Opinion by: Salman Avestimehr, co-founder and CEO of ChainOpera
With a market value expected to reach over $1.3 trillion by 2030, AI has captured the imagination of users and investors alike, enabling the world to work smarter and make more informed decisions. Yet, the rapid progress of generative AI applications has occurred mainly under a centralized development model wherein two key contributors — users and developers — are largely disempowered with no voice, stake or ownership over the development process.
The centralized model of app creation, combined with the lack of inclusion of users and developers, means that agents and apps today lack the true personalization that will take AI agents to the next level of adoption and growth toward an expected trillion-dollar AI economy. Through decentralizing AI app and agent creation, users and developers will be empowered with ownership, and better, more innovative applications will result starting in 2025.
The generative AI application ecosystem
With more than half of the global population already owning a smartphone and generative AI smartphone shipments set to grow 364% year-over-year in 2024, the potential for AI to elevate the user experience through on-device personalization is vast. The current generative AI application ecosystem has serious flaws that harm the user experience, frustrate AI developers and hamper sustainable growth.
Concerns abound that AI models such as ChatGPT and other LLM-based chatbots may not protect user data or prioritize data quality. There have also been issues with AI agents expressing political bias or contributing to disseminating outright false information. In addition, many AI companies are running out of data on which to train their large language models and AI algorithms, with 25% of data from the highest quality sources restricted between April 2023 and April 2024.
To date, developers have mostly been limited to the usage of models released by large, centralized organizations, which curtails the possibilities for innovation and inhibits developers’ ability to address issues around user data privacy. AI developers want to create better apps with more unique data and human knowledge that improves the quality and customization of agents, and users want more personalized experiences that protect their privacy and reward their contributions.
Decentralization is the answer
Centralized models made a mistake in leaving out the two key stakeholders: developers and users. Centralized entities created GPT-4 and the most currently available AI apps and agents behind closed doors. With no transparency or inclusion of developers and users in the building process, developers must trust the resulting models, and users must accept the resulting applications. There is another way.
Recent: Google unveils new quantum computing chip: Clock ticking for crypto encryption?
Developers and users are empowered with a voice and ownership by deploying a decentralized approach to app and agent creation. The foundational layer of app development should involve a decentralized network of GPUs, which ensures the development process remains open and transparent rather than hidden behind a centralized compute provider, who may also have undue influence.
An added benefit of decentralized computing is that it can be much more affordable than reliance on centralized providers, which are experiencing a massive bottleneck in demand due to the proliferation of AI applications and associated data processing needs.
Community members should have greater control over what data is leveraged by developers in the creation of applications supposedly tailored to their needs. The only way to incentivize community participation is through value-based rewards, which can be best achieved using the monetization capabilities enabled by blockchain technology. When people are economically rewarded for contributing their data, they are likely to do so.
As users become increasingly aware of online privacy risks, they demand safe and secure data usage. Centralized providers collecting sensitive information run the risk of storing such data in databases that become honeypots to would-be hackers. We’ve seen the cascading repercussions of database hacks like the 2017 Equifax hack, in which 148 million Americans were affected.
The consequences of a hack to an AI company database will likely be far worse and more far-reaching. Decentralized networks have no such central database to attract hackers, making this model much more secure.
When backed by secure technology and with proper incentives, developers can source high-quality, personalized data, which can then be deployed to create applications best suited to users. Most centralized models function by scrubbing the internet for information, but there is a limit to how customized an agent can become by that method.
If, instead, people securely share their personal health, finance, and other sensitive information via a decentralized system that simultaneously protects and rewards those data inputs, then the possibilities for tailored agents who can advise us based on our unique health, finance or educational profiles are near-infinite.
Looking ahead
Democratizing AI app development and creating communities where all stakeholders are valued and rewarded will foster AI’s sustainable growth. Using quality human knowledge and private data through decentralized networks will enable apps that dramatically enhance our productivity, communication and social engagement while protecting user data privacy.
From personal health and wellness apps that provide tailored health advice and nutrition recommendations to intelligent financial planners that can analyze our spending habits and establish financial goals or virtual stylists that suggest clothing and accessories based on our style preferences, we can only begin to envision the possibilities for agents built in a genuinely collaborative and decentralized way that prioritizes developers and end-users.
Salman Avestimehr is the co-founder and CEO of ChainOpera.
This article is for general information purposes and is not intended to be and should not be taken as legal or investment advice. The views, thoughts, and opinions expressed here are the author’s alone and do not necessarily reflect or represent the views and opinions of Cointelegraph.