AI set to benefit from blockchain-based data infrastructure

The rise of Chat­G­PT has been noth­ing short of spec­tac­u­lar. With­in two months of launch, the arti­fi­cial intel­li­gence (AI)-based appli­ca­tion reached 100 mil­lion unique users. In Jan­u­ary 2023 alone, Chat­G­PT reg­is­tered about 590 mil­lion visits.

In addi­tion to AI, blockchain is anoth­er dis­rup­tive tech­nol­o­gy with increas­ing adop­tion. Decen­tral­ized pro­to­cols, appli­ca­tions and busi­ness mod­els have matured and gained mar­ket trac­tion since the Bit­coin (BTC) white paper was pub­lished in 2008. Much needs to be done to advance both of these tech­nolo­gies, but the zones of con­ver­gence between the two will be excit­ing to watch.

While the hype is around AI, a lot goes on behind the scenes to cre­ate a robust data infra­struc­ture to enable mean­ing­ful AI. Low-qual­i­ty data stored and shared inef­fi­cient­ly would lead to poor insights from the intel­li­gence lay­er. As a result, it is crit­i­cal to look at the data val­ue chain holis­ti­cal­ly to deter­mine what needs to be done to get high-qual­i­ty data and AI appli­ca­tions using blockchain.

The key ques­tion is how Web3 tech­nolo­gies can tap into arti­fi­cial intel­li­gence in areas like data stor­age, data trans­fers and data intel­li­gence. Each of these data capa­bil­i­ties may ben­e­fit from decen­tral­ized tech­nolo­gies, and firms are focus­ing on deliv­er­ing them.

Data storage

It helps to under­stand why decen­tral­ized data stor­age is an essen­tial build­ing block for the future of decen­tral­ized AI. As blockchain projects scale, every vec­tor of cen­tral­iza­tion could come to haunt them. A cen­tral­ized blockchain project could suf­fer gov­er­nance break­down, reg­u­la­to­ry clam­p­down or infra­struc­ture issues. 

For instance, the Ethereum net­work “Merge,” which moved the chain from proof-of-work to proof-of-stake in Sep­tem­ber 2022, could have added a vec­tor of cen­tral­iza­tion to the chain. Some have argued that major plat­forms and exchanges like Lido and Coin­base, which have a large share of the Ethereum stak­ing mar­ket, have made the net­work more centralized. 

Anoth­er vec­tor of cen­tral­iza­tion for Ethereum is its reliance on Ama­zon Web Ser­vices (AWS) cloud stor­age. There­fore, stor­age and pro­cess­ing pow­er for blockchain projects must be decen­tral­ized over time to mit­i­gate the risks of a sin­gle cen­tral­ized point of fail­ure. This presents an oppor­tu­ni­ty for decen­tral­ized stor­age solu­tions to con­tribute to the ecosys­tem, bring­ing scal­a­bil­i­ty and stability.

But how does decen­tral­ized stor­age work? 

The prin­ci­ple is to use mul­ti­ple servers and com­put­ers world­wide to store a doc­u­ment. Sim­ply, a doc­u­ment can be split, encrypt­ed and stored on dif­fer­ent servers. Only the doc­u­ment own­er will have the pri­vate key to retrieve the data. On retrieval, the algo­rithm pulls these indi­vid­ual parts to present the doc­u­ment to the user.

Recent: Tok­enized mort­gages can pre­vent anoth­er hous­ing bub­ble cri­sis, says Casper exec

From a secu­ri­ty per­spec­tive, the pri­vate key is the first lay­er of pro­tec­tion, and the dis­trib­uted stor­age is the sec­ond lay­er. If one node or a serv­er on the net­work is hacked, it can only access part of the encrypt­ed data file.

Major projects with­in the decen­tral­ized stor­age space include File­coin, Arweave, Crust, Sia and StorJ.

Decen­tral­ized stor­age is still in a nascent state, how­ev­er. Face­book gen­er­ates 4 petabytes (4,096 ter­abytes) of data dai­ly, yet Arweave has only han­dled about 122TB of data in total. It costs about $10 to store 1TB of data on AWS, while on Arweave, the cost is about $1,350 at the time of publication.

