• Tag Archives AI
  • AI writing product / place reviews

    My Dad and I had a bit of a back and forth email exchange about AI and Google filtering search results a week or so ago.
    For me, the end result was that it seemed like we both had valid points, but a lack of solid information about what really was going on made it hard for either of us to drive our respective points home.

    This article wont help.

    https://www.businessinsider.com.au/researchers-teach-ai-neural-network-write-fake-reviews-fake-news-2017-8

    For many people, online reviews are the first port of call when looking for a restaurant and hotel.

    As such, they have become the lifeblood for many businesses — a permanent record of the quality of their services and products. And these businesses are constantly on the watch for unfair or fake reviews, planted by disgruntled rivals or angry customers.

    But there will soon be a major new threat to the world of online reviews: Fake reviews written automatically by artificial intelligence (AI).

    Allowed to rise unchecked, they could irreparably tarnish the credibility of review sites — and the tech could have far broader (and more worrying) implications for society, trust, and fake news.

    Its a bit of a read, but the bottom line is that they have taught an AI how to write reviews that read like a human wrote them. Indeed, I failed the little ‘test’ at the bottom of the article.
    The point they make is scary. Reviews are one thing, but there is nothing stopping the AI from writing news. Or blogs. Or… yeah, you get the idea.

    One of the points I was making with my Dad is that while software engineers have written the AI, I am not so sure they are still in control of it.
    In this case, the AI reads reviews to learn how to write reviews. What happens when there are more AI reviews than real human reviews and so the AI is reading its own reviews? Who wrote the code for that? (And are they even still working for the company?).

    Don’t get me wrong, I am all for AI and ML, and while I rely (perhaps a little too much) on real people reviews of the products I buy, the idea of a computer writing a review on a product so it will rank higher in my search results leaves me a little cold.


  • Honey, lets put a camera in the bedroom

    All sorts of chorting, laughing, crying and pounding of a head on a desk with this one.

    Far out. I just…. Well, you know me, takes a bit, but honestly, I’m a little bit speechless…..

    https://www.engadget.com/2017/04/27/what-amazon-gets-out-of-putting-a-camera-in-your-closet/

    Years ago, if someone told you she was buying a camera for her bedroom, you’d imagine she was planning something naughty. But times have changed, and Amazon believes you’ll spend $200 on a camera that’s both an extension of your smart home and the ultimate fashion accessory. The Echo Look is designed to help you look your best every morning, guiding and improving your style every time it’s used. But what it represents, and what Amazon gets out of it, could be a much bigger deal for the future of fashion.

    Echo Look resembles pretty much every other smart home security camera you’d care to find, but it’s not about protecting your property. The unit is designed for fashionable types who like to document their daily outfits and make sure that they’re always looking good. Look comes with a depth-sensing camera, bolstered with four LEDs that turn even the dingiest bedroom into a half-decent photo studio.

    If you take a selfie in a full-length mirror before leaving the house each morning, then the Echo Look is its modern-day replacement. Rather than holding a phone across your body, you can simply speak and ask the device to take a picture for you. The unit is voice activated and comes with Amazon’s chatty computing platform, Alexa. In addition, the depth-sensing camera will automatically blur the background to hide how messy your place is.

    Chunk of quoted text yeah, but I warned you, I am speechless with this one.

    Clearly, I am not the target audience for this thing, but I am not sure that’s even my point….. You are putting a web enabled camera in your bedroom/bathroom/change room/wardrobe…… What were you thinking? What did you expect was going to happen?

    Ok as if we are over that fact and have moved on… Here is why you are going to do it…..

    The other half of Echo Look’s sales pitch is Style Check, a feature that builds on the company’s Outfit Compare platform to keep your “look on point using advanced machine learning and advice from fashion specialists.” Simply submit two snaps of you wearing different outfits and the system will tell you which one looks best. It does that by crunching what’s trending, what fashion experts are saying, how well it fits you and what colors are in season.

    AI. Yup. You are going to ask a computer how the outfit looks.
    Now, see, this oddly enough I am comfortable with… Blokes the world over are tired of the trap question ‘Honey, does this outfit make my butt look big?’ This clanger is right up with the whole, which of these 6 dresses should I wear out today?
    Clearly Amazon knows we blokes are tired of this game and have programed a computer to take our place. Sweet.

    But, every share comes at a price;

    Amazon is, of course, storing every single one of those images, along with hundreds of pieces of contextual information. That data will be constantly crunched not only to understand what style suits you but also what outfits you pick depending on weather, mood or season. In addition, the information will be used to train a machine learning system that can offer better suggestions for every Echo Look user.

    Every image is stored. Every. Image.
    Did that sink in?
    See, that’s the thing with machine learning, you need to feed it data, lots and lots of data…. Topic for another blog.

    I have already taken up enough of your time on this one, but the article goes on to talk about AR, where you can try on virtual outfits and see how they look in a monitor/mirror. Yeah, you guessed it, AR is another blog topic…..


