> Like their human counterparts, Aaru’s bots are fallible; for instance, they wrongly forecast a Kamala Harris victory in the 2024 election. Fink said the firm has significantly improved its models since then.

Anyone else do a mental double take after reading this bit? What other big things has it gotten wrong, then?

I don't know who to trust in this article. Aside from the reporter.

Person A with high praise of their startup, is an investor. Company B that would like to further use this service, is quoted though an executive as if written by an LLM.

Coca-cola is trialing the service but has no information to share. Alongside the other testimonials this doesn't say much about reality. (Does WSJ do advertorials?)

Also, good luck trying to predict real people.

What's funny, is seeing these young adult entepreneurs dress up like 40 year old tech bros. Will turtlenecks be the next style to circle back in the scene?

Using predictions of past events as the benchmark where the LLMs have already been trained on the results seems quite flawed to me

what the actual fuck is going on