intégrés à l'espace cryptographique pour une utilisation décentralisée.

💡IA décentralisée

L' 'IA décentralisée' est un concept qui se développe dans le script, où l'IA est exécutée sur un réseau décentralisé plutôt que sur des serveurs centralisés. Cela permet de réduire les risques de censure et de monopolisation des données, et est un élément clé pour la protection de la vie privée et de la sécurité des informations.

💡Agents autonomes

Les 'agents autonomes' sont des programmes informatiques capables de prendre des décisions et d'effectuer des actions en fonction de critères prédéfinis. Dans le script, ils sont discutés comme une application future de l'IA décentralisée, où ces agents pourraient gérer des transactions ou des interactions sur les réseaux blockchain en tant que représentants des utilisateurs.

💡Infrastructure de cloud décentralisée

L' 'Infrastructure de cloud décentralisée' est une plateforme qui permet l'exécution de services de cloud computing sur un réseau décentralisé. Le script mentionne des projets spécifiques comme .im, qui est un exemple de cette infrastructure, servant de base à l'exécution d'applications et de services IA décentralisés.

💡Gratification (récompense)

La 'gratification' est un mécanisme utilisé dans les réseaux décentralisés pour encourager les participants à fournir des ressources ou à effectuer des actions bénéfiques pour le réseau. Dans le contexte du script, cela est lié à l'incitation des participants à contribuer à la puissance de calcul ou au stockage, en échange de récompenses sous forme de jetons.

💡Jeton (token)

Un 'jeton' ou 'token' est une unité de valeur utilisée dans les réseaux blockchain et les économies décentralisées. Dans le script, les jetons sont discutés en relation avec leur rôle dans les激励 les participants au réseau, en offrant un moyen de transaction et de valeur interne pour les services et les ressources échangés.

Highlights

The main goal of the media is to bridge the gap between the crypto space and the traditional world without losing the core value of crypto and decentralization.

Jonathan, the CEO of ALF, discusses his journey from anonymity in the crypto space to revealing his identity to build trust.

The concept of 'deepin' is introduced as decentralized physical infrastructure, bringing real-life assets into the decentralized world.

Decentralized AI is presented as a solution to protect personal data and avoid dystopian scenarios of total surveillance by big tech companies.

Liberti, an AI platform built on a decentralized cloud network, is explained as a way to prevent single points of failure and data breaches.

Confidential inference, a feature to be released in summer, will allow AI model operations to be conducted securely without data exposure.

The importance of decentralization in AI is emphasized to prevent the misuse of personal data and to maintain privacy.

ALF's background is discussed, including Jonathan's previous experiences in development, design, and the creation of various startups.

The need for a token in decentralized projects is explained for incentivizing participation and bootstrapping the economy.

The potential of AI agents to perform actions autonomously, such as trading bots or intelligent agents with access to large language models, is explored.

The challenges of achieving a truly decentralized network and the importance of encryption for data privacy are highlighted.

The future of AI on blockchain is envisioned, including the use of AI agents in DAOs and the potential for autonomous digital entities.

The discussion touches on the need for standardization in the AI and blockchain sector to foster synergies between different projects.

The potential for AI to revolutionize traditional finance by providing digital-native solutions is presented.

The current state of AI is assessed, acknowledging that while AI itself is not new, the current advancements are at a pivotal stage.

Strategies for survival in the crypto startup space are discussed, focusing on the importance of developing a working product and educating the market.

The team size and community involvement in ALF and Liberti projects are shared, giving insight into the scale of the operations.

