RE: AI learns to play Flappy Bird - the impact of machine learning on Steem Blockchain by cmplxty

steemstem · @cmplxty · Mar 9 '19

$0.03

Machine learning is an interesting prospect in some manners but in others it’s disconcerting. In the medical field, they are trying to assist radiologists in their diagnosing of cancers. 
I’m not sure that I would be looking forward to having an AI look at posts on the trending page to be any kind of representation of the content of the Steem chain. There are too many junk posts that make it to the trending page sadly. It would be hard to tell it what kind of content to emulate as well because all our opinions of what constitutes a quality post vary. Interesting work on the game though! Those things are certainly easy for the software but humans excel at other things that the software does not and vice versa. 

Posted using [Partiko iOS](https://steemit.com/@partiko-ios)

👍 cmplxty, neavvy

`author`	cmplxty
`permlink`	cmplxty-re-neavvy-ai-learns-to-play-flappy-bird-the-impact-of-machine-learning-on-steem-blockchain-20190309t223907430z
`category`	steemstem
`json_metadata`	{"app":"partiko","client":"ios"}
`created`	2019-03-09 22:39:09
`last_update`	2019-03-09 22:39:09
`depth`	1
`children`	1
`last_payout`	2019-03-16 22:39:09
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	0.022 HBD
`curator_payout_value`	0.007 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	781
`author_reputation`	790,167,012,563,000
`root_title`	"AI learns to play Flappy Bird - the impact of machine learning on Steem Blockchain"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	80,984,373
`net_rshares`	42,528,069,144
`author_curate_reward`	""

properties (23)vote details (2)

Thank you for your reply @cmplxty > In the medical field, they are trying to assist radiologists in their diagnosing of cancers. That would be really awesome if they succeeded. > There are too many junk posts that make it to the trending page sadly. It would be hard to tell it what kind of content to emulate as well because all our opinions of what constitutes a quality post vary. I agree with that, but I think there are already solutions to these problems. There is already a platform that learns about your content preferences (basing on content that you read most often) and displays only posts that may interest you. There also algorithms that detect which posts were upvoted by bidbots, so they are able to detect posts that are rewarded with highest rewards from the community. > Those things are certainly easy for the software but humans excel at other things that the software does not and vice versa. That's the point :)

voter	weight	wgt%	rshares	pct	time
neavvy	0 B		10,744,382,922	35%
cmplxty	0 B		31,783,686,222	100%