Ally Haire from Lilypad Shares Perspective on Significance of Open-Source AI and Importance of Accessing Compute Power

We recently connected with Ally Haire, CEO and Founder of Lilypad, which is described as a serverless, distributed computing network for AI, ML, and other computational processes.

Recently, it was reported that Elon Musk is going to build a “gigafactory of compute,” and while the name might need tweaking, the compute is needed.

The rapid development of AI over recent years, makes it easy to assume that the technology will continue on a fairly linear trajectory as it matures – perhaps similar to the evolution of the internet. However, there are currently forces at play that could seriously disrupt such a path, hampering the development of the technology or limiting access to it.

Perhaps the biggest and most imminent risk is that demand for computing power to feed AI algorithms could outstrip available supply – and soon. Our World in Data shows that across virtually all domains, compute requirements to feed AI algorithms are increasing exponentially.

According to the company, the Lilypad team is working to unleash idle processing power by leveraging decentralized infrastructure networks, such as Filecoin. This is expected to unlock a new marketplace that will make AI more accessible, efficient, and transparent for developers and users.

Ally, a former Lead Developer Advocate for the Hybrid Cloud team at IBM and Senior Software Engineer at the University of Sydney, discussed the capacity issues currently faced by major AI players and the need to move towards a decentralized future.

Our conversation with Ally Haire is shared below.


Crowdfund Insider: Open-source AI seems to be a buzzword of late, why is it so important?

Ally Haire: The rapid advancement of artificial intelligence (AI) is a defining moment in our era. We see how it is reshaping industries, transforming how we work, and altering our daily lives in ways we couldn’t have imagined just half a decade ago. As AI begins to influence so many industries and grows increasingly important, we have a collective responsibility to make sure it is not controlled by the few – that it remains accessible and available.

Many of the AI advancements we see in the mainstream today are developed, trained, and inferenced in closed cycles, with little to no transparency on process and data use. These closed systems can thrive because access to reliable and decentralised compute power for AI is rare – so only the largest organisations, such as the centralised AWS, Google, and Microsoft can facilitate the compute requests needed for large-scale AI inference. Decentralisation checks that AI development isn’t dictated by any one organisation or region and the open-source AI movement works to ensure AI development is verifiable, auditable, democratised, and equitable.

Crowdfund Insider: Why is compute power, and specifically access to compute power, vital for the future of technological development?

Ally Haire: The modern web stack is all about data – moving data around, storing data or making data useful – and there are three main components on which the modern web stack is built: networking, storage, and compute.

However, compute has so far been a missing piece of the decentralised infrastructure story. Traditionally, access to compute is granted by centralised entities that can control the cost and distribution, making the barriers to entry unassailable for many. There are large scale complexities involved in building a fully distributed compute network and also the fact that, within the compute space there are other problems seen as more pressing to solve i.e internet-scale storage or block scaling solutions. However, this is quickly changing with the explosion of AI and ML driving exponential demand for compute and compute resources.

This demand isn’t just focused on training large foundational models like ChatGPT either. The demand for AI compute also includes running inference models like stable diffusion text to image or LLM prompts or fine-tuning models via LoRA (low rank adaption) or other emerging algorithms. That’s why at Lilypad, we make AI and compute democratically accessible and available by opening access to compute by repurposing global idle GPU capacity for AI inference and other computational tasks.

Crowdfund Insider: What is Lilypad? Why does the world need distributed compute?

Ally Haire:

Lilypad is a distributed compute network harnessing idle processing power to fuel an open AI economy Click to Tweet
. Compute is being framed as a rarefied resource. For example, Jonathan Ross, CEO of Groq, argues that compute is the new oil. However, latest studies published by the American Association for the Advancement of Science state the computational power required for sustaining the AI rise doubles every 100 days. Current GPU capacity is being widely under-used all over the world.

Distributed computing utilises this available GPU and unites it to serve the needs of developers in the fields of AI, decentralised science (DeSci), gaming, DeFi, and more. Lilypad rewards those who contribute GPU to their network with Lilybit_, the IncentiveNet points program, and coordinates workloads to ensure GPUs are utilised to their full potential. Lilypad’s compute coordination platform becomes the backend for compute where users and providers can interact and feed into an open infrastructure stack to extend global access to efficient compute power.

Crowdfund Insider: How does the current market landscape for decentralised AI present opportunities for Lilypad?

Ally Haire: The current technological era is consumed with AI, making the market a very interesting place for Lilypad. As a whole, tokenisation works to empower the decentralised open infrastructure space. Tokenization allows assets to be divided into smaller units, enabling fractional, decentralised ownership. This lowers the entry barriers for investment, making it accessible to a broader audience and giving more power to crowdfunding mechanisms.

Traditional AI is very vertically integrated, however, projects like Lilypad help create a horizontally distributed AI stack with more value flowing to all contributors of the decentralised AI ecosystem. We give fairer distribution to power and in this case, compute is the power. And what’s more is, all parts of the data economy can be tokenised here as well. Research lakes are possible ie. sharing of research data without revealing the data to create models that can help address things like rare diseases.

Tokenisation also allows us to reward our IncentiveNet contributors for their efforts with Lilybit_, a points programme that will be redeemable for Lilypad (LP) Mainnet Tokens at launch that is expected later this year. IncentiveNet contributors can interact with the testnet through contributing GPU, testing network resilience, finding bugs, or validating operational game theory logic.

Crowdfund Insider: What do you see as the main barrier to widespread distributed compute adoption?

Ally Haire: Adoption of distributed compute is still in its very early stages, and with all early stage innovations, some kinks need to be ironed out such as the overall usability of distributed compute. Distributed compute systems are hard to comprehend and hard to debug. To see widespread uptake of distributed compute the more difficult parts of working with compute need to be abstracted away, to make for a more accessible entry to the space. Furthermore, to be successful in building networks large enough to compete with entrenched oligopolies in the compute space, we need to make distributed compute more accessible to the broader users.

Adoption of distributed compute is still in its very early stages, and with all early stage innovations, some kinks need to be ironed out Click to Tweet

Another issue in the space is ease of access to distributed compute, as so few reliable coordination platforms between users and suppliers exist. This limited access to decentralised compute also serves to benefit centralised compute suppliers – corralling the market to their products.

Ultimately, the distributed compute sector must also balance data encryption, access control, and privacy without compromising transparency or auditability, much like their centralised counterparts, if broader adoption is to take place.

Crowdfund Insider: What does the future have in store for Lilypad? Are there any major milestones or new features that you can share?

Ally Haire: Lilypad is currently rolling out the Lilypad Incentivised Testnet Program – IncentiveNet – which will significantly enhance the platform’s capabilities and user experience. This will pave the way for a broader, more diverse pool of developers to access compute and create new horizons for the future of AI, DeSci and more.

Going forward, we are continuing to onboard GPU providers and work towards Mainnet launch, where Lilypad will enable everyone to access high-performance computing through a serverless, distributed multi-chain network. If anyone would like to join the IncentiveNet more details can be found on our website. Follow along and join the Lilypad community via Discord or X.


Register Now
Sponsored Links by DQ Promote

 

 

Send this to a friend