Conversation

From 2015, but if you were to read one piece on AI... this is the one. Stands the test of time surprisingly well! waitbutwhy.com/2015/01/artifi

131

hardmaru

@hardmaru

Aug 29, 2017

Well it's almost 3 years since this article was published. And I have yet to feel any hockey stick "exponential" progress as forecasted! :-O

Shivon Zilis

@shivon

Aug 29, 2017

Do you think that is true for both applied AI and trajectory of research? I posted this to see reactions on both sides so would love to hear

agibsonccc

@agibsonccc

Aug 29, 2017

Don't even get me started on markets outside the US. Let alone on enterprise procurement of non stock dell cpu boxes. (gpus etc)

Replying to

and

Oh! What are you seeing here?

5:36 AM · Aug 29, 2017Twitter for iPhone

Replying to

and

Our 2 biggest markets outside the US are japan(work with softbank and other cos) and china (large scale healthcare). EU is sparse yet.

Replying to

and

EU is still heavily regulated and only getting worse. Hard to get work done. See deepmind and the friction they've had in healthcare.

Replying to

and

Our work in telco in europe, we have also seen conservative data access and budgets.

Replying to

and

In china, the market is highly fragmented. There are different pockets of talent with more applied research.

Replying to

and

Large scale research largely requires a tech company or govt backing (we have the latter)

Replying to

and

In japan, the companies are behind with few companies building their talent locally.

Replying to

and

Companies that need AI mostly have managers. No builders. Training and education in japanese are needed.

Replying to

and

China is owning applications research with top down access to resources but it's still reserved for few individuals.

Replying to

and

Japan and even korea where we have some presence has a lot of hobbyists but not a lot of people with concerted effort.

Replying to

and

Preferred Networks (who we are ironically just downstairs from) is a notable exception in a see of marketing.