Fab @GaryMarcus @WIRED piece on future of #AI & 3 big questions for @GoogleAI’s DeepMind:
Is DeepMind on the right track scientifically?
Are investments of this magnitude sound from Alphabet’s perspective?
And how will the losses affect AI in general?https://www.wired.com/story/deepminds-losses-future-artificial-intelligence …
@IntuitMachine @earnmyturns @ESYudkowsky I felt like you got me all wrong last week; here’s what I actually think, couldn’t fit it all in a retweet.
-
-
Perhaps you should have including other research groups like OpenAI, FaceBook, Microsoft, Tesla, Baidu, Nvidia that are also investing a lot of money in Deep Learning based technology. DeepMind isn't the only one.
-
agreed, but an 1100 word essay has to be scoped...
End of conversation
New conversation -
-
-
Genuine question: should the use of DRL for reducing Google's datacenter costs count against the losses (https://deepmind.com/blog/article/deepmind-ai-reduces-google-data-centre-cooling-bill-40 …)? I don't know what % of DC costs are cooling but apparently data center costs are a big chunk of Google's 25$ billion capital expenditures for 2018.
-
the losses are profits minus revenue, which includes the cooling revenue. unclear whether DRL is actually involved. i reported the revenue #, per Alphabet filing that i linked
- 1 more reply
New conversation -
-
-
How do children acquire language and come to understand the world, using less power and data than current AI systems do? Very important question. I see that a child learns by repetition and continuous observation. Amount of time a kid spends is not less by any means.
-
Worth noting that kids don't really learn from unstructured data in 2 years, they have at least 100k years of co-evolution with culture, and selection on culture and language tailored to concepts and culture that allows human mental structure to learn.
- 2 more replies
New conversation -
-
-
A boy takes 2 years to start talking. This along with the info and knowledge he received in his genes. I guess we may be under estimating the evolution here.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Spinning research as """losses""" is disingenuous. Yes, the article walks this back a bit but the title is clickbait. & the numbers are big because of big salaries, which is because the rest of the economy believes that such people are very valuable/good investment
- 1 more reply
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.