Stunning New AI Assistant From Google


Again, on other use cases, fully understood. But that’s the thing, it’s not that we must accept the technology into everything. We need to say that it some situations it’s great, in others less so. Otherwise we end up having one of those “everything made of bakelite” phases.

So, my verdict on using nlp and nlg for trivial things like hairdressers appointments: problematic parlour tricks. Love the tech, think it’s got a ton of value, but think this “my time is far too important to spend 90 seconds talking to this person” uncoverers some pointed questions.


Ah right, yh I get what you’re saying dude. We’ll just have to wait and see how this plays out I guess. Will be interesting either way that’s for sure.


Sorry for the short answer to your elaborate and well thought out expose. But I think it’s not about programming AI anymore, but about programming humans. And I mean that not in a scary way. It’s about user experience producing predictable outcomes, it’s a design thing, lowering the threshold to adoption by making it intuitive. In this case, reintroducing mannerisms that make it sound less human would not add perceived value in this particular use case. It would create a high threshold to adoption because people would think twice about letting an obvious artificial agent represent them in a conversation, however trivial.

Google is about making and keeping data flowing, which needs as many people as possible to be on board and use its services without friction or resentment. They have stayed the course on a mind bogglingly long term strategy, with some spectacular failures along the way (Google+, Duo, Allo), but this one might very well pay out for them.


MKBHD weighs in on Google Duplex. He features a short clip of Sophia on her date with Will Smith in the video. Lots of buzz forming around this AI breakthrough.



You definetly got it right. Ums and ahms are really necessary for a proper rapport. I think it might be possible in the future that there will be some conversational therapy like this for people which are in trouble.
But of course it would be necessary that the ai reveal their status to be congruent. Like it’s necessary in a conversational therapy like as Carl Roger’s stated. here I guess
I personally would rather talk to a machine which is unbiased, logic and of course functional. Then to a stressed stranger. With unlimited patience.