At the UN Open Source Conference, I joined a discussion on how to help beginners start contributing to open source.

Midway through, I asked "Will AI lower barriers to OSS contributions, and how are projects handling these influxes?"

The responses were unexpectedly blunt:

cultivating sovereignty and stewardship // one must imagine Sisyphus happy // prev consensys / chainbound / NPR // https://fevenm.substack.com

so their current stop gap has been:

explicit policies in their code of conduct or contribution guidelines that AI generated PRs are given low to zero acceptance rates

The maintainers I talked to were pretty clear about this:

"AI will change everything, but we're not close to ready."

They lack the processes and mechanisms to verify that AI generated code isn't:
→ Unknowingly malicious
→ Full of security vulnerabilities
→ A literal ticking time bomb

On the surface, the promise of AI for open source is exciting: 

AI may lower the barrier for contributions from beginners and non-technical folks who were previously blocked by high technical barriers.

But there’s a catch...