Undoubt­ed­ly, decen­tral­ized stor­age has a long way to go, but high-qual­i­ty data stor­age can boost AI for real-world use cases.

Data transfer

Data trans­fer is the next key use case on the data stack that can ben­e­fit from decen­tral­iza­tion. Data trans­fers using cen­tral­ized appli­ca­tion pro­gram­ming inter­faces (APIs) can still enable AI appli­ca­tions. How­ev­er, adding a vec­tor of cen­tral­iza­tion at any point in the data stack would make it less effective.

Once decen­tral­ized, the next item on the data val­ue chain is the trans­fer and shar­ing of data — pri­mar­i­ly through oracles.

Ora­cles are enti­ties that con­nect blockchains to exter­nal data sources so that smart con­tracts can plug into real-world data and make trans­ac­tion decisions.

How­ev­er, ora­cles are one of the most vul­ner­a­ble parts of the data archi­tec­ture, with hack­ers tar­get­ing them exten­sive­ly and suc­cess­ful­ly over the years. In one recent exam­ple, the Bonq pro­to­col suf­fered a $120 mil­lion loss due to an ora­cle hack.

Besides smart con­tracts and cross-chain bridge hacks, ora­cle vul­ner­a­bil­i­ties have been low-hang­ing fruit for cyber­crim­i­nals. This is main­ly due to a lack of decen­tral­ized data trans­fer infra­struc­ture and protocols. 

Decen­tral­ized ora­cle net­works (DONs) are a poten­tial solu­tion for secure data trans­fer. DONs have mul­ti­ple nodes that pro­vide high-qual­i­ty data and estab­lish end-to-end decentralization.

Ora­cles have been used exten­sive­ly with­in the blockchain indus­try, with dif­fer­ent types of ora­cles con­tribut­ing to the data trans­fer mechanism.

There are input, out­put, cross-chain and com­pute-enabled ora­cles. Each of them has a pur­pose in the data landscape. 

Input ora­cles car­ry and val­i­date data from off-chain data sources to a blockchain for use by a smart con­tract. Out­put ora­cles allow smart con­tracts to car­ry data off-chain activ­i­ty and trig­ger cer­tain actions. Cross-chain ora­cles car­ry data between two blockchains — which could be fun­da­men­tal as blockchain inter­op­er­abil­i­ty improves — while com­pute-enabled ora­cles use off-chain com­pu­ta­tion to offer decen­tral­ized services. 

While Chain­link has been a pio­neer in devel­op­ing ora­cle tech­nolo­gies for blockchain data trans­fer, pro­to­cols like Nest and Band also pro­vide decen­tral­ized ora­cles. Apart from pure blockchain-based pro­to­cols, plat­forms like Chain API and Cryp­toAPI pro­vide APIs for DONs to con­sume off-chain data securely.

Data intelligence

The data intel­li­gence lay­er is where all the infra­struc­ture efforts of stor­ing, shar­ing and pro­cess­ing data come to fruition. A blockchain-based appli­ca­tion using AI can still source data from tra­di­tion­al APIs. How­ev­er, that would add a degree of cen­tral­iza­tion and could affect the robust­ness of the final solution. 

How­ev­er, sev­er­al appli­ca­tions are tap­ping into machine learn­ing and arti­fi­cial intel­li­gence in cryp­to and blockchain. 

Trading and investments

For sev­er­al years, machine learn­ing and arti­fi­cial intel­li­gence have been used with­in fin­tech to deliv­er robo-advi­so­ry func­tion­al­i­ties to investors. Web3 has tak­en inspi­ra­tion from these appli­ca­tions of AI. Plat­forms source data on mar­ket prices, macro­eco­nom­ic data and alter­nate data like social media, gen­er­at­ing user-spe­cif­ic insights.

The user typ­i­cal­ly sets their risk and returns expec­ta­tions, with the rec­om­men­da­tions from the AI plat­form falling with­in these para­me­ters. The data required to deliv­er these insights is sourced by the AI plat­form using oracles. 

