Jump to content
House Price Crash Forum
Sign in to follow this  
wonderpup

Google = Skynet?

Recommended Posts

I came across this interesting take on what Google is really trying to create;

“At first glance, you might think that Google is beefing up its AI portfolio to improve its search capabilities, since search contributes 80 percent of its revenue. But I think that’s backward.

Rather than use AI to make its search better, Google is using search to make its AI better. Every time you type a query, click on a search-generated link, or create a link on the web, you are training the Google AI.

When you type “Easter Bunny” into the image search bar and then click on the most Easter Bunny-looking image, you are teaching the AI what an Easter bunny looks like. Each of the 12.1 billion queries that Google’s 1.2 billion searchers conduct each day tutor the deep-learning AI over and over again.

With another 10 years of steady improvements to its AI algorithms, plus a thousand-fold more data and 100 times more computing resources, Google will have an unrivalled AI. My prediction: By 2024, Google’s main product will not be search but AI.”

http://www.dailyreckoning.co.uk/penny-shares/private-companies-are-building-an-artificial-intelligence/

So in effect every time you perform a Google search you are acting as a tutor for their AI system. That means that at least some of us are already training the system than will eventually take our jobs- and not even being paid for the service. :lol:

Share this post


Link to post
Share on other sites

I came across this interesting take on what Google is really trying to create;

http://www.dailyreckoning.co.uk/penny-shares/private-companies-are-building-an-artificial-intelligence/

So in effect every time you perform a Google search you are acting as a tutor for their AI system. That means that at least some of us are already training the system than will eventually take our jobs- and not even being paid for the service. :lol:

https://en.wikipedia.org/wiki/Technological_singularity

Share this post


Link to post
Share on other sites

You are getting over excited. There are two main reasons for the recent surge in interest in AI.

1. A tweak to the neural net methodology (that had been failing for decades to be useful) has finally generated some useable results. The tweak in question has to do with the methodology for training the neural networks, which even with the recent improvements, typically takes weeks or months using many, many CPUs and kilowatts in a data centre. The keyword you need to google (no pun intended) is 'deep learning'. Google and all the others who are jumping on this bandwagon have been very public about the method. There are lots of useful videos on youtube. Please do educate yourself about it.

2. The advance mentioned in (1) has happened to co-incide exactly with the growth in cloud computing, which means that cloud operators have access to a vast set of training data. The algorithm advances are of no use to those without access to sets of labelled training data.

Before you get too luddite, bear in mind that the energy cost of training the networks and then actually using them to perform *inference* (in which an image or spoken sentence is classified) costs orders of magnitudes more energy than what the human brain uses to do the same job. And there are no obvious computational strategies using conventional semiconductor processes that will address the power consumption problem. Quantum computing *may* address the power issue but it still remains confined to a very few universities with no obvious commercialisation in sight.

All that said, its a great time to be working in these sectors, but that does not mean its a clear and present threat to man-on-the-street in the near future. It *does* present a real threat to professionals whose job is pattern recognition and prediction - analysts, traders, advertisers and so on. But these are not man-in-the-street jobs.

Share this post


Link to post
Share on other sites

quote taken from here http://www.telegraph.co.uk/news/uknews/crime/11964655/Police-to-be-granted-powers-to-view-your-internet-history.html

"Police are to get the power to view the web browsing history of everyone in the country.

Home Secretary Theresa May will announce the plans when she introduces the Government's new surveillance bill in the House of Commons on Wednesday."

I was searching for a new speedometer for a land rover defender earlier and the search results had a line at the bottom stating that all child p0rn results had been excluded. If AI thinks that PRC7373 is dubious, should I be worried or paranoid or just amused.

Share this post


Link to post
Share on other sites

quote taken from here http://www.telegraph.co.uk/news/uknews/crime/11964655/Police-to-be-granted-powers-to-view-your-internet-history.html

"Police are to get the power to view the web browsing history of everyone in the country.

Home Secretary Theresa May will announce the plans when she introduces the Government's new surveillance bill in the House of Commons on Wednesday."

Trivially circumventable by anyone with even the barest technical nous - just connect to a any of the numerous anonymising VPN services.

Share this post


Link to post
Share on other sites

I have recently turned to using bing. Its been at adily improved over the year s and IMO now better than google in many ways.