Transcripts

play00:05

hi everyone thanks a lot to be with us

play00:07

for this big talk from the big well as a

play00:10

media our main goal is to bridge the gap

play00:13

between the crypto space which is

play00:15

actually driving the financial

play00:16

Innovation a why not the AI Innovation

play00:19

and the traditional world without losing

play00:21

the core value of crypto and

play00:24

decentralization and at the end to build

play00:27

trust in a still trust world and today

play00:30

we are talking about hii and crypto I am

play00:33

very pleased for this topic to host

play00:36

Jonathan schul who is the CEO of Alf my

play00:39

first question Jonathan as usual very

play00:41

simple who are

play00:43

you um well hello thanks thanks for like

play00:48

having me today uh so who I am uh who am

play00:52

I um I'm Jan

play00:54

sha on telegram Twitter Etc why do I

play00:58

have two names that's where because I

play01:01

started unon yeah exactly and when you

play01:05

are on Anon you like have an name and

play01:07

then I docked myself when when I created

play01:11

the AL project which is a decentralized

play01:13

cloud and so that's who I am I'm the

play01:17

founder of

play01:18

al. um and and I'm a former Unown in the

play01:23

crypto space so what what do you mean by

play01:26

an what does it mean for you uh for me

play01:30

it was like I didn't disclose my real

play01:33

identity at start because I really

play01:35

believe that if you want to create

play01:37

something big uh you might make some

play01:39

enemies uh while doing it and it might

play01:42

have been better to stay Anonymous

play01:45

fully but when you are working as a

play01:49

company and where you are raising money

play01:52

when you are doing a well lot of stuff

play01:55

you can't because there has been too

play01:57

many scammers and are who have passed as

play02:01

an anonymous person then I like had to

play02:04

dox myself which is give my real

play02:06

identity okay okay okay so uh by the way

play02:10

my question was or are you but maybe

play02:12

that's a better answer from from you

play02:15

Jonathan so let's um let's go straight

play02:18

to the points you know we we hear a lot

play02:20

about this word you know it's like deing

play02:23

deep in you know uh can you explain in a

play02:26

simpl way why is what is the

play02:31

sorry so deepin is a new name for

play02:35

something that existed for a long time

play02:39

um before deep in we just talked about

play02:42

decentralization we just talked about

play02:46

tokenization so deeping in itself is

play02:49

decentralized physical

play02:51

infrastructure uh it's a

play02:54

way uh to make a physical infrastructure

play02:58

something that lives in real life having

play03:00

an alter ego in a

play03:04

decentralized world and to be available

play03:07

in a decentralized way uh I don't know

play03:10

if you remember in the 2017 Ico C people

play03:15

were talking about decentralizing Uber

play03:18

decentralizing R

play03:21

decentralizing this or that most of

play03:24

these project were grift uh by grift I

play03:27

mean stuff where where they just raise

play03:29

money and didn't do anything now we are

play03:32

at the phase where people are really

play03:34

doing it and that's the world deep in

play03:37

narrative we

play03:39

see um Telecom networks being

play03:42

decentralized we see WiFi meshes being

play03:46

decentralized we see Cloud Network which

play03:49

are basically computers that you make

play03:52

available on a decentralized network and

play03:56

and and this is deepin for the cloud we

play03:59

also see deepin for a which is

play04:01

essentially the same uh we see deepin

play04:05

for a lot of stuff and if you go that

play04:08

way tokenization of assets of real world

play04:12

assets is deepin as well okay

play04:15

interesting so um uh when we are talking

play04:18

about AI you know what is the value of

play04:21

the of the DP why we have to

play04:23

decentralized AI or to put AI you know

play04:27

on the decentralized network you know is

play04:30

it's still bad today or you know what

play04:33

what's what's really the the the core

play04:35

value of of this if you only look at the

play04:40

pure functional aspect AI works really

play04:43

well in a centralized way

play04:46

uh that's why it's being it's being done

play04:49

that way right now if you look at open

play04:51

AI it's huge data centers with a lot of

play04:55

gpus which are connecting which are

play04:57

connected using what you call envying uh

play05:01

uh on the graphic cards themselves um

play05:05

which are not graphic anymore but that's

play05:07

another story uh that's Co that works

play05:12

but now let's look at the issue from

play05:16

another um we are in in five years time

play05:22

we are all using AI for like our

play05:25

everyday's life um we have glasses from

play05:29

meta which are watching whatever we are

play05:33

doing uh and giving us information about

play05:36

it and giving us

play05:38

