Conference Presentation, Fireside Chat, Panel
A Fireside Chat with Himanshu Tyagi, CEO & Co-Founder of Sentient Foundation ft. FINOS
- Community contributions to open-source AI are expected to persist for eight to nine years while shifting from generating "bragging rights" to enabling contributors to monetize ownership as assets appreciate.
- Future AI models are predicted to evolve beyond expressing opinions to actively governing and making investment decisions on behalf of users.
- The open-source ecosystem is anticipated to generate a multi-billion dollar asset that distributes value back to its contributors as the movement transitions from an activist cause to a creator economy comparable to YouTube or TikTok.
- Open-source models will likely be required to declare their constitutional alignment, with communities utilizing open red-teaming to audit these models for deviations.
- New regulations, including the Cyber Resilience Act and EU AI Act, are expected to impact the open-source sector, potentially driving tariffs and national boundaries despite the increasing international nature of the next generation.
- Sovereign models may emerge in countries beyond China, the US, and France that fork from existing baselines and close their models, though a baseline of open source must be maintained to ensure continued advancement.
- Historical patterns observed with OpenAI, which closed models to generate revenue, may repeat as entities seek to monetize their work.
- The next generation of developers, described as 10- and 15-year-olds, is expected to own their contributions and demand direct payment rather than relying on non-monetary recognition.
- Future incentive structures may include barter systems where trade secrets like hyperparameters are tokenized to prove value, and data contributors could receive revenue when high-value data, such as medical information, increases the value of high-value tokens.
- Monetary incentives could potentially impede the growth of certain networks, referencing the argument that Wikipedia could not have succeeded if contributors were paid.
- Coordination across decentralized compute, where participants invest $10,000 each, may create ambiguity regarding liability and enforcement for resulting AI models.
- There is an expectation that regulatory frameworks will attempt to reach these technological evolutions, with a hope that such regulation does not halt progress.