Sierra co-founder Clay Bavor on Making Customer-Facing AI Agents Delightful

[@jamesdetweiler]

This video is a rare, crystal clear glimpse into the future

[@NandoPr1m3]

Fantastic Interview! It's very tough to talk about AI in general because it's still moving quite fast. I equate it to trying to fetch water from a river using a strainer. I'm grateful for the caliber of people like Clay working on turning it into real world solutions.

[@chanrox69]

Awesome interview. Working on AI ethics and fairness. There are really good open source frameworks

[@sirishkumar-m5z]

The Strawberry project by OpenAI is fascinating! It's exciting and possibly transformative for education that a single AI model can cover such a wide range of topics.

[@DylanKane-u2j]

Kinda took a dark turn when he basically admitted to misleading or hiding information from customers, whereas this tech could be used to educate, cure and advance society.

[@foregroundtreble05]

0:15 -- useful insight for the most common user

[@darkmatter9583]

Given the rapid evolution of AI in the past 18 months, how do you envision its impact on hardware optimization for large scale deployment in the next decade?

[@420_gunna]

0:15 "They're better at detecting errors in their own output than not making those errors in the first place." Haven't listened yet but there's a GDM paper "LLMs cannot self-correct reasoning yet" that refutes this, assuming he's talking about intrinsic self-correction without shitty pre-hoc prompting.

[@railtorail]

Looks like he's a good fund raiser given his story telling abilities.

[@sergiopassos5579]

C 3:42

[@vicaya]

Pretty much admitting that the current AI solutions fundamentally don't scale, just like all previous enterprise software that needs heavy post sale customization for many customers, albeit with even less certainty.

[@FriscoFatseas]

Delightful Good, gimmie the juice

[@dillanhoyos9181]

Meh