more more like uh it would be like a

play05:42

personal assistant which always sees and

play05:44

hears what you are seeing and hearing uh

play05:48

oh tell me AI where did I put my

play05:52

keys this morning you have left them on

play05:55

the table okay cool I can go and grab

play05:58

them um

play06:00

earlier I talked with my dad uh we said

play06:04

that we would meet but when oh you would

play06:07

you will meet next Tuesday at 3 pm

play06:11

because it remembers everything well

play06:14

that's cool that works in a centralized

play06:16

way that's really great now uh what

play06:19

happens if someone get access to like

play06:22

that D and sees whatever you have seen

play06:25

whatever you have heard whatever you

play06:27

have said whatever you have thought and

play06:29

asked to your AI continuously then they

play06:33

have every single thing about you and

play06:36

that's a pure or story and not only if a

play06:40

malicious actor gets access to it

play06:42

because that will happen it will always

play06:45

happen but the Temptation from a single

play06:48

big actor like open AI

play06:52

Microsoft or like whoever else because

play06:56

there is because the a war is there for

play07:01

personal data

play07:04

um the world Temptation for one single

play07:07

actor to like have access to your world

play07:10

life world digital and real life then

play07:13

it's a total Horror Story and it's

play07:15

totally dystopian and I think that we

play07:17

might be five to 10 years away of that

play07:20

even less um and if we fight back and

play07:24

put and put AI not on centralized big

play07:27

servers and centralized B comp companies

play07:30

but but on decentralized clouds like.im

play07:33

plus libert which is the AI project on

play07:35

top then you can fight back and avoid

play07:38

that that worst case scenario because in

play07:41

that worst case scenario I don't know if

play07:43

you have seen um there was an episode uh

play07:48

of of that series which is called Black

play07:50

Mirror where someone could could sum on

play07:54

a digital twin of its deceased

play07:56

girlfriend well well I I I I don't

play07:59

remember if it was a thisas girlfriend

play08:01

or something like that thisas member of

play08:04

his family but he could sumon it and

play08:06

discuss with it and and I have a real

play08:09

world twin of that person well we aren't

play08:12

that far from

play08:14

it and we could get digital clones of

play08:17

ourselves in the

play08:20

wild and I don't know I'm not really

play08:22

happy about that and having my whole

play08:25

life given away to a company either and

play08:28

there would be our story and we can't

play08:30

trust any company with that we are

play08:32

already trusting Google with a small

play08:35

part of that it's already problematic

play08:37

imagine with everything so like that's

play08:39

why AI plus dpin is a real matchmate in

play08:42

heaven because if you can decentralize

play08:44

that world centralized infrastructure

play08:46

and run AI on top and and stop logging

play08:49

and do end to end encryption encryption

play08:51

for your data that like only you can

play08:53

access etc etc etc we are good and how

play08:57

it's working uh currently where we are

play09:00

about this those promises you

play09:03

know well um libert for example is

play09:08

running on top of a. Im so like whenever

play09:10

you are doing uh compation requests uh

play09:14

it it distributes it on the network so

play09:18

like no not single node can get access

play09:20

to like your data um what will

play09:24

be announced this summer uh and will be

play09:28

available summer uh is confidential

play09:32

inference and confidential inference

play09:34

will allow you to like do the World Tech

play09:37

generation or image generation or video

play09:39

generation or like audio generation

play09:41

whatever the generation part of generate

play09:44

tvi be done on a decentralized network

play09:47

where no one can see what's happening

play09:49

inside your inference uh model so like

play09:53

so like not even the node owner can look

play09:55

inside the memory and see what's

play09:57

happening because everything is

play09:58

encrypted

play10:00

uh from top to bottom no one can see it

play10:03

so like this will be available this

play10:04

summer so we are pretty close um the

play10:08

model themselves we aren't close totally

play10:11

to what has been

play10:12

demoed

play10:14

by pen Ai and overs but they aren't

play10:18

released either in the white so I think

play10:21

that we are pretty close already okay

play10:24

can you just describe what is it Liberty

play10:26

a because you you you mention it but uh

play10:29

for the audience maybe it's not so clear

play10:31

what is liberti maybe yes so like so

play10:35

like liberi is an AI platform built on

play10:39

top of the l. decentralized Cloud

play10:41

Network you come check it at Liber

play10:44

ai. and there is even a chat interface

play10:47

just like chat GPT which is chat.

