Mozilla ran an interesting, meticulously documented, time-bounded effort to create an in-house, open source, large language model based chatbot.
With this goal in mind, a small team within Mozilla’s innovation group recently undertook a hackathon at our headquarters in San Francisco. Our objective: build a Mozilla internal chatbot prototype, one that’s…
- Completely self-contained, running entirely on Mozilla’s cloud infrastructure, without any dependence on third-party APIs or services.
- Built with free, open source large language models and tooling.
- Imbued with Mozilla’s beliefs, from trustworthy AI to the principles espoused by the Mozilla Manifesto.
Heads up. I’m on sabbatical, attempting to do a bit of education on deep learning, generative models, and AI broadly. Case studies like these are really useful.
Despite the hype train on generative AI, there’s promise, … and peril. Expect to see quite a bit more of this in my feed.