Perspectives
Why We're Being Upfront About What We Haven't Proven Yet
It would be easy to publish confident numbers before we have them. Here's why we're choosing not to, and what that means for how this newsroom works.
AI companies publish a lot of confident-sounding numbers: accuracy percentages, user counts, time saved. Some of those numbers are real. A lot of them are early, rounded, or quietly optimistic in ways that are hard for an outside reader to check. We would rather be the kind of company that is slower to make a claim, and means it when we do.
What this looks like in practice
Why this matters more for us than most companies
A company that says it's building AI to understand Rwanda accurately has a particular obligation to be accurate about itself. If we are not careful about what we claim, there is no reason anyone should trust us to be careful about what our systems claim either.
If you ever notice us overstating something, we'd genuinely like to hear about it. Reach out through our contact page.
Elyssa Irankunda
Founder & Chief Technology Officer
Leads AI research and product development at ATAS, with a focus on Kinyarwanda NLP and low-resource language modeling.
research@atas.rwRelated from the newsroom
Discussion (0)
No comments yet
Be the first to share your thoughts on this article.