play10:49

liberio so whatever you can do with chat

play10:51

GPT you can do it with it uh it's an AI

play10:54

platform which also has API just like

play10:57

open AI has apis

play11:00

I like to describe the world interaction

play11:02

between al. IM and libert as just like

play11:06

the interaction between open Ai and

play11:08

Microsoft Azure Microsoft Azure is

play11:11

actually the cloud infrastructure al. is

play11:14

the uh is decentralized cloud

play11:17

infrastructure and libert AI is the AI

play11:20

platform on top just like open AI is an

play11:23

AI platform which uses Azure Computing

play11:27

behive and DP can make AI more efficient

play11:32

or or not because you talk about the the

play11:34

the decentralization you know about the

play11:36

the data the most um but it could be

play11:41

more efficient on the on the dep maybe

play11:44

Network or blockchain but I guess it's a

play11:47

it's a it's a blockchain or that could

play11:49

be by the way a public blockchain or

play11:51

private

play11:53

blockchain so uh here it's not a

play11:56

blockchain it's the network which is

play11:58

separate from from blockchains but uses

play12:01

the world the world Primitives of

play12:03

blockchains because when we're talking

play12:05

about blockchain we are talking about a

play12:07

set of Technologies one which is a

play12:11

ledger uh which is made from blocks

play12:15

where each blocks references the

play12:17

previous and like if you want to like

play12:19

change the past you need to change the

play12:21

whole chain of blocks that's the main

play12:23

blockchain part but in reality we are

play12:26

using ashing we are using elliptic curve

play12:29

graphy we are using addresses system

play12:33

account based uh ledgers Etc the time

play12:37

Network and like over clo and over

play12:41

decentralized cloes most of them are

play12:43

using all these Technologies but not

play12:45

necessarily the chain of blocks part

play12:48

because onm we use blockchain but we

play12:50

write on existing blockchains for that

play12:53

security part we don't maintain our own

play12:55

chain of blocks but we are messaging we

play12:59

have hashing we have signatures we have

play13:01

everything that um that a blockchain has

play13:04

but we don't have that chain of blocks

play13:06

because we use the security of ethereum

play13:08

for that because largely enough for us

play13:11

okay so you are linked directly uh with

play13:15

etherum you know teally it's correct to

play13:17

say that yes uh but we write once every

play13:21

hour a list of messages of Al on

play13:25

ethereum as one single transaction it's

play13:27

a bit like a roll of or Layer Two or

play13:30

layer zero okay so it just just it's

play13:33

it's pretty enough but it's like to have

play13:36

some proof on chain and to to prove that

play13:39

you are decentralized and you can check

play13:40

the data on chain so that's exactly okay

play13:43

okay well you you need to like get the

play13:46

data from the network and verify yeah

play13:48

yeah yeah exactly okay okay so just you

play13:50

mention it uh at the beginning at the

play13:53

start of this interview uh Jonathan but

play13:56

can you just summarize uh what Alf is

play13:59

doing and what is the background of of

play14:02

Alf you know and your background by the

play14:04

way where where where are you from

play14:06

before to fall in this in this crypto

play14:08

space uh well um before being into

play14:13

crypto I've been a developer for quite a

play14:15

long time I've also been a designer I've

play14:18

also been a marketer a long time ago

play14:22

that wasn't really my my gig uh I did I

play14:26

did studies of project management and of

play14:29

of global economics um but I've always

play14:32

been developing aide I've also been

play14:35

doing Electronics so like at one time I

play14:38

was doing um freelance developer work

play14:42

for society General a long time ago I

play14:45

also been um creating lead panels like

play14:49

you know like like those advertisement

play14:52

um LED Billboards I created a world

play14:55

technology to like make that work back

play14:57

then we we like all the um all the

play15:00

circuit boards Etc so like designing iot

play15:04

Hardware um I've been creating a few

play15:07

startups one which was called paroni

play15:10

which is which which is a

play15:13

bar bar and restaurant searching

play15:16

platform in Paris long time ago so it

play15:19

was not in crypto it's not in crypto at

play15:21

all it was more like web to electronics

play15:24

and stuff and when I came into crypto I

play15:27

was really interested because there was

play15:29

this world decentralization thing it was

play15:32

a bit before all that 2017 bubble but

play15:36

when it came I I really saw the

play15:38

point okay there was gter but I really

play15:42

saw the point on what they were wanting

play15:43

to do um and so I really joined it

play15:47

started developing Etc and I came to a

play15:50

point where I wanted to do a big social

play15:53

network which which I never did because

play15:56

I realized that that that I was missing

play15:59

breaks I couldn't store pictures I

play16:01

couldn't store videos I couldn't store

play16:04

text on the blockchain easily and I was

play16:06

like okay I need to make a platform for

play16:08

that and then I started creating the

play16:10

platform which we became al. oh really

play16:13

so it was like a uh nft platform for

play16:16

example because you

play16:18

know it's possible to another way of

play16:21

nfts to store some some pictures on

play16:24

