Recently we highlighted the ways in which Livepeer has evolved and grown over the past several years, and how this growth positions us for future opportunities. Today, we want to dive further into certain aspects of the network’s design and evolution that are especially exciting for the future.

Since its inception, Livepeer has been focused on building out decentralized infrastructure to support the future of streaming video, and we have made enormous progress on this front. And these technical foundations open up further exciting possibilities that intersect with some of the most important developments in video as well as in the technology space more broadly.

Growing the Role of Orchestrators

Orchestrators are key to the design and functioning of the Livepeer ecosystem. Currently, these node operators are heavily focused on a single function: transcoding, which is a crucial function for all streaming video services and applications.

Broadly speaking, a continued focus on streaming video is the right posture for the network in the near term. With that in mind, there are two areas of Livepeer’s development roadmap where we intend to focus efforts to expand its capabilities.

The first of these is Job Type Abstraction.

In simple terms, this means broadening the type of work Orchestrators can perform. Currently, the marketplace operates on an assumption that all work performed, and therefore all payments exchanged, relate to transcoding. But we are now at a stage where Livepeer should begin to evolve to a stage where multiple job types are advertised – for example, to automatically generate closed captions for videos. This type of additional capability furthers the power and attractiveness of decentralized streaming, while at the same time addressing use cases whose salience will continue to grow with increasing use of Artificial Intelligence (AI) by video creators.

As a decentralized network, Livepeer looks to its growing community to help push its goals forward. With that in mind, there will be an open contributor opportunity through grants for a small team or individual to lead work on Job Type Abstraction. Please join the #developer-lounge channel to introduce yourself and discuss if you’re interested.

The second area of near-term focus is Security & Verification Research. The objective here is to develop processes for users to verify that the different job types performed by the network were done correctly, particularly in the non-deterministic GPU environments that will increasingly become the norm as AI becomes both expected and embedded in basic video production.

Research into this concept has been ongoing since the inception of the project, and the team as well as members of the community continue to work to advance Livepeer’s capabilities in this area. As this progress continues, the definition of “verification” itself is likely to expand to encompass concepts like slashable stake deposits and reputation, as well as governance-based mechanisms such as judiciary panels that play a role in certain disputes.

We encourage members of the community to form working groups and collaborate with core contributors to expand the network’s understanding of security and verification during this time of rapid technological progress.

AI and Decentralized Video

These initiatives take place against the backdrop of the rapid growth of AI. This is an area where Livepeer has an opportunity, over the medium to long term, to really establish itself as a linchpin of a future in which decentralized video interacts with AI models used by creators. Specifically, Livepeer’s decentralized infrastructure has the potential to support a key aspect of the industry’s growth that has received relatively less attention amid debates about regulation and corporate concentration.

At their most fundamental, AI models require massive amounts of computing power to be applied on videos in real-time, and to process large numbers of actions very quickly. The only chips with enough power are called Graphics Processing Units (GPUs), which are currently in short supply.

In light of this situation, researchers and entrepreneurs are forced in many cases to turn to large cloud computing services such as Amazon Web Services and Google Cloud. This concentration of control over the world’s compute power and the hefty fees the incumbents charge replicate the centralization and concentration that has increasingly characterized the legacy internet.

But just as decentralized infrastructure can help democratize streaming video, the same can be true in many ways of AI compute power. In fact, in R&D, Livepeer software has already been used for AI-based video compute, including scene classification, content type detection, closed caption generation, and object detection in video.

The explosion of demand for GPU based AI compute therefore creates an opportunity for Livepeer. The project’s software is already used in many elements of the video pipeline, from ingest, to origin services, to content delivery networks, and in interfacing with storage providers and indexing services. Now, the recent advances in AI and the associated compute requirements mean that Livepeer can become the leading video compute network for specific, important subcategories such as generative video, deep fake detection, and realistic avatar generation.

This would represent the fulfillment of a long-held ambition for the network. Moreover, it would align with Livepeer’s core values. Using decentralized infrastructure to address some of the key challenges surrounding the future of AI video, and its implications for creators and creative content, is exactly the kind of challenge the technology has been developed to solve for.

The Cutting Edge of Video

Livepeer is designed to facilitate an unlimited number of processes associated with streaming video – and more. As the growth of AI continues to reshape all areas of the modern world – not least streaming video – Livepeer is positioned to be the backbone of a powerful, decentralized, AI-enabled streaming ecosystem.

Join us in our community Discord today to get involved and start contributing to the future of streaming video.