Login

Sign up

OR

Can't sign in? Did you forget your password?

x

Login


If you don’t have an account, click right here


Can't Sign in? Did you forget your password?

x

Please fill in these details

to create your profile


Click on the image to upload your profile pic

x
x

At Awaremonk, I hereby pledge that

I will not troll or abuse any member of the community.
I will be a good listener.
I will check my facts before stating them and attribute them as well.

x

Forgot Password?

Enter your email and we shall mail you the password reset link

x

Reset Password

Enter new password and confirm password

x

Physics

A community for the scientific researchers, enthusiasts, explorers and thinkers

Vedansh Singh , highly opiniated Aug, 03 2016

Google's applications still take manual decisions.We are not ready for AI decisions


Google Now cannot scale to the whole internet. It can do a lot, and certainly enough to be a great product, but it can't do everything, not until we have 'true' HAL 9000-style AI, which is a long way off. Contrast this with Google Maps, which is also partly a manual product. Hundreds or thousands of people drive Streetview cars down roads and hundreds or thousands of people check the images and edit the maps. There's lots of machine vision too, as well (and crowd-sourcing), but the back-stop is lots of paid employees. The algorithms alone aren't good enough yet. Editing the internet by hand was just too big, but editing maps isn't - it's just very expensive. Google is willing to spend that (and Apple is going to try too).
Google Now cannot scale to the whole internet. It can do a lot, and certainly enough to be a great product, but it can't do everything, not until we have 'true' HAL 9000-style AI, which is a long way off. Contrast this with Google Maps, which is also partly a manual product. Hundreds or thousands of people drive Streetview cars down roads and hundreds or thousands of people check the images and edit the maps. There's lots of machine vision too, as well (and crowd-sourcing), but the back-stop is lots of paid employees. The algorithms alone aren't good enough yet. Editing the internet by hand was just too big, but editing maps isn't - it's just very expensive. Google is willing to spend that (and Apple is going to try too).

Akhil Singh , Co-Founder at Awaremonk.com Unlearning all th Aug, 03 2016


This reminds me of another discussion on Awaremonk where the discussion was about distinction of AI and machine learning. Current trend suggests majority of work in AI is a function of huge amount of data and smart ways to manipulate that data by machines.


Vaibhav , Aug, 03 2016


I think Apples strategy here is a little bit off, its not that algorithms are poor at selecting music that I would enjoy, its that they are poor at discovering new music that I would like to hear before the algorithm has a chance to interpret and iterate through multiple rounds of categorisation/quantification. There are multiple ways of fingerprinting/analysing/clustering the existing discography; at least the popular stuff, the problem is the newly released music. If Apple partnered with DJs that focused on their locality/region whilst give some air time to the up-and-comers more broadly it would tie Apples radio efforts much more closely with discovery, creating a more intimate experience for listeners and providing a new discovery avenue for artists (an aspect that is sorely lacking in spotify, and pretty much most streaming services). Local curation would also provide the initial fingerprint/signal into algorithmic curation. Such a model would add to the richness of the music sector, allowing local scenes to come through the radio/discovery channels; to swing back away from the "top 20 hits" model that some of these services seem to enforce.
I think Apples strategy here is a little bit off, its not that algorithms are poor at selecting music that I would enjoy, its that they are poor at discovering new music that I would like to hear before the algorithm has a chance to interpret and iterate through multiple rounds of categorisation/qua


Shanu , Aug, 03 2016


Because of the current limitations in AI and any other computer-based analytical methods, human curation is essential at some point to get a result that can be relied upon and acted upon. The only question is whether the user is expected to provide the curation or whether it should have already been done inside the service. In the case of Google Search, human curation happens when a user browses the search results. For Maps, human curation has to happen internally because users expect 100% accuracy and because it is the platform technology for automated driving directions. For advertising, human curation happens when a user shows interest and decides to click on it, which is a very rare event. For music streaming, human curation can happen by the use of the "skip" button, but in some situations, a more accurate playlist is preferred. For Google Now and other personal assistants, if they are to replace human secretaries, they will have to be pretty accurate and hence need a form of human curation. Another way of looking at is that a human created knowledge base allows others to build services upon that data, but not so much for an automatically built one. This is certainly the case with Maps as opposed to Search, and is very much the case in Bioinformatics.
Because of the current limitations in AI and any other computer-based analytical methods, human curation is essential at some point to get a result that can be relied upon and acted upon. The only question is whether the user is expected to provide the curation or whether it should have already been


Sanjay , Aug, 03 2016


With Google Search, nobody expects the results to be complete. Nobody expect to use the results as is. The end-user must make a final human judgement on which page they want to visit. If this is the goal of the service, then AI is sufficient. With Google Maps, the results must be complete. The suggested route must not lead you to the middle of a desert or into an airport runaway. The user is not expected to examine the route and manually confirm its accuracy. In these cases, manual curation is necessary.
With Google Search, nobody expects the results to be complete. Nobody expect to use the results as is. The end-user must make a final human judgement on which page they want to visit. If this is the goal of the service, then AI is sufficient. With Google Maps, the results must be complete. The sugge


Namita , Aug, 03 2016


The problem with curated text is that the text itself must change to suit the curation. Its not enough any more to curate a list of ten blue links or ten news stories and then let the reader sort through them to find out the information. The raw text itself has to be prepared and curated by humans with the purpose of curation in mind. Consider ten articles written about John Lennon and how much repetition each one would have. The article format itself is the problem curation on the web works better as a database, snippets that are curated first, and then joined together by data. Check Googles search pages they are a mishmash of links, news, videos, and shopping. They doesnt tell a story because *Google can only collect what is already out there*. Curation isnt some add on, but much deeper. the entire product, from first to last must tell a story
The problem with curated text is that the text itself must change to suit the curation. Its not enough any more to curate a list of ten blue links or ten news stories and then let the reader sort through them to find out the information. The raw text itself has to be prepared and curated by humans w