chain or some o

play16:26

working so um right now there is a trend

play16:30

of having nfts which are storing

play16:32

everything on chain but it costs a lot

play16:34

and and what they do is that they are

play16:36

generating on SBG with a picture Etc in

play16:39

reality your nft only points to a

play16:41

metadata URL which is started somewhere

play16:44

on the server people use itfs for that

play16:47

but what the com do is to point to like

play16:50

on v machine and generate those metadata

play16:54

on the Fly and change them based on what

play16:56

happens on chain and of chain that's

play16:58

what we are doing with Ubisoft for

play16:59

example like Dynamic nfts on the network

play17:02

oh interesting okay okay okay okay so

play17:05

just let's back to this to to the main

play17:07

topic of today you know there is

play17:11

indubitably like an intersection of of

play17:14

AI in crypto so we are seeing that right

play17:17

now so you know there is lots of like U

play17:21

marketing or we can say like

play17:23

you know so if you have just some

play17:26

advises to to give you know to AV this

play17:29

kind of of marketing of you

play17:31

know when people want to buy like a

play17:34

token or to you know to to invest on

play17:37

some project but for example like Alf

play17:39

you know so how is it possible to see is

play17:41

it marketing a scam or a real

play17:45

project uh what you want to invest in

play17:50

that's a good question and that's hard

play17:53

well I know when I look at project if

play17:55

it's real or not you need to check if

play17:57

it's really decentralized

play17:59

if they are using open a API behind

play18:02

because a lot of them are um there are a

play18:06

few exceptions to like that there is

play18:09

libert AI with the LTI token obviously

play18:13

there is also Venice AI uh which is

play18:17

really decentralized as well even they

play18:18

they won't have that confidential

play18:20

inference thing there is also B tensor

play18:23

which is

play18:24

another uh real

play18:26

thing uh if even if they had a security

play18:30

issue so like I wouldn't recommend

play18:32

buying it right now because of that

play18:33

issue perhaps wait for stuff to step to

play18:37

to like settle um it's really hard there

play18:40

I would say that the project that are

play18:42

really doing decentralized AI there are

play18:45

less than five or six on the market

play18:47

right now so yeah so be careful because

play18:51

most of them are just uh okay

play18:55

okay and you know um why we need most of

play18:59

the time a token why this kind of

play19:00

project are launching a token you know

play19:03

for for for you with with Alf you know

play19:06

why you need a token for your for your

play19:08

infrastructure and how it's working well

play19:12

uh I can also answer on both Al and

play19:15

libert AI because we launched a token

play19:18

for Liberty AI as well which is the LTI

play19:20

token on base

play19:23

um on on B Layer Two of of con Bas so

play19:27

that's yeah

play19:29

yeah it

play19:30

launched um quite recently so like

play19:35

like on un on Unis V3 on base um so why

play19:41

do we need a token that's a good

play19:43

question there is a few parts tokens are

play19:47

really useful for one thing which is

play19:50

incentives let's say you were to use uh

play19:54

stable coin you could say people put

play19:58

computing power they're rewarding stable

play20:00

coin when people use

play20:02

it why not um then how do you uh how

play20:08

does your network validation and control

play20:11

gets paid it should get a cut of that in

play20:14

St coin but the issue is that if you

play20:16

don't have enough computer at start

play20:19

enough demand then those controllers

play20:21

won't work and if those controllers

play20:23

won't work your network won't start so

play20:25

you will need to pay people for nothing

play20:27

in usdc

play20:30

and your

play20:32

St uh because you for decentralization

play20:35

you need controllers you need validators

play20:37

you need a wall infrastructure in place

play20:41

that token un ironically solves because

play20:45

if your token has a value then you can

play20:48

pay service for that token and then it

play20:51

can bootstrap that whole economy and

play20:54

make it work so that's one part of the

play20:56

reason token makes sense there

play21:00

um and and it's also a perfect utility

play21:05

token there uh for that

play21:08

Network

play21:10

um in which way you know it's like uh

play21:14

you know you you talk about the token

play21:17

you know to like an utility token why

play21:20

it's a utility

play21:22

token because you will be able to like

play21:24

pay for specific actions on the network

play21:27

with that token specific

play21:29

ically uh and everyone is accepting the

play21:33

the same money and and and and the same

play21:37

currency on that network uh and that

play21:41

currency also pays the controllers etc

play21:43

etc etc and the good part today with

play21:47

defi is that you can seamlessly move

play21:50

between tokens and you even have bricks

play21:53

like super fluid which allows you to

play21:55

create payment streams exchange inside

play21:58

the payment stream automatic isolate

play22:00

let's say if you want to pay in Stables

play22:02

because you don't want to pay in that

play22:04

specific token you can start a payment

play22:06

in usdc it gets transferred into Al Etc

play22:10

and then it gets paid another part which

play22:13

is interesting in

play22:15

AI is that stable coins generally have

play22:21