  • AI vs missing kids

    Here is a pretty cool use of AI and facial recognition.

    https://www.engadget.com/2017/05/10/baidu-ai-facial-recognition-finds-abducted-son/

    Facial recognition tech has loads of application other than the creepy ones that put your privacy at risk. In China, for instance, it has helped a couple find their son 27 years after he was abducted. Fu Gui thought it was strange that he only had vague memories of his childhood, so he uploaded a photo taken when he was 10 years old to a website called Baobeihuijia, which translates to “Baby Come Home.” Little did he know that his parents would also upload his childhood photos on the website a few months later and that Baidu’s facial recognition tech would bring them back together.

    This is pretty cool. Sure, for now, both kid and parents have to upload a photo, and they have to be taken within 6 years of each other, but it’s early days. The more we feed it data, the more we train the AI model, the better it will become.

    Granted, there is always a flip side, this would be a massive backdoor for all sorts of tracking to take place…. But lets just pretend for a micro second that computers will ‘do no evil’….. (Let me know how that works out for you).


  • Moody Facebook

    File this one under, ‘this should not have surprised me, but sorta did’.
    Should have seen this coming a mile off, but wow, not only did it surprise me, but it really creeped me out.

    Let me say up front, I have never been much of a Facebook user. It has confused me from the day I joined up. Never really made peace with it.
    Been trying to post some photos to it the past 2 weeks. My first posts in years. Before this, I would check my feed probably once or twice a month.
    In other words, I hardly use it.

    So, why the creepy surprise?

    https://betanews.com/2017/05/02/facebook-emotions/

    A leaked internal document shows that Facebook is capable of identifying people according to their emotional state. The document, seen by The Australian, shows how the social network can monitor users’ posts and determine when they are feeling “stressed, defeated, overwhelmed, anxious, nervous, stupid, silly, useless, or a failure.”

    The leak pertains to Facebook’s Australian office and suggests that algorithms can be used to detect “moments when young people need a confidence boost.” It raises serious ethical questions about Facebook’s capabilities, but the company denies it is doing anything wrong.

    With all that I have blogged about AI and such, it really should not have come as a surprise to me that Facebook was machine analyzing every post by every user and coming up with a metric for each user. But it did.
    Its pretty sicking.

    The hint is that Facebook then uses this information to target advertising to its users.
    Ewww.

    I guess the short version is that we should not be surprised about what our online activities are getting used for these days.
    One that jumps to mind is gmail. They are up front in saying that they read every single email to deliver better ads for us. That’s the price you pay for free email. I’m now thinking that it would just be a slight of hand for Google to say, well, we have to read the email to figure out what ad to show, so may as well read his email and figure out what sort of mood he is in.
    What about WordPress blogs. There are millions of blogs hosted in their cloud, it would be trivial to feed an AI with the wordage from each blogger and come to all sorts of conclusions.
    Boggles the mind.

    In other related news. Just got off the call with the IBM Watson guys about getting our hardware to talk to their machine learning algorithms… Wonder if I can tweak it to figure out the mood of an air compressor or refrigeration unit?


  • Virtual you for eternity

    Every time I blog I wonder how much I should give my thoughts on the thing I am blogging about… My mate Matt (BA) tells me that I should speak my mind because it’s my blog and people read it to find out what I think about said blog… Others (nameless to protect them) have said, no, just tell me what’s new and tell me about stuff I would never find on my own….. Don’t add any commentary… So I duno.

    None of that really has anything to do with this blog, other than this topic really made me sit back and think hard about what I thought of this.

    Today in my RSS feed popped up a post about companies are springing up that are, or plan to, using AI to learn about you now, so that when you die, your AI can live on.

    http://www.liveson.org/connect.php

    Liveson focuses on Twitter. They look at what you have tweeted in the past, and track each tweet and like now while you are alive and builds a profile of you. When you stop tweeting, it takes over for you.
    Their motto? “When your heart stops beating, you’ll keep tweeting”.
    Slightly AIish.

    http://www.deadsoci.al/

    These guys are less about AI and just storing messages for later release on Twitter and Facebook. A kind of time vault. There are many of these services around since they are nothing more than a reminder (usually via email) every week or so, if you don’t reply, it unlocks your vault of social stuff and that is that.

    http://eterni.me/

    This is the big one.
    These guys are the one that want to keep an AI version of you around for a long time. They want to read your past Tweets, Facebook, Instagram, Emails and everything you let them have. On top of that, you interact with the AI (your AI?) for around 10 minutes a day (not at once, spread out) and it learns you. All it can, about you.
    When you’re gone. You’re not really offline for very long before your AI kicks in and keeps the ball rolling. Your ball.

    Just try and let that sink in. An AI that Tweets and posts on Facebook and Instagram just like you would……. WHOA.