Bit­coin Loop­hole and Numerai are exam­ples of this AI use case. Bit­coin Loop­hole is a trad­ing appli­ca­tion that employs arti­fi­cial intel­li­gence to pro­vide trad­ing sig­nals to plat­form users. It claims to have over 85% suc­cess rate in doing so. 

Numerai claims it is on a mis­sion to build “the world’s last hedge fund” using blockchain and AI. It uses AI to col­lect data from dif­fer­ent sources to man­age a port­fo­lio of invest­ments like a hedge fund would.

AI marketplace

A decen­tral­ized AI mar­ket­place thrives on the net­work effect between devel­op­ers build­ing AI solu­tions at one end, and users and orga­ni­za­tions employ­ing these solu­tions at the oth­er end. Due to the application’s decen­tral­ized nature, most com­mer­cial rela­tion­ships and trans­ac­tions between these stake­hold­ers are auto­mat­ed using smart contracts.

Devel­op­ers can con­fig­ure the pric­ing strat­e­gy through inputs to smart con­tracts. Pay­ment to them for using their solu­tion could hap­pen per data trans­ac­tion, data insight or just a flat retain­er fee for the peri­od of use. There could also be hybrid approach­es to the price plan, with the usage tracked on-chain as the AI solu­tion is used. The on-chain activ­i­ties would trig­ger smart con­tract-based pay­ments for using the solution.

Sin­gu­lar­i­tyNET and Fetch.ai are two exam­ples of such appli­ca­tions. Sin­gu­lar­i­tyNET is a decen­tral­ized mar­ket­place for AI tools. Devel­op­ers cre­ate and pub­lish solu­tions that orga­ni­za­tions and oth­er plat­form par­tic­i­pants can use through APIs. 

Fetch.ai, sim­i­lar­ly, offers decen­tral­ized machine learn­ing solu­tions to build mod­u­lar and reusable solu­tions. Agents build peer-to-peer solu­tions on this infra­struc­ture. The eco­nom­ic lay­er across the entire data plat­form is on a blockchain, enabling usage track­ing and smart con­tract trans­ac­tion management.

NFT and metaverse intelligence

Anoth­er promis­ing use case is around non­fun­gi­ble tokens (NFTs) and meta­vers­es. Since 2021, NFTs have been viewed as social iden­ti­ties by many Web3 users using their NFTs as Twit­ter pro­file pic­tures. Orga­ni­za­tions like Yuga Labs have gone one step fur­ther, allow­ing users to log in to a meta­verse expe­ri­ence using their Bored Ape Yacht Club NFT avatars.

As the meta­verse nar­ra­tive ramps up, so will the use of NFTs as dig­i­tal avatars. How­ev­er, dig­i­tal avatars on meta­vers­es today are nei­ther intel­li­gent nor do they bear any resem­blance to the per­son­al­i­ty that the user expects. This is where AI can add val­ue. Intel­li­gent NFTs are being devel­oped to allow NFT avatars to learn from their users.

Recent: Uni­ver­si­ty stu­dents reveal new Web3 solu­tions at ETH­Den­ver 2023

Matrix AI and Althea AI are two firms devel­op­ing AI tools to bring intel­li­gence to meta­verse avatars. Matrix AI aims to cre­ate “avatar intel­li­gence,” or AvI. Its tech­nol­o­gy allows users to cre­ate meta­verse avatars as close to them­selves as possible. 

Althea AI is build­ing a decen­tral­ized pro­to­col to cre­ate intel­li­gent NFTs (iNFTs). These NFTs can learn to respond to sim­ple user cues through machine learn­ing. The iNFTs would become avatars on its meta­verse named “Noah’s Ark.” Devel­op­ers can use the iNFT pro­to­col to cre­ate, train and earn from their iNFTs.

Sev­er­al of these AI projects have seen an increase in token prices along­side the rise of Chat­G­PT. Yet, user adop­tion is the true lit­mus test, and only then can we be sure that these plat­forms solve a real prob­lem for the user. These are still ear­ly days for AI and decen­tral­ized data projects, but the green shoots have emerged and look promising.