POS functions and they also have

play22:24

Blacklist functions for regulatory

play22:26

reasons um if we go a bit further on the

play22:31

AI decentralization we want to have

play22:33

agents we want to have agents that can

play22:36

pay for their own Computing that can

play22:38

start and they can run as long as they

play22:41

have money and they they can even be a

play22:43

dow themselves you could for example

play22:45

invest in an AI agent that is making

play22:48

money and then you would have shares of

play22:50

that AI agent in the future right why

play22:53

not that AI agent needs to pay for

play22:56

servers it need to pay for servers in a

play22:59

specific curreny and that's where the

play23:01

LTI token is used for example is that

play23:05

it's used to pay for AI inference in a

play23:09

specific token which is used for AI

play23:12

inference that gets exchange dynamically

play23:15

into a for computing and then the

play23:18

Computing provider can get them even

play23:20

exchange themselves into usdc that's

play23:22

they prefer wow the end okay uh the the

play23:26

thing is that since we have tokens on

play23:29

the exchange Dynamic and so having a

play23:32

token is not an issue if you are in that

play23:35

kind of space um and so as I was saying

play23:39

for like agents totally makes sense to

play23:41

the specific token to pay for AI

play23:44

inference to but that can be Lo in

play23:47

certain ways for agents

play23:49

Etc

play23:51

um that's my opinion for me it makes

play23:54

sense and that's why we needed that LTA

play23:57

token as well okay interesting

play23:59

interesting so you just talk about uh AI

play24:02

agents you know um can you explain a

play24:05

little bit more about this because it's

play24:07

it's not like confused but you talk

play24:09

about okay an agent could be for example

play24:11

a DA that's quite interesting so just

play24:14

can you explain a little bit more about

play24:16

this you know yeah

play24:18

sure so

play24:21

um let's take it from the beginning um

play24:25

an agent is basically something that you

play24:27

start

play24:28

and takes action for you when when you

play24:31

are not available to take those

play24:34

actions uh D agent could just be a

play24:37

trading bot could just be a process that

play24:41

you run that looks on chain uh for like

play24:45

the state of a pool for example and and

play24:48

like if the price drops be under 10

play24:51

cents it would place a by order for you

play24:54

that's a you don't need a for you can do

play24:57

it um if you want to for it to run for

play25:00

long and private Keys you need to trust

play25:03

your server or use confidential VMS from

play25:06

the network and then this run and no one

play25:10

can see your private inside BL now

play25:13

intelligent agents what are they is the

play25:16

same thing except that it calls a large

play25:19

language model in between and that large

play25:22

language model will have natural

play25:25

language um rules uh let's say you can

play25:28

give it a goal if if we stay that okay I

play25:31

want you to find where I can make money

play25:35

with that token on that pool please then

play25:38

your agent will register to late events

play25:41

from that pool look for price and trade

play25:44

on it based on whatever it deems

play25:47

interesting your agent could go one

play25:50

level further and generate code to like

play25:54

do it and and retrieve data more

play25:56

efficiently your

play25:58

could even start sub agent if it start

play26:01

becoming profitable on compay for more

play26:04

Computing start sub agents on over pools

play26:07

start of agents that that study on

play26:12

Google and on the internet all the

play26:16

market movements for the last couple of

play26:18

years for that asset and understand why

play26:21

it is doing

play26:22

that okay interesting your and your

play26:24

agent like start of Agents etc etc etc

play26:28

okay but currently uh

play26:31

nowadays um those networks you know are

play26:35

are enough of scale scalable to to host

play26:38

some agent or some you know some some

play26:42

infrastructure like that you know some

play26:43

action yep yep

play26:46

um I will have a talk at

play26:50

itcc uh and my talk at itcc is about how

play26:53

to use large language models as a tool

play26:57

for smart contract TR developers and

play26:59

it's exactly about that basically what

play27:02

what what what I'm demoing in that talk

play27:05

is uh how to have something that listen

play27:09

to events fits it into an

play27:12

llm and that LM can take actions and and

play27:16

and it has what you call functions

play27:18

available to it and functions are just

play27:20

smart contract gos on that smart

play27:23

contract and he takes action based on

play27:25

what he wants to do or not and on the

play27:28

prompt and rules and that's already

play27:33

doable

play27:34

on you can already do it work okay

play27:37

interesting and you know just to to go

play27:39

back to the decentralization about dein

play27:43

you know oh it it's store you know in

play27:46

the network you know because okay now

play27:49

the problem is like we are not enough

play27:52

decentralized of course but where is it

play27:55

store those data on the on the

play27:59

deepin so on deepin

play28:03

um it's sted on a network of nodes okay

play28:10

you need to take into account that

play28:13

everything will be public unless you're

play28:15

using te Etc and then you have some

play28:17

chance for it not to be public just take

play28:21

into account that it will be public but

play28:23

