As j-james and wrkyle have said, the key issue here is sources. A reference work contains sources. It's not about annoying UI (easily fixed), or even accuracy. A reference work can point to inaccurate information, as long as there is a source so we know who is making the inaccurate claims, we can evaluate their reputation, and so on.
Is Kagi aspiring to be a reputable reference work, an index to the web? In that case, all information it presents must be sourced. This is non-negotiable, and any journalist, librarian, or even non-fiction author is very familiar with these practices.
Or is Kagi apiring to be an source of information itself, to answer questions directly? (e.g. "Who was president of the U.S. when superconductivity was discovered?") This would be a very different product, and the quality of such a product depends entirely on the trustworthiness of the answers. (Which is currently nil. You cannot trust a source that is only sometimes accurate.)
A product that "searches" and gives references is operating in a completely different realm than a product than gives its own opinions. This is a very old distinction that, again, every journalist is very aware of. This is much more fundamental than UI, or "a feature that users want". It's about which code of ethics you'll be held to.
My opinion is that Kagi is already an excellent "search" engine, and when it tries to operate in two conflicting modes, it makes the whole product untrustworthy. You can no longer trust what you see on Kagi.
