why aren't all 15 resolved? a statement only gets an assessment when the public
record can support or contradict it. opinions and what-ifs never can, and 0 checkable
ones are still open, waiting for their date. predictions held up or didn't;
assertions are supported or contradicted. on every card:
▮▮▮▮▮ certainty ·
▮▮▮▮▮ debate potential. speakers are clickable
Insight
Intelligent AI agents should never pretend to be human
“Now, number one, do not pretend to be a human.”
Insight
AI agent developers should over-invest in reasoning engines over parsing
“And I think there's probably even more value in whatever agent that you go engineer to over invest in the reasoning engine.”
Disclosure
Mortensen: x.ai has no backup plan to rely on human labor
“What we kind of set out to do is to kind of create a setting for where there is no plan B. There is no kind of backup plan, and I know SoftBank kind of, you know, shit their pants whenever I say that, but there is no plan B of, ah, perhaps we can just have hum…”
Opinion
The tech industry fell in love with conversational UI too early
“I certainly think we just fell in love with the conversation of UI a little bit too hard, too early, because it certainly was fucking sexy, to be honest.”
Insight
Giving Google Analytics logins to non-analysts is naive
“Editors should be editors, not analysts. I think that would work in pretty much any industry, because the idea of handing out random logins to Google Analytics and believe that your organization is wiser seems Naive tomorrow, perhaps slightly smart today, but …”
Insight
Recruiting 30 engineers for a meeting-scheduling AI is extremely difficult
“You try to hire 30 engineers on the idea that they should schedule more meetings? That is fucking hard.”
Opinion
Voice commands are an inferior interface for adjusting TV volume
“Screaming to your telly that you wanted to kind of increase the volume is actually not a good UI. My remote is just way more powerful.”
Insight
AI product designers must ensure low-accuracy predictions still provide value
“You need to design for a system where if you make a prediction, even if it is of low accuracy, you should make sure that it counts as some sort of value.”
Disclosure
x.ai priced its agent under $200 to avoid human-salary expectations
“We priced it at 19, 39, 199, some price for where it doesn't even resemble a human being at, you know, 50 K cost in Manhattan.”
Disclosure
Mortensen: x.ai features no user interface beyond email CC
“So there is no interface. There is no dropdown box. There is no set of possible times which you can click on. There's only Amy, which exists in dialogue as a CC.”
Assertion Not checkable as stated
Half of newspaper article views in 2012 originated from homepages
“About half of all the article views on any given day for any one of these properties were driven off the homepage or the section front page.”
Disclosure
x.ai shifted meeting preference setup from conversational text to web UI
“We initially implemented it against the idea that you would set your preferences over the conversational UI. Like, ah, I really like meetings on Monday. I come in a little bit late. I need to leave early on Fridays. Got to pick up the kids. And a whole long id…”
Disclosure
Mortensen: x.ai is built solely to schedule meetings
“We've set out to do nothing more than schedule meetings. No more, no less.”
Assertion Not checkable as stated
Dennis Mortensen had 1,019 meetings and 672 reschedules in 2012
“I did a 1019 meetings. And I had 672 reschedules.”
Assertion Not checkable as stated
Mortensen: 87 million US knowledge workers schedule 10 billion meetings annually
“So, there's eighty-seven million US knowledge workers today scheduling about ten billion something formal meetings every year, which is a problem you just can't solve with humans.”