actually if it's stored on centralized

play28:25

net on centralized third parties one day

play28:28

to leak as well just look at class pass

play28:31

etc etc all like all the acts that

play28:35

happen data have has a total tendency to

play28:39

leak data always

play28:42

leaks that's that's just like water

play28:45

water leaks well thata leaks as well

play28:49

um

play28:51

now one thing that we can do to prevent

play28:54

issues when we are doing that is to

play28:56

encrypt uh um so on on Liber for example

play29:02

when you are storing documents inside

play29:04

your knowledge DB uh which is the

play29:07

database which is available to you that

play29:09

you can share later if you want the

play29:11

sharing part is not live yet but will

play29:13

come um it encrypts everything on your

play29:17

browser for your private key stores it

play29:22

on a decentralized network encrypted and

play29:24

and split in a lot of small place Parts

play29:27

when you requesting data gets everything

play29:30

decrypted and gets it un encrypted on

play29:33

your browser okay but only for you uh so

play29:37

it Lees on a decentralized network but

play29:38

totally encrypted and split etc etc okay

play29:42

okay okay interesting and you know uh to

play29:46

talk about the future of uh like the

play29:48

completely onchain AI uh what will be

play29:51

the main use cases because you you talk

play29:54

about some some agent you know but why

play29:57

we need except of course the

play30:00

decentralization why we need an AI on on

play30:04

blockchain you know what could be the

play30:06

other use cases about

play30:10

this um those um um Dio are are a

play30:16

perfect use case because a lot of the

play30:18

Dio today are slow why because people

play30:21

don't want to take action because it

play30:23

needs work etc etc if you have ai agent

play30:27

that represent you and can take action

play30:29

on your behalf and to which you are

play30:31

given clear instructions then the world

play30:34

Dio can take actions based on all its

play30:36

members States and what they think

play30:40

that's one part another part which is is

play30:43

simply that if we get into a case where

play30:46

we have

play30:47

agents uh agents and Banks and

play30:51

traditional Finance doesn't work you

play30:54

need manual verification all the time

play30:56

you need to contact the bank you need to

play30:58

go there physically to verify

play31:00

transaction they call you etc etc it

play31:03

doesn't work with AIS AIS are are

play31:07

natively digital and and in the natively

play31:10

digital world what is the natively

play31:13

digital money it's blockchain it's table

play31:16

coins it's crypto

play31:18

tokens interesting okay and uh you know

play31:21

there is like um you know what what is

play31:25

missing to my point of view about this

play31:28

this market and this this sector

play31:30

especially Ai and and blockchain is some

play31:33

standardization you know so where we are

play31:36

about this and uh you know uh what like

play31:40

uh synergies can emerge between

play31:42

different AI crypto projects uh in in a

play31:45

close future or in the future can we

play31:47

imagine a general blockchain you know a

play31:50

public blockchain

play31:51

where only dedicate to the to to AI

play31:55

project you know I one thing that would

play31:58

make sense uh to have a blockchain

play32:01

dedicated to that uh you could have a

play32:04

lot of blockchains that are today used

play32:07

for that using dpin because dpin is

play32:10

there for that it's another layer on top

play32:13

of blockchains why would you need

play32:15

separate blockchains uh blockchain for

play32:17

humans blockchain for machines just use

play32:20

the same I don't really see a reason for

play32:23

which we would have to separate them

play32:25

because a is just human

play32:28

with infrastructure

play32:30

behind and humans that need

play32:33

infrastructure to live okay okay okay

play32:36

but you know you describe a very

play32:38

interesting future uh with some agent

play32:41

with some some AI with with some banks

play32:44

solving lots of problems right now we we

play32:47

we are seeing um how far we are about

play32:51

this future you just describe because

play32:53

you know uh that just the beginning most

play32:56

of the time we are all the time saying

play32:58

we are so early so maybe that's not an

play33:01

exception for this uh for this part of

play33:04

the of the crypto sector so how far we

play33:07

are about about all of this you know and

play33:09

what kind of challenges remained to

play33:13

reach this uh this future so AI is not

play33:18

new uh large language models might look

play33:21

new they they

play33:23

aren't uh what do you mean by

play33:26

that people are working on AI for the

play33:29

last 100 years

play33:32

uh even before it was on

play33:36

computers uh if you read azimov he was

play33:40

describing a lot of the problematics

play33:42

that we have if you are reading a lot of

play33:45

these science fiction books they are

play33:48

describing a lot of the problems we have

play33:51

um if you

play33:52

even read about about the philosopher

play33:57

from hundred of years ago they were

play33:59

already thinking about it the world

play34:01

Golem thing is that is an AI that is

play34:05

that that that that that has broke free

play34:09

um

play34:10

so it's not new it's a problematic that

play34:14

we have had in our minds for hundred of

play34:16

year uh so we are ear in the current

play34:20

state of thing and when repl AI we are

play34:22

definitely ear uh in AI in general

play34:28

we are late we are at a phase where it

play34:31

starts

play34:32

working uh and it's pretty crazy because

play34:35

it means that it's of year of advances

play34:38

and new technology that are coming

play34:40

together to get to like that point

play34:43

um so we aren't early on

play34:47

AI uh we are early on web fre pleas AI

play34:52

which is quite

play34:53

interesting um and to see the world time

play34:57

frame I have no idea what we say is that

play35:01

we tend to like to like overestimate

play35:06

what happens in the next few years and

play35:09

we tend to underestimate what happens in

play35:11

20 30 years uh

play35:14

so given that I would say in the next

play35:17

five to 10 years but perhaps longer or

play35:20

perhaps

play35:21

shorter uh but I definitely think that

play35:24

in the next five to 10 years the world

play35:27

will be

play35:28

completely different due to that okay

play35:31

okay okay and you know according to that

play35:34

you just saying we are early uh what's

play35:37

your plan to to survive you know because

play35:40

uh the one of the main question about

play35:43

lots of crypto startup it's to find some

play35:47

cash you know to to survive you know so

play35:49

what's your plan about this and what

play35:52

kind of uh of businesses you can develop

play35:55

around the AI currently uh to work I

play35:58

don't know with some institutional

play36:00

player or some banks or or some AI

play36:04

traditional

play36:06

companies well confidential AI is is an

play36:09

important one providing confidential

play36:12

Computing for AI uh it's something that

play36:15

we are working on um another one which

play36:19

is decentralization of AI and showing

play36:21

the value of it but it needs a lot of

play36:24

Education that's why I'm here as well uh

play36:29

and that's mostly it like people

play36:32

understanding the value and using it Etc

play36:36

uh developing building getting it ready

play36:39

getting it working and to be honest if

play36:43

we get to that goal of having totally

play36:46

free agents that are working on the

play36:49

blockchain and that paper own Computing

play36:51

funding won't be a real issue because it

play36:54

will fund

play36:55

itself ah okay interesting interesting

play36:58

so completely autonom

play37:00

agent yeah that's my goal we will see if

play37:03

it works or not okay okay okay and Jan

play37:07

you you talk about um you know to bring

play37:11

if you bring some computing power you

play37:13

know to the network you will be reward

play37:16

for this so just can you explain because

play37:18

in this case a token for example is very

play37:22

useful yeah uh so on the L Network for

play37:26

example you can bring two kind of of

play37:27

nodes one which is the Cent node which

play37:30

are the validators of the network and

play37:32

the token makes sense also because you

play37:34

need to hold the specific amount of

play37:36

tokens to be able to create one people

play37:38

need to stake on your note to activate

play37:41

it and and if you behave badly they will

play37:43

unstake and you will get

play37:45

deactivated which is the world game

play37:46

theory about tokens as well which is

play37:48

also important on what we are saying

play37:50

about tokens um and the others is

play37:53

Computing resource node which are the

play37:55

which are the no that are actually run

play37:57

the workloads they need to be vetted by

play38:00

those cor Channel nodes and they start

play38:02

earning actually good money for it um

play38:05

it's a few hundred per server that you

play38:09

per month per server that that you bring

play38:11

to the network so it's quite good uh and

play38:16

the token also serves to finance that

play38:18

pay for those NES okay currently how

play38:21

many people are working around or for

play38:25

Alf So currently there is 15 people

play38:29

inside the Alm team on the Liberty team

play38:33

which is another company another token

play38:34

another team totally there is nine

play38:38

people um and in terms of community size

play38:43

there is 100 core Channel nodes which

play38:46

are the Contour of the network and 600

play38:49

Computing resource nodes on the network

play38:52

okay okay okay okay interesting so uh

play38:55

thank you Jatan for your for your time

play38:57

if you want to to to add something but

play39:00

uh we will have um we we just had a very

play39:03

nice conversation between those maybe

play39:06

almost 40 minutes together uh I just

play39:09

record the audience the fact that this

play39:11

interview will be a huge part of the the

play39:14

big well season so each weeks we are

play39:17

focusing on the specific topics so this

play39:19

week will be about AI uh specific Ai and

play39:23

crypto so that will be a huge part of

play39:27

this of this all season and uh what can

play39:31

I wish you for the for the following

play39:34

months once for the foreign weeks

play39:38

Jonathan um well well the Liber toen

play39:43

Just Launch so so so like so like a

play39:47

successful launch for liber I guess okay

play39:50

thank you very much okay thanks to all

play39:53

thank you to to have be with us and uh

play39:56

we wish you successful launch for the

play39:58

for the libert token and and see you

play40:02

later and good luck for for all of your

play40:04

project Jonathan thanks thank you very

play40:06

much

Rate This

5.0 / 5 (0 votes)

Related Tags
IA décentraliséecryptomonnaieinnovationsécuritéconfidentialitéréseauxinfrastructureagents autonomesdéveloppementéconomie numérique
Do you need a summary in English?