I liked google again for a bit when they had their blogsearch feature. That obviously gave too many resukt that were on topic and not ad related so it was switched off. When really only the first two pages are of any importance I can see at some point other search engines will enter the scene and bring some sense back to the idea of search. Maybe distributed / open source.

Share this post


Link to post
Share on other sites

I was searching for a new speedometer for a land rover defender earlier and the search results had a line at the bottom stating that all child p0rn results had been excluded. If AI thinks that PRC7373 is dubious, should I be worried or paranoid or just amused.

Maybe that's why?

Share this post


Link to post
Share on other sites
Before you get too luddite, bear in mind that the energy cost of training the networks and then actually using them to perform *inference* (in which an image or spoken sentence is classified) costs orders of magnitudes more energy than what the human brain uses to do the same job.

Unlike a human brain however once you have trained your AI you can propagate that capability far more easily. A single AI network could- via the internet- replace a lot of individual human brains whose advantages in energy costs are far out- weighed by their need for such trivia as salaries, healthcare benefits, holiday pay ect.

Most employers are not going to sit down and do a cost benefit analysis based on the relative energy consumption of organic verses artificial brains- if Google can offer them brainpower on tap that works out cheaper in financial terms than a roomful of humans with all their messy organic collateral costs the temptation will be strong to replace those people with AI.

All that said, its a great time to be working in these sectors, but that does not mean its a clear and present threat to man-on-the-street in the near future. It *does* present a real threat to professionals whose job is pattern recognition and prediction - analysts, traders, advertisers and so on. But these are not man-in-the-street jobs.

I'm not sure you are right about the 'man in the street' either. Anyone whose work is mainly sitting in front of a screen manipulating data and drawing inferences from data is going to be vulnerable to a cloud based AI solution of one sort or another. Unless by 'the man in the street' you meant the guy who walks around with a pointy stick picking up litter and so forth- people doing this kind of manual work will probably be a lot more safe I agree- for a while anyway.

Edited by wonderpup

Share this post


Link to post
Share on other sites

It is debatable whether this is really artificial intelligence or just machine learning. If you look at the changes they have made to include deep learning into the search process it is clear that there is no 'secret source'. The process takes highly skilled scientists and a lot of fine turning to get the algorithms to learn well.

What will be interesting is when the combination of deep learning (or wide learning) combined with reinforcement learning brings in the years to come. It is the reinforcement learning element that will provide the 'magic'. The half billion dollar purchase of the deepmind experts and the amount of data google has clearly will give them a head start on this. However with the internet everyone has the ability to get gather enough data to build AI systems. I imagine that the real breakthroughs will come from individuals working alone or in small groups without the constraints of a big PLC like google. They have snapped up all the key players in deep learning but the tech is available for all to see as the likes of Geoff Hinton are academics by trade and have published many papers.

Share this post


Link to post
Share on other sites

It is debatable whether this is really artificial intelligence or just machine learning.

Quite - machine learning (which is what we are actually talking about) is not 'artificial intelligence' as it is bandied about in the press. Its very useful, but "intelligent" is probably not the right term to use unless you are in marketing.

Share this post


Link to post
Share on other sites

It has always been googles intention to build the ai. A mate of mine who was there in early days and used to spend time with Larry and Sergei in work and social situations used to tell of how they would refer to the machine. When it comes around 2045 this is where it come from.

Share this post


Link to post
Share on other sites
Quite - machine learning (which is what we are actually talking about) is not 'artificial intelligence' as it is bandied about in the press. Its very useful, but "intelligent" is probably not the right term to use unless you are in marketing.

It's a moving target. Before a computer beat a chess master a chess playing computer would have been 'intelligent' But once it happened the definition of 'intelligent' was revised to exclude chess playing computers.

The same thing is happening with self driving cars- before Google's cars appeared this was a task that required 'intelligence' but as things progress the meme is building that driving is not really something that requires true intelligence after all.

What we seem to have here is an axiom that is self validating-" if a computer can do it then it does not require true intelligence- therefore anything a computer can do is not truly intelligent."

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

  • Recently Browsing   0 members

    No registered users viewing this page.

  • The Prime Minister stated that there were three Brexit options available to the UK:   59 members have voted

    1. 1. Which of the Prime Minister's options would you choose?


      • Leave with the negotiated deal
      • Remain
      • Leave with no deal

    Please sign in or register to vote in this poll. View topic


×

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.