There’s been an awful lot of hype about AI in 2017. Whilst some of that hype may be blowing things out of proportion a bit (the singularity certainly has not arrived yet!) there are certainly some predictions for AI in 2017 that we are rooting for.
4 Ways AI In 2017 Will Change The World
2016 was hailed as the year of VR, and it turned out to be a bit of a damp squib compared to the rise of AR. Pokemon Go and the continuing rise of Snapchat stole the stage in the consumer arena, whilst sales of VR hardware struggled to meet their projections.
Whilst both VR and AR continue to grow and develop, another tech player is the belle of the ball for 2017. AI is developing at an awe-inspiring, and sometimes terrifying, rate.
There is some seriously jaw-dropping stuff going on in AI right now, so we’re pretty confident that it could live up to its pre-determined 2017 status. But TDMB, you ask, what is this jaw-dropping AI stuff you’re talking about? Well, reader, let’s take a look…
AI in 2017: #1. Generative Adversarial Networks
Let’s just call them GANs (because that’s what everyone else calls them, plus it’s shorter). GANs are systems comprised of one network that generates new data by learning from a training set, and a second network that tries to work out which data is real, and which is fake. A bit like deciding whether or not to believe the articles on your Facebook news feed… except not. Between the two networks, they are able to produce very realistic synthetic data.
The applications of this tech are pretty awesome. GANs can be used to generate scenery for video games, apply design changes to CGI designs, and – you’ll like this – de-blur pixelated video or photographic footage. That will come in handy for making sense of that video you and your mates made on the late train home last Friday night.
Joking aside, however, GANs represent a way for computers to learn from unlabelled data, which may be the key to further improving their intelligence in the future.
If you fancy geeking out on GANs then this is a proper geeky GANs video:
AI in 2017: #2. Language
Trying to have an in-depth conversation with Siri or Alexa can be more frustrating than reading Donald Trump’s Twitter feed, albeit without the unhinged tyrant aspect. But imagine a time when you can chat away to your personal AI assistant and she can chat back in a way that feels natural. Undoubtedly, a better mutual understanding between woman and machine would make our AI pals a lot more useful. But the task is not an easy one.
Language is subtle, complex, and infused with strange and puzzling idiosyncrasies. Look at how we employ metaphor, for example, often without even noticing we’re doing it. So if you’re an AI, you’re going to struggle. Hey, it even takes children a good few years, and their brains are pretty sophisticated.
The development of Deep Learning has made a substantial impact on progressing advancements in how computers learn to perform complex tasks. AlphaGo, for example, famously slaughtered Lee Sedol, the world’s number one player of the world’s number one most difficult game using its Google Deep Mind neural networks.
However, deep learning has not, as yet, made much leeway in the realm of language. China is probably the world leader in this right now (as they are on loads of aspects of tech), with Baidu’s AI lab already making fantastic progress in voice recognition and language processing. We could learn a lot from China, actually. They’ve stopped trying to copy the West, and using their own highly skilled talent to make in-roads to areas in which we are lagging behind.
When they do crack the language thing, whether in China or Silicon Valley, artificial intelligence capability will skyrocket at a pace hitherto unseen. Whilst we’re not sure this breakthrough will come to pass in 2017, we certainly hope it does, because it’ll be amazing.
AI in 2017: #3. A Resurgence for Microsoft?
Microsoft may still dominate the PC market, but it has struggled in recent years to match the glamour and corresponding sales figures of its main rival, Apple. Nonetheless, it’s not out of the race by any means.
Patrick Moorhead, president and principal analyst at Moor Insights & Strategy, predicts great things for Microsoft in 2017:
“Microsoft surprised everyone in 2016 and I’m expecting more surprises in 2017.”
Those 2016 surprises included the launch of its chatbot, Zo, in October, which has so far engaged over 115,000 users. The Microsoft Bot Framework has attracted over 67,000 developers so far, and claims to be working hard at natural language processing, and improving cognitive function in its bots. Microsoft’s new service, Calendar.help, also uses AI to help with diary management and meeting scheduling, whilst also introducing its Cortana Devices SDK to hardware manufacturers to increase the smartness of their smart devices.
Moorhead has also stated that he reckons a new intelligent personal assistant from Microsoft may knock the socks off Amazon Echo, whilst following its characteristic brand route of focusing on business related APIs that use AI.
AI in 2017: #4. Better AI Personal Assistants
Echo really stole the show last year, leaving even the likes of Google Home far behind. Almost everyone you speak to has one, though some – like this very sweary man – cannot use them to their full capacity yet.
Though shouting ‘Trevor!’ at your Echo and smacking it around may be at the bottom rung of understanding, a lot of anecdotes so far centre around the fact that Alexa seems to get used just for playing music, playing Simon Says, and taking a barrage of playground insults from the kids.
Few of us have all the bells and whistles of the smart home enabled yet, nor can easily integrate all of Alexa’s potential ‘skills’ into our everyday lives. But we are at very early stages with the Echo right now. There are reports that Amazon plan to release a new version with a 7” touch screen, and better audio capability. Obviously, the new version will have a bigger price tag, but that’s just the way it goes, isn’t it?
And what’s more than certain is that, as time goes on, Alexa’s skill base will keep expanding. This will make her more useful to a wider range of users, and before long, we’ll be over the first hurdle to savvy mass adoption. If you listen to your grandparents’ tales of when the TV first arrived on the scene, and what a battle your granddad had making the darn thing work, you’ll see that there’s hope for the Echo yet.
Not to be outdone, of course, we anticipate that 2017 will see Google up their game in the digital assistant space, to try and compete with Amazon’s success, and we may see a new arrival from one of the other major players. We’re banking on Apple coming to the table, but when it does, it’ll probably still be the Paris Hilton of the smart assistant world: all glitzy good looks and shoddy brains.
AI in 2017: Your Predictions
Well, these are our predictions for AI in 2017… but what are yours? Let us know on Twitter! Use the hashtag #AskTDMB, or just